Skip to main content

tv   After Words  CSPAN  November 5, 2023 1:20am-2:15am EDT

1:20 am
i think we can do better than that. the book is sensational. it will for sale outside the doors, the holidays coming. martha is here to sign for you thank you so much for coming. state. martin baron, of course, has been the top editor at such newspaper, says the miami herald, the boston globe and most recently the washington post adamant along the way that
1:21 am
the journalist isn't story. but marty baron, you decided write your own. tell us how you came to the decision to write this memoir titled collision of power. trump bezos and the washington post. well, i knew that i was living through history. an important moment, history for the united states and an important moment, moment for the press and for the washington post. so the paper was sold by family that had owned it, the graham family for 80 years was then sold to one of the richest people in the world. and then along comes a presidential candidate unlike any we had ever seen before, and a president unlike any we'd ever seen before. and so i was living through history, an important bone, as they say in american history. and for democracy and i thought should tell that story. i was at the post for more than eight years during, that story and i think at some point somebody should tellbody shouldi
1:22 am
was the one to do it. you know, you had been previously, as i mentioned, at these other papers at the boston globe, although an important newspaper, not one, followed by all of america you, really became much more of a household name and figure, thanks to. the movie spotlight, which looked at what your enterprising core of investigative reporters did to the astonishing range of the child abuse crisis in the church in boston. and as it out, beyond, what was it like have been in one of the key subjects of this major hollywood film before ascending to the national in sort of your own right, as you did at the washington. yeah. the film actually came out long after i joined the post. so i joined the post and the beginning of 2013, and the movie came out was a 2015 movie. so it came out as i was the post. it was a look. i mean, i don't think any
1:23 am
journalist does or should work to be portrayed in a movie and we certainly did not do that at the at the boston globe. in fact, the first entreaties that we got from hollywood to have a movie made, we rejected. but then years later, we were approached again and we thought it was an important story to be told, the story of how investigative reporting works, the story of listening to. two survivors holding powerful institutions to account, basically explaining how investigative reporting works and why important in this country so. that to me was the importance of the movie. obviously it gave me some degree of celebrity. i guess mine or celebrity, i would say. but it's not something. i expected not something i was seeking. it just happened to me. and you try use it for a good purpose. i try to use it for the purpose of talking about the role of an
1:24 am
independent and free press in this country. and frankly, the world. so that's what i that's what i endeavored to do with that that and the movie and the timing, of course, being important as corrected me for that. and i appreciate that beyond that question beyond that that that i'm sure out body experience in some ways for you what it like to take step from going from this you know incredible important newspaper the boston globe to something of the stature of the washington post the journalism in some ways the same what made this so different? well, was joining a newspaper that had a national and international reputation. and i was almost certainly best for its role in the watergate investigation, which ended up bringing down a of the united states. it had, even if it had not been read around the country or people around the country were
1:25 am
certainly familiar with, it had an image of the washington post from perhaps the movie all the president's men, that was about the watergate investigation. and so, you know, and you're moving to a i was moving to the nation's capital and knew that i would be covering the most important stories in the country on a regular basis. and so you recognize that there's a huge responsibility there obviously i was sort of confronting the legacy of ben bradlee i legendary editor of the washington post and asked me whether i was intimidated by that. and and i said that i intimidated by it but i was by it. and that is, in fact the case. so there was a legacy that i, i felt that i needed to uphold a certain set of institutional standards that had been set down by. the graham family and by ben bradlee, by his successors as editor, as and and so it's a huge weight on your shoulder,
1:26 am
frankly. and i felt that going to the post. but i also knew that they had an enormous number of financial difficulties at the time that they were kind of at sea, as many news organizations were. this country, particularly newspapers, and that it was going to be an enormous challenge. what were the stakes for you personally and professionally? right. i was secure at the boston globe, although i was frankly ready to move on. i had been there for 11 and a half years. it seemed like a good time to move on and be two were facing enormous financial pressures although we had recovered really from the worst and you never know how things are going to work out once you go to a new news organization. they may not like you. you may not like place all of that, but i had taken risks before my my career. i had left the los angeles times
1:27 am
where i had worked for 17 years. and i went to the new york times. you never know how that's going to work out that i went to to in 2000. in 2000, i went the miami herald to be editor there working for a new publisher who was new to me at the and then i went to the boston globe where i really didn't know i knew hardly anybody in the city, really, only one couple. and i knew nobody the in the newsroom and so that was a risk. so i was accustomed to risks and i felt that this was one that was well worth taking both professions professionally for me and that i thought, you know, if i did a good job that we could do something really for the washington post and in doing something worthwhile for the washington post, i felt that we would be doing something worthwhile for the country. you were hired, if i recall correctly, by by. one of the sort of scions in a
1:28 am
sense, of the graham family. katharine weymouth was the publisher at the time the niece of don graham, former chairman of that. and she brought you board relatively soon in your tenure there. she met with you to explain that she would not be your boss any and in fact, there would be a transition in ownership what what happened and how did you absorb new information? she invited me to a to have a drink at a hotel across the street from the what was then the post's headquarters sort of iconic building in that in that sense and that was surprising to me because it was at five about 5:00, we did normally have drinks at 5:00 that i was busy. that's about getting close deadline time and so i had a sense that something might be up. you know, you develop a second sense after a while in this business. i've gone through a lot of turmoil in the in the news business and, you know, when somebody says, let's have a drink 5:00 and you're not expecting it, you think
1:29 am
something might be up. but i did not expect this. i did not expect that the graham family would be selling the post. i don't think anybody expected the family had been incredibly devoted to the post for 80 years. the parent in the name of the parent company was the washington post company and the name of our newspaper was, the washington post. so this was a total to me. you know, my feelings were mixed at the time. i actually i guess my initial reaction was that, well, first of all, shock. but second of all, that it might be good for the post. the truth is that we had, or at least the graham family and everybody had pretty much run out of ideas for to turn the place around and were facing an enormous number of cuts. i was beginning to develop plans for the next year, budget plans and they called for more cuts on staff. i had already done about i was expected to do about cutting 35
1:30 am
people in our newsroom in the first year. they said they were going easy on me and then the second year was going to be more like 50 or more. and i expected that perpetually, frankly. so was somewhat encouraged that somebody who who had a reputation for growth would be purchasing the post that he would invest he would come with some new ideas because lord knows we needed a lot of ideas in the business. so i thought would be it might be good for the post, but i didn't know if it was going to be good for me. the usual equation is new owner, new editor. so you know i expected that i might be i might be dismissed and that jeff bezos would find a new editor not what as it turned out, you know business, in some ways one could look empirically at, you know, strands from a wikipedia entry, say, well, here's a guy who should have been a walking conflict of interest. you had so much business in front of the government and wanted more. who turned out wanted to develop
1:31 am
things in washington dc who it turned out got involved in personal scandal and controversies with an ensuing president. but there's a constant thread through the book about you know it's really an admiration of who the kinds values bezos expressed and also the kind of civic leader he turned out to be. talk a little bit about that. what what is there that is so, so about jeff bezos, but of his stewardship of this important presence to tissue? yeah, i guess is a surprise for a lot of people because lot of people have a clear of him based on what they've read or or seen in a different context than the washington post. what i wrote about was my own personal experience with him. i personally observed at the washington is his ownership, his his management of the of the place and of me and other people on the staff and. so, you know, i mean certainly he has huge commercial interests
1:32 am
mean there's no doubting that and he's a controversial for a variety of reasons from labor practices at amazon to or intrusions into private c to their lobbying efforts in washington and you name it. there are a whole range of things so and it was our duty to investigate all that of course and to continue doing so despite his ownership even so my experience was that he he was going to going to give us our independence and did give us our independence completely, he said at his first meeting with the staff in a town meeting that we could cover him and cover amazon any way we'd like without any interference his part. and he really that on several occasions to me and that's exactly we did we certainly put it to the test we cover it aggressively issues at amazon we covered aggressively you know all aspects that company and of him his own personal when he got divorced and had you know had an affair and you name it and he
1:33 am
did this while coming under tremendous pressure from donald trump who was was initially the presidential candidate and then president, clearly, i would say the most powerful person in the world and was who was regularly attacking bezos and seeking to undermine his his business. and the clearly the primary of his wealth. so and he stood by us throughout the whole time. he never he never caved. he never submitted to that pressure. he didn't interfere in our and in the content of our coverage any way, even when it was about amazon, when it was about himself, in no instance did he interfere in the coverage. it seemed to me over time, as an outside observer, that if anything, your coverage of, amazon only became more rigorous and tougher and more, more deeply diving into questions
1:34 am
that amazon as a corporation not have wanted a great public scrutiny. right that that you guys felt comfortable doing that let's talk let's get into the journalism that you experienced and let's start with early example of it. the the question. barton gellman, a former washington post. national security reporter, who walks in with what is in some ways one of the scoops of the new century. and that's what, you know, people now think of as the snowden papers talk about challenge that presented and the opportunity you as a relatively new editor at the post. yeah well bart i came into our newsroom and he informed us that he had these incredibly sensitive documents, access to them or was going to get access to and that he wanted to know and they were going to reveal a surveillance regime on the part of the intelligence community in the united states, a surveillance practices that swept in a lot of information about american citizens, as well
1:35 am
as on citizens around the world. and he wanted know whether this is something that we would be willing publish. and we listened to him, he was very careful. he was very thoughtful. he was very cautious extremely cautious. and, you know, i, i it was clear to me that he was explaining that would be of not just a curiosity, but actually of true public interest. because one thing i think americans value and that i value is privacy. and i think we should value privacy in this country. and this was an exceptional intrusion into the privacy of americans. and so, you know, one of our primary principles and national security information is whether there's a true public interest at stake. so i that we were willing to go with that and then that evening, i thought a lot about it
1:36 am
because, look, i mean, this huge, hugely sensitive, the most confidential information in the u.s. government, it comes it came at a time of deep public concern about terrorism. i had lived through 911. of course, i had been in boston at the time. two of the planes flew out of logan in boston. those two planes were the ones that hit the world trade center towers. i was very sensitive to that. and so, you know, i did i was very concerned that we do the right thing, that we not endanger national security, but also that we make, the public, aware of a level surveillance that was certainly not certainly not public knowledge and which i don't think that a lot of americans approve of. if they had known. and so i thought a lot it that evening, i did some research. i was i had printed out the espionage act of 1970 and i was
1:37 am
highlighting it. i thought more about it. i was i was thinking maybe i should change course. but as i thought more about it, i thought, no, i think this is the right thing to. do and you look, why did you print out the espionage act? was it important for the executive editor of the washington post to look at the actual text of that century old law to help gauge or not, this was right to print because there was a risk to the institution that i represented to the washington post. it could have suffered huge fines. it's possible people on staff, including ownership, could have been to prison. we certainly it's possible that we could have been prosecuted. that had not happened in the history of the united states, but it didn't that it couldn't happen. and so i thought that i should take a look at the law and think about and think about the the risks the publication posed to the institution. i think that's the responsible
1:38 am
thing to do. and you know, and i did think about and and i knew that this was a this was putting the institution at risk. and ultimately, you know, within days i told my publisher, katharine weymouth, and ceo of the parent company don graham. well, i told i told, katharine weymouth, she told don and and they trusted us to make the right call and they did not interfere. they trusted my judgment. they trusted bartz and allowed us to proceed. you talk about the importance of privacy, which i think was a key element of what was shown that under in the obama years, that that was still being sort of disregarded in some when national security was invoked by senior officials and the like. i wondered if we could play out
1:39 am
that question as it confronted you in in the years that followed, as it out, so much hacking happened of key political figures in the 16 race, particularly the democrats, as a source of disruption and then more that it was treasure trove or news organization as well as for political actors on the scene and if i recall correctly the post was among those that you know reported aggressively when it felt the news was warranted. how do you in retrospect you come down on that about the use of such materials and about the the sort of principles that should guide the use of such materials, given that it's both revelatory and incredibly intrusive, you yeah, well, i think it's a really important question that the press needs to grapple with. we were the first to report that the democratic national committee computers have been hacked. we certainly raised the question of the likelihood that this was done by russia or russia in some
1:40 am
sort of collaboration, maybe formal but indirect was providing information to wikileaks, which then released that information. and and a lot of these emails very hard to ignore. you can't really ignore those emails. can't pretend that that didn't happen. and in fact, those emails were creating a stir even within the party, whether we're and we're going to create a controversy, whether we were in publishing them or not, the reality is that they suggested that the democratic national committee, its leadership, was favoring the hillary clinton candidacy over bernie sanders. and ultimately they revealed some speeches hillary clinton had given on wall street and how much she was paid for those speeches and the content of those speeches, which was highly controversial as well. and so i don't think we could just ignore that. at the same time, though, i
1:41 am
mean, i think that we probably if i were to do this over again, is that i would devote as much in the way of resources to the origin of that, to the hacking itself, to the reasons for the hacking, the motives for that hacking. as i we did to the content of those emails. so i think we needed to do both and i think we focused overwhelmingly on the content of emails. we certainly highlighted that russia likely involved, but you we we did on that at the time than we did on the content of the emails and part of the reason for that is that the intelligence leadership was unwilling at point to say that russia was responsible for that or that it happened for nefarious purposes. but it did seem you know it emerged of course, that russia was seeking help. the trump campaign, that was the ultimate conclusion.
1:42 am
the intelligence community and. i think that that's something that's a theme we should have emphasized more forcefully as sooner in the coverage of that campaign. it's funny, i remember i had with donna brazile, who is a top party leader, as well as a cnn pundit at the time, and she just said, you know, david, i'm not going respond to the substance of the questions, which i personally found pretty interesting in the moment, because i don't want to give credence to the idea that it's okay for foreign actors. steal private communications and you know that did set me back that made me think a little bit about story i was doing at the time and it made me think about this question. obviously, we've all now had to think about this question lot. does it matter if such material comes from somebody who presents as a whistleblower, as edward snowden did, or somebody who presents as a chaos creator, as the, you know, agents acting on behalf of the russian
1:43 am
government, does or does it matter most that the material itself is of important public knowledge? yeah, well, those are all really good questions, david. i think that's something we have to grapple with, as i said, i don't see how we can ignore the information i recognize that our appetite for that information is going to be an incentive for people to hack in the first place because they know that we're going to eat up and that we're going to publish it and we're going to jump on it and all of that. i do think that i say what i think we need to do is, really balance that with. our coverage of the motives for that information being released in the first place. and so think that's what we we should have done more of that in that particular presidential election. so let's turn to now there's a moment in the book where you write, in preparation for trump, i could be getting the timing slightly off, but i think it was after he was elected or he took office that you set aside, i believe, five, half dozen books to read them.
1:44 am
timothy snyder's on tyranny. another about how democracies fall and a few others. what did read and what did you take from your self assigned curriculum there there. well i took from that that trump had real authoritarian impulses that a lot of what he had said and a lot of what he had done suggested that he he a not only admired authoritarian leaders around the world, which he had kind of made clear that point, but that he actually would like to have authoritarian powers and so i was very sensitive to the kinds of measures he might put in place that would allow to to act a more authoritarian authoritarian way. and there were a lot i his condoning of violence at his rallies what's an example of that he always found an excuse for that violence usually by saying well they were angry as if that were sort of
1:45 am
justification for violence which of course it's not his admiration for certain foreign leaders his willingness to demonize his certainly the press but the language that he used about globalists and all of that sort thing, some of which actually verged on being anti-semitic, i think you could say, but so all of that, i think, suggested that he he had authoritarian tendencies and that that was something that needed to inform our reporting on him. and i needed to keep i needed keep that in mind. it sounds like you wanted to you were and wanted to go in very clear about what you perceived his is operating framework instinctive approached a power was yet when asked about your your own approach to this you said we're not going to war, we're going to work. tell me you meant by that.
1:46 am
and then about the public discussion that ensued or. well, keep in mind on the very day of trump's presidency, he went to the cia and speaking to agents there in front a memorial for fallen cia. what did he do? he decided to talk about the press and he said, as you know, i'm at war with the media. the that i took from that. and i think he was to enlist those agents in his own war with the media because here it is. here he was the president of the united states who was at war. and and they worked for the president of the united states. so a couple of weeks later, i was asked for my reaction. and i said, we're not at war with the administration we are at work. so did i mean by that? i meant we have to remember what our fundamental mission is here and our profession. and we have to think back to why we have a first amendment in this country, why we have a free and independent press in this
1:47 am
country and. you know, james madison, who was the principal author of the of first amendment, he talked about freely examining public characters and so let's dissect that, first of all, free, freely operating, independent examining meaning. it's not meaning that we actually hold government officials to account real to account. we hold the people, the powerful individuals and institution ones that influence public policy to account. so examining the public characters are the politician, the other government official from the people who influence them and the measures or the policies that are implemented. so is why we have a first amendment in this country. that was the original intent. it's something that i believe any should understand, particularly they're taking an oath to the constitution of the united states, which they do.
1:48 am
and it's that we in the past have to always keep mind our job is not stenography. it is journalism. and journalism beyond simply recording, people say, but looking at really looking at who's behind these policies, who are the policies intended to benefit, what kind of practical impact will they have? what are the motives? you name it, what all all of these sorts of things. and that's what what i wanted us to do. and so my view is it's not a war it's work, it's evidence is our work. it's been our work from the very beginning. and so that was the statement that i was making. i'm not doing we're not doing this out of any animus toward donald trump. we're not doing it out of this. something that was done with presidents as well. it always has any fair of the history of the washington post or other organizations. and even at the local level that is what the has done and what it should do. so my view is that's not war.
1:49 am
this is work. and yet it sparked in some quarters, including newsrooms, including some corners of your newsroom, the response, you know, this isn't enough. this isn't to meet the moment. you know, media critics like jay jay rosen made that. but there are journalists who who would argue that as well. why, in your view, is that wrong? why, in your view, is it inappropriate as comes across strongly in the pages of this book to step outside conventional values in the face of an almost unprecedented stance from president administration towards the the not just the credibility but the legitimacy of the press itself? yeah, well, i think that if we if we present as partizans, if we are partizans, then we lose credibility with the public. all right. our job is to our job is to hold all politicians to account.
1:50 am
i don't care what party they're in. i don't care what ideology is. if it's a democratic president, have an obligation to do that as well. so i don't see ourselves being in a war here, and i don't think it's actually necessary in fact, i think it's it's destructive and it erodes public confidence. and if we are seen as we are seen as partizans, if we act like partizans, and if we see ourselves in a war, then we begin to act like partizans. and i want us to act like partizans. i want to just be journalists. and that is true, as i say, certainly to do a lot of to do a lot of reporting, to be rigorous, to be comprehensive, to understand that our goal is to gather the facts, to get of the truth as best we can and and then tell the public what we've really found. we're not fighting that from the public. there's no point in doing all this reporting unless we're going to tell the public what we're what we found. so i think that that works really well.
1:51 am
and i think i mean, i think also think history the history of the press shows, that it does have a huge impact. i can point to a lot of journalism that had a huge impact, whether it's the watergate investigation of the washington post, the pentagon, work at the at the new york times, a whole range of coverage over long period of time. that not partizan in nature. there was simply fulfilling our mission as journalists. and i think that that has a lot more credibility with public and should have a lot more credibility the public and as soon as we forget as soon as we think that we are warriors as opposed to journalists, i think we're going down the wrong path. well, in certainly, you know, during your tenure during the trump term the post did an astonishing array of not just political coverage but accountability reporting, investigative reporting, you know, whether it involved the secret service, it involved questions about what was
1:52 am
happening on the border with with mexico and the treatment of people into this country without documentation or legal status or whether it was david fahrenthold, essentially crowdsourcing questions about what trump had done with donations. talk a little about the inventiveness that went into what the post did in trying to surround this, you know, completely politician at the head of the levers of the american government. well, a lot of it was traditional and some it was new. so what david fahrenthold did, for example, was new. i mean, initially we reported on, you know, trump had avoided had ducked out of a debate and he held a rally. he said he going to give a donation to, veterans groups and all did was really just follow and try to find out where the money go who was a given to and the trump camp. the trump team couldn't give
1:53 am
straight answer as to where money went. so he went on to social media, which in a previous era would not have been available to us to out asking whether any veterans got this money and and wasn't hearing back that anybody got the money. so we just it was hard just to stop for a second, marty. i he took, if i recall correctly, a notepad in which he would write out and say, this is what i got can you guys add to it. you know, it was it was almost as though you were sitting over the reporter's shoulder watching him do his work. it was incredibly engaging, almost entertaining game of the the idea of some rather serious reporting. that's true. that's true. and david was brilliant at it and inventive and came up with some real creative to practicing journalism. and they worked in the current era where social media is so
1:54 am
important and it did give them insight into what he had and what he didn't have and to see whether they could add to his his knowledge and, it did force donald to actually distribute that money. and he finally informed them that he i mean, they had said that the trump had already said that they had already done that, but there was no evidence it had. and then and then they did it after he reported on all that throughout the book, you're very generous about pointing out the insights and muscularity of the reporting, the editing, your colleagues there. it did seem as though with fahrenthold, there was one moment at which you said, well, if he's doing this meaning trump with this one charity, what's to say not doing this with all kinds of other things and sort of almost launched? fahrenthold a cottage industry of reports on this how important is you know you were editor of the globe for i think 11 years and of the post for nine if i'm not mistaken, how important?
1:55 am
go ahead. i'm sorry, a little over eight, you know, a little eight. how important is the tone set from the top for an editor? how much can you steer these battleship types in appreciable over the course of a ten year you know well a newsroom, a collaborative enterprise. you work i think that's it's of the things maybe that's being lost when people are not actually in the office so. i often use the example of fahrenthold example. so we ran into each other at the elevator bank. i congratulated david on the work that he had done and talked about how great it was. and then as i was speaking, i don't know, it just occurred to me to ask. i said, well, if trump did this sort of thing where something was something that was so public, what really happened with all his his supposed charitable donations that were done that were supposedly made without a lot of public attention and i don't think it would have occurred to me really to ask unless i had just bumped into david at the elevator bank and he took that and he ran with
1:56 am
it and he recognized, i think, a good idea and and did a brilliant job, job without that kind of. serendipity happens a lot in newsrooms, at least when people were actually showing up in newsrooms physically. and i think it's incredible, incredibly important. so the editor can have a tremendous influence, a lot of different ways. certainly you can order up stories, but it's better actually have conversations with people. the staff ask a lot of questions i think it's important to just ask questions and draw people out to see what are the unanswered issues. so some of the the secret service, carol leonnig had written a couple of stories about the secret service that real mishaps and blunders on their part and. i you know, i asked carol this was part a bigger pattern whether this is just there a systemic problem at the secret service because these were incredibly embarrassing and we a meeting and carol that she had
1:57 am
leads on some other stories and that it didn't look like a a bigger problem. and so we then there was a group of editors in office and we authorized her to spend full time on. and she did and and produced a tremendous series of about the secret service. and it's blunders and and that won a pulitzer prize her and for the washington post you know one once i'm hearing you say this out loud it does remind me a little bit of the genesis of, of course the great spotlight reporting on the catholic church how much did that prompted in part by your prodding? i mean, really at the very opening of your tenure as editor in boston, how much did that influence your thinking about systemically in that way? well, i was really from the very beginning. first, i just to get to the basic facts. and i mean, i think it's really important for people in editing roles and reporters as well to
1:58 am
just keep asking questions like, what don't we know? what do we need to know what are the unanswered questions to? what has truth been left unknown or is there a way to get at it? so the case of the catholic church, when i arrived in boston, there was a column written by a fantastic columnist, eileen mcnamara. she about this case of a priest having accused of abusing as many as 80 kids. and at the end of column, she said the truth may never be known because these documents that might reveal were under court seal. and at the time, the lawyer, the plaintiffs said the cardinal himself was aware this abuse and yet reassigned this priest parish to parish without telling anybody. and the archdiocese denied it. so that was absolutely not true. and so at my first meeting on on my first day working at the globe, everybody talked about their meeting about what they were doing for the day. and nobody happened to mention this particular case. and i said. well, what about this? i mean, this incredibly
1:59 am
important story, if it's true that the cardinal knew about this this abuse and reassigned this priest and essentially enabled him to abuse again and again and and i asked, can't we get to the truth? can't we get beyond beyond one side saying one thing and one side saying something else and and people pointed out that the documents were under seal. and i said that i knew that because i read that. but had we thought about going to court to get those documents arguing that there's a public interest in it and certainly people were surprised to hear that suggestion at that first meeting. but did talk to our lawyer later and then we the parallel investigation journalists took investigation talking to survivors talking to lawyers talking to anybody we could priest and and we produced that sort of stories beginning in january of 2002 and over the course of the next and a half, we probably produced about 900 stories and and we can still see
1:00 am
the impact of that that set of stories today. very much so. that idea of can't we get at the truth undergirds much of what you present in this book of your your experience, your perspective, of your years at the post to. take it back to the trump years. it seems to me that were editor at a time when you as an editor the post as a newsroom we all as part of this nation experience of a series of i don't know shocks to the system. the election of trump and the the just completely unprecedented nature that administration from stem to stern really you know until you know what we saw january 6th on his way out the metoo movement that really exploded i would say in 2017, starting really 2016 and changing nature of how many journalists looked at reporting on questions of abuse, of mistreatment, of harassment of women in the workplace and beyond. and then, of course, the social
1:01 am
movement that exploded in the streets, but really also in the newsrooms across country with the murder of george floyd and subsequent events and revelations about how a law enforcement other systems work. it seemed to me that that was in a relatively short number of years. some major social upheaval and, not just recalibration, but rethinking tectonic plates on the at times in ways that were hard for news leaders to manage as their own journalists started questioning prior assumptions. how did that play out at the post and how did that play out for you as the news executive, that newsroom? yeah, well, you know, i, i think our that covers a lot of subjects there so i played a little bit different from one subject to the next. but, you know, i mean, i think that people started to feel a need to express themselves more particularly on social media.
1:02 am
they had strong views what was happening in this country terms of racial justice in terms of justice for women who had been harassed or assaulted. obviously for, you know, all of what was happening in the trump administration, whether it's appeared to be an attack was on attack on basic, basic public civility, on on free expression, on norms of political discourse, all that once civil rights people felt so so i think people were motivated to express themselves more particularly on social media. and that was not something that i was in favor of. i remain strongly opposed to that. i don't think it's helpful to us i keep asking people, you know, can you show me the journalists tweet that changed the world? i haven't seen one yet, but i have seen a lot of journalism
1:03 am
that has done a lot of good. and it does go back to reporting let's find out what the facts are. let's tell people what the facts are ultimately. the public gets to make these decisions about, what to do, what where we where we head, where we're heading as a society who we elect to govern us. all of that on our call to make that that decision our job is to give people the information they need and to know in a democracy and so that's where i think we that's our lane and i think we should stay in that lane and not become those of us who work in the news department should not as opposed to let's say, opinion writers and editorial writers, people like that. we should be and stay in the lane of reporting, reporting hard reporting aggressively doing our job and telling the public what we find directly, forthrightly unflinching, frankly. and i think there's a lot that be accomplished with that. it has been throughout history that kind of reporting has
1:04 am
accomplished a lot and. i haven't seen much accomplished by going on social media and, expressing people's personal views. in fact, you write here too often, this is from your book, too often it became a venue for personal opinions, advocacy, anger, snark, sniping filled humor, virtue signaling, personal animus, and a rush to judgment. never more so than during the trump years. there came a point where colleagues, your colleagues my colleagues, people in other newsrooms, not all, but some, you know, really felt that being and being a truthful, required in some ways something that went beyond a strict recitation of facts or syntheses of analyzes. and it required in some ways lived experience and personal personal, personal understanding of things to make things more powerful and authentic. and that that was the distance between the media and the audience. they felt come out in a different place in this book.
1:05 am
explain why. i think people's personal experiences are highly relevant. that's why we really needed diverse often a newsroom we need people from all different backgrounds who have lived different lives, their stories that other people will detect that i wouldn't even aware of their perspectives other people have that i couldn't possibly have. we should bring that all of all of that should be brought into our newsroom and we should discuss that among ourselves and talk about that as the kinds stories that we should pursue and how we should those stories and things like that. but the stories that we do have to conform to institutional, these are well-established boundaries in our newsrooms and you know, when you walk into the washington post, there's a name, the washington on the building what does that mean? does that mean something or does it mean nothing? are we just a random collection of individuals working in a building that has washington post on its name or does working
1:06 am
in that building mean that we actually work as a team that we work collaboratively, that we discuss things collective, play that we make, that we make decisions in a collaborative way as opposed to each individual expressing themselves on social media however they want, whatever they want, on whatever subject they want, all of that generally, impulsively, instantaneously. that, to me is not journalism. that's something else entirely. many instances, it's advocacy, and that's not our role. i we can accomplish a lot if we discuss of this together. we as i said, we need very diverse staff. we need people from different backgrounds. they do need to bring up the subject. so that we can discuss them. they can inform coverage and do inform our coverage. but every person just going off on their own, doing their own thing means that you're not really functioning and you're not functioning an institution.
1:07 am
it's just a a bunch of individuals should work saying whatever it is they want, however they want, never want. so let's take stock for a moment about where the trump years and its successors have left. we are at a time where it like we are awash with misinform and you know, whether it's recent events in israel and gaza, whether the fight between ukraine and russia, questions about vaccines in this country, questions about the elections past and upcoming. and even as the post and other media organizations do what they are laboring at their best to do, which is rigorous, factual based reporting, they are also how do we at a time of so many attacks on and its credibility, how does the press or journalists.
1:08 am
to establish credibility with more than those who feel that they're rooting interests are represented. any given story? yeah, well, it's a fundamental question for our profession right now. clearly very important. i think, first of all, it's important to remember something that you just said, and that is we are fallible, are human beings. so the reason we make mistakes, just like every other human beings mistakes, it doesn't mean there's a conspiracy. it doesn't mean there's animus. it doesn't mean, you know, incompetence. it doesn't mean of that. it means that people are human and they make mistakes. and the important thing to do is to correct those. when you become aware of them, if in fact, they are they are mistakes. so so that's one step that we certainly should take and we should recognize our own fallibility and our own limits actions. and to recognize that, i think it's very important that we go into our journalism asking the right questions, not assuming that we have answers, but go
1:09 am
seeking those answers, asking the right questions is hard enough. job knowing what those questions are. we are often seeing the world through a keyhole. that's certainly case in many areas the middle east, particularly in gaza, because covering that, that that place now is just extraordinarily difficult. so i think we need to in this country, i think we the press need to get out more into the community. we need to talk to all people in all corners of our society, regardless of regardless i mean, all races all ethnicities of people, all classes, all all experiences you name it. and all professions and and to get a sense of their concerns, their worries, but also their and their hopes and and reflect that very fairly and, without any condescension in our in our
1:10 am
work judgment about them, in our in our in our pages and on the air and what on on our sites, what you so i think that's important for us to do. another thing that's important for us to do to lay out the evidence for people we have the tools now where we can show people things that we could show them before. so if covering a court hearing, we show them the evidence been presented in court, we can them the ruling. we can annotate that ruling to highlight the particularly relevant portions of that ruling. we access to videos. we have access to to audio that we can present online so that people, if they don't trust, they can go watch it for themselves, listen for themselves, read it for themselves. so we need to be as transparent as we possibly can. and in almost every because we should just go this with the assumption that people won't trust us. so we're present. here we are here it is here's
1:11 am
the evidence, here's the basis on which i wrote story. so those are some of the things that i think that we ought to be doing a couple of quick questions before we wrap up. among the things that you expressed for jeff bezos, the stewardship of the post was that he didn't treat as simply as a charity. you know, this guy could cut for a very long time before he noticed anything really diminishing tens of billions of dollars, personal wealth. and yet he said this has to be a viable and in under your stewardship and trump years you know the post grew in size and stature in digital subscriptions which is really the motherlode for for most newspapers at the moment. they're seeking viability and some sort of stability, and that looked pretty good in recent days and weeks we've learned that the post obviously well since your departure but has embarked on an effort to buy out 240 of its employees about half of which will come from the newsroom.
1:12 am
that represent roughly 10% of the newsroom. and they'd also a modest, more modest of staff earlier this year. the bezos is appointed acting ceo said our anticipated growth revenues and particularly subscriptions just wildly out of whack. they couldn't up to real scrutiny as things now. what do you see now at a slight remove in terms of what we should what conclusions we should draw, what what we should how we should assess the post's strength and stability at the moment? well, i think it remains very stable. it's owned by such a wealthy individual also. i think that it has a great franchise. i think it's important to remember, as you say, that the post grew tremendously over the years. we had six straight years of profitable kitty. that money was all reinvested. when i got there, there were about 580 people on the staff we were having to make cuts, about 35 in the first year.
1:13 am
so and then probably another 50 or more in every year in. the years ahead. what even after the cutbacks that they'll have at the post now in the newsroom, they'll be left with, i believe, about 940 people, which is not far from where it was when i left. that's that still represents significant for the post and a very strong foundation for sustainability. they did overinvest they didn't adequately prepare for a post-trump era a lot the growth at the post was driven by, keen interest in politics a desire to hold trump accountable. all of we needed to prepare for a post-trump era. it was particularly apparent for two reasons. one, after 2018, when the gop lost the house and then of we could see, for example, at the new york times that they were diversifying. they had invested hugely in
1:14 am
cooking up, they had acquired product recommendations, service. they had started games initiative, building on their famous new york times crossword puzzle, all of that. they were basically insinuating themselves into people's daily routines in a way that had nothing whatsoever to do with news. i was evident years ago. and we did need to prepare for that. i urged us to prepare for that. i think that there was a sense that things were going fine at the post. we did have six straight years of profitability and but we didn't we did diversify. we did invest in a lot of different things. we expanded our technology coverage tremendously, we added some other things, but it was not to the same extent as the new york times. and so now they have to make an adjustment because they did overinvest they did over expect in terms of what their subscriptions would be and what traffic would be like.
1:15 am
and i think it's important to not just to project from one year performance into the into the future. i mean i think it's important to recall that people predicted the post would slide into oblivion that was about the time that i arrived that people were looking it very pessimistically and also even the new york times. it wasn't that ago that people were saying the new york times would slide into bankruptcy and that it would be overtaken by buzzfeed news and huffington and all the upstart digital outfits. that didn't happen. the new york times turned itself around. what that tells me is that we shouldn't just extrapolate from one year financial difficulty into the future, but that we should we should understand, that news organizations can't turn themselves. and i think that the post will i think that bezos is very committed to future of the post. he said so many times believe that. and right now they have to make some sort adjustment and, you know, look and debate whether it's the right approach or not, but clearly, if you're expected to lose, as they projected, $100
1:16 am
billion this year, it's not a probably not a good call to do nothing. but let me ask you this last brief question. marty baron, what surprised you as you had time to reflect as you pulled from decades of hurtling headlong into these daily deadlines and long term enterprise investigations, what insights surprised you about your own tenure or about your own trade of journalism for all these years? i don't know that it was a surprise. mean i think that there is i a lot on the tensions within our profession i think a lot of it is a reaction to donald trump and his attack on the press and it's a tank attack on norms political norms in this country and worries about what what his movement if you want to call it that could usher for the united states. and so you know i certainly found myself grappling with that and what my own views were on
1:17 am
that. and and how other people view it differently. and but, you know, i came to the conclusion that i don't want to say i came to the conclusion that i was right. but i do have i do have some firm views that and i think it's really important for us while we have to change the way that we deliver information to the public technology is changing that people will want to receive their information in on different devices and in different forms and we have to change with that and we have to our jobs are going to change but i think should be true here to the traditional values. i don't think the response to trump's obliteration of norms should lead to the press obliterating its own norms, own standards, abandoning those standards. i think we should stick to those standards if we go back, if we fear from those standards, i think we run a real of merely contributing to the erosion of democracy in this country.
1:18 am
marty baron, thanks so much. thank you,i decided after i reah
1:19 am
1:20 am
1:21 am
letter to an open letter that had hundreds of experts in the field sign onto it. and one of the things they said is that they were in, you know, looking at how we can control and contain and do something that's already out there it's not something in the future. it is here. and now with us and they asked for a pause hit the pause button and let everybody collectively are we doing this right? how do we go forward and reading that letter and then just it seems like every week more articles and news information about i see people's head shaking it's constant. so started reading books and we're fortunate to get three of the wonderful authors here today we have jacob ward author of a
1:22 am
hold his book the loop how technology is creating a world without choices how to fight back. fight back. we have david auerbach at the end here, and his book is meghan. it's how digital forces beyond our control commandeer our daily and inner realities. and calvin de lawrence here hidden in white sight. how ai empowers and deepens systemic racism. so we've got the voices, we have the books. i will just reiterate what he said. show the authors some love and buy the books, but not to show them love. just because you're just going to get a taste of of what they've got to say today. i want to start the program with of the authors taking a few minutes to to tell them what i tell you, why they research, why they thought about it, why they
1:23 am
work in the field and. so i'm going to start with jacob ward and as i said, he is the loop author of the loop how technology creating a world without choices how to fight back, jake is the correspondent for nbc news and has been a reporter for cnn. he's been on al jazeera and nat geo the discovery channel hosted, a documentary series with pbs nice hacking your mind and he was the former editor chief of popular science magazine. he's written for the new yorker, new york times, etc. and jake argues that unless we act fast to fight back artificial. intelligence is about to amplify the most primitive version of who we and spit it back at us for entertainment and profit. so i will let jake now speak directly. good morning, everybody. good afternoon. thank you so much for being here. yeah, so i my sort of
1:24 am
transformative. moment was i got to, as carolyn mentioned, got to do this pbs series called your mind, which was this sort of rapid fire world tour for me of some of the top minds, behavioral science, behavioral economics, social allergists, political scientists trying to sort explain to me and through me, to our audience the past sort of, 30 years of of breakthroughs in that. and one of the big sort of fundamental takeaways is this idea that we have inherited and some very deeply progress, manmade evolutionary systems that really make us move around the world on automatic pilot even today in most sophisticated form, we're all sitting here and just so nicely and, you know, cooperating together in a world that experience of doing that documentary totally transformed my view the world. and at the same time, in my day job as a technology reporter, i was bumping into more and more and more technology, employing people in that stuff who were trying to bring in the top minds
1:25 am
of behavioral science to teach them how to make typically their their products more addictive. but in other ways trying to increase the amount of time you would spend with it over the long term what insights they could gain from you think all of us have experienced rate the to which technology companies try to mine your behavior in order to figure as much as they can about you. the saying if the product is free you are the product right and then the third part of it was then bumping into everybody, exploring these a.i. models that were very, very good at taking data, picking invisible patterns out of it, and regurgitating for the use of these private companies. and i went to this dinner basically in 2016. i think, which i literally sat and watched these two newly minted phds in addiction. they're just out of studying addiction. we're speaking to a group of fledgling young app makers, entrepreneurs, explaining how they would use their insights
1:26 am
about addiction to make your your app, any app you wanted to hire them to to improve as addictive as possible. and they were pulling from all of these threads, science, surveillance, economy stuff and from a.i. and i just thought there's something going on here that i have to try and get out in front of and at time the book was called paranoid and speculative i had a guy blow me up in the wall street for, you know, saying, i'm just of this kind of handwringing. and now it's dated like a year and a half later. my book is like a little behind the times of the paperback comes out on tuesday, and i had to scramble and write the new introduction because chatty betty had not really been released. when i was first focusing on this stuff and the race to the bottom of trying to make money as quickly as possible off of all us using these systems was is a big part of it. i just want a nod here to both the work of david and carl and david you know his thesis about there is sort of no one in charge really and this is just sort of taking on a life of its
1:27 am
own. i absolutely agree with and calvin, who's who's thesis about systemic racism being deep by this stuff i have encountered personally again and again as i talk to these the makers of these systems, the unconscious ways that they are just pushing stuff out in the world that i think is is putting paid to of these theses. so i'm very honored to be here. thank you so much for having me. thanks, jake and david we're going to bring david are into the conversation now. the author of magnets how digital forces mega nets mega nets. you know, we talked about how do they work? yes. as in if it's i should have seen it coming. i know it's your fault. you warned me not to do how digital forces beyond our control, commandeer our daily lives and into realities. realities? david is a writer, technologist and software engineer who's worked for ten years as a software engineer at google and microsoft. his writing has appeared in the times supplement mit technology,
1:28 am
the nation book forum and more. he was slate's columnist from 2013 to 2016 and teaches the history of computation at the new center for research and practice. and here is david. thank you so much for having me. carolyn thank you very pleased to be on the panel. and you know, we were the three of us were speaking at the other day and we realized that, you know, so many of our concerns did overlap. and in fact, you know, some of us were talking about that in. our sense of some of the most dangerous areas were were the same and that, you know what our differences were more in our backgrounds and in you know what we what we perhaps saw as possible avenues forward. the impetus for me write to write mega-hits and to and to coin a new term as ill advised as that may be came.
1:29 am
i think during my time as an engineer because there was this conversation i would keep having with people who were in policy or academia. am i coming out loud enough? okay, great. and people would say like, wow, you know, social media is so toxic, youtube is so toxic. everything is so toxic or annoying or or discriminatory, you know, why can't you just why don't you just fix this? are you that greedy and, you know, other engineers and i would just sort of talk about this and be like people really we have more control over this stuff than we actually do. and, you know, there was this ongoing sense, especially in the last ten years or so, that there would be this series of scandal. they would just be nonstop scandals, whether they were privacy related or discrimination related or surveillance related, you name it. and there would always an outcry and there always be like talk
1:30 am
having to fix it. and yet things did not seem to get better and. i wanted to write this book because i felt that we were missed the problem and that we looking for solutions that that couldn't there. and one of the examples i give is in i think 2018 or so facebook. had had one of many controversies, i think neo-nazi propaganda in in scandinavia and and there was an internal that was leaked in among facebook's executives where they said i want to quote this exactly. they said to avoid limit the effect the quotes that limit the meme that were slow to spot misuse and can't control facebook limit the meme that we cannot control our systems or to slow to spot these different types of abuses. and i stress this memo because this people often, oh, facebook
1:31 am
could just crack down on this. they weren't so interested in money. but i think this actually and while the incentive still there and i have no desire to absolve the companies of of they are doing because in some ways this is that you break it you bought it scenario if it's out of your control you're still it's still your system but if we're asking them to do something that's impossible, then that's going to happen. and i think we have to realize that these systems have gotten so large, so fast and so out of control. in fact, they are so consistent. they consist so of feedback loops to take a term from. from jacob's book that they inevitably do end up building on themselves and exacerbating their own worst tendencies. many of these tendencies being ones that the calvin chronicles in his book of reinforcing and reevaluating discrete discriminatory practices that
1:32 am
are encoded in ways in these algorithms that may be that may be incredibly hard to detect. so what i to say, you know, is that these problems may be even more profound, more difficult to address than than we even think and to think of them merely as as issues of, financial self-interest is misleading and. i don't think that's re-assure i think that, you know, there's a there's a great comfort conspiratorial thinking because you think well, someone's got the power and they could just, you know if they just chose to do differently, everything would be great when in actuality the combination of hundreds of millions people interact and constantly nonstop with these gigantic tech worldwide network computer networks and constantly changing their data streams so that you cannot, you know, step into the same data stream twice,
1:33 am
so to speak, makes it very to track why something will happen and why something will happen, why something has happened, and to predict what will happen in the future. and that was another thing when i was an engineer, i actually stayed away from as much as possible because you can if something went wrong, you could never figure why. it was like it was like it was like banging on a pinball machine to sort of shunt in the right direction and while the whole world is becoming pinball machine, as we're deploying a.i. because there's things i can do that you can't. so why did i coin a new word? because much of this loss of control stems not from systems in and of themselves, but the fact that we are inputting so much human data into in a way that we can't screen or filter the air, all these eyes, all of these chat ts consist of being trained on inordinate amounts of human generated data, texts,
1:34 am
videos, whatever. there is clearly of this data that can ever be pre filtered, you can try to eliminate in the most in the most coarse grain of ways. but the whole point is that these systems are bigger than than any human or any coordinated human entity can actually screen. so what does that mean then if you're if you're putting in this human data? well, what's really happening is you're getting these systems are combinations. both of these human elements that i think jacob is talking about in terms of, you know, that are getting mirrored in these systems and and opaque systems that are often ai related. they don't have to be a related because even if you take the ai component, there's still large enough that, you know, even a facebook basis, you're seeing stuff on facebook that clearly, you know, shouldn't be there. and yeah, even with human moderators, there aren't enough them. so i coined the term mechanism
1:35 am
to refer to that assemblage of. humans and computers interacting at lightning speed in a way that in effect even the companies themselves that administer these systems cannot catch up with them and we are forever closing the barn door after the horse has bolted. and it's not completely hopeless. but i do think have to reframe how we look at it. hopefully we'll get to that later. thank you, david. our next conversation panelist is calvin de lawrence and his book hidden in white sight how i empower cars and deepen systemic racism. calvin, a distinguished engineer at ibm, a member of ibm's corporate air ethics board as well as its academy of technology. he holds numerous technology patents. he's a global speaker and thought leader in the field of ethical and responsible artificial intelligence.
1:36 am
he serves on multiple corporate and, nonprofit innovation and technology boards. he's also a proud of clark atlanta universe city, where he earned a bachelor's and master's degree in computer science with a focus. focus on ai. calvin now. welcome, welcome. thank you. wow. i when i think about the the impetus that caused me to write book is quite personal. to be quite honest, it was probably if i wanted truly be honest. i don't think i've said this personally, but as as she just mentioned, i got my master's degree in computer science and a mid-nineties with a in a i with a specialty air. so you're talking about the and from that time we didn't do a whole lot of i, i it was it was a lot of, you know, i kind of stuff, but it was in order to
1:37 am
say that you are an air person, you was a geek. so nobody raising their hands to be an air person. but i guess about or eight years ago i became a big around facial recognition and systems. so i was at that time for a facial recognition smarter cities organization. so when i make this statement guilt i've built lots of these systems that we're about. i'm a hands on is what i do. i mean, i'm not a not a reporter. i'm a journalist. i'm practitioner. i work with, apply, computer science and intelligence. so when i kind of think about it, i think it's important to really maybe level set. well, just a definition. i mean like what is a i mean i really is is the science of of of basically making computers that can think, judge reason and
1:38 am
learn like a human and the key there the key differentiator that's from all the other programing languages technologies that we've used is this ability to reason and to learn to mimic human behave. so therein kind of where my guilt came from because i start running into people that look like me who was having issues. one such example is and i say in the book i actually start my book off in a practical and before i give the example, i would say that this little small book was decides on because i didn't write for technologists i it for my mom i wrote it for people weren't technology focused at all because i knew what was happening because. i've built these systems, so i wrote it in such a way that i story told. i use examples of real people friends of mine that had stories
1:39 am
a.i. and you. so you got to know my world right? the world that i came in everybody. i'm in computers, but they don't know what do they think? i put memory in computers and i couldn't do that if i if you make pay me just one example i was at my mom's house and my mom is 80 years old. i guess it was years ago, four years ago. and phone rang, her cell phone rang and i my mom getting i could feel her getting on the phone and the voice on the other set on the other side said. hello, mrs. lawrence called her by name and she said yes, this is mrs. lawrence and bot on the other end of the phone, tell that my mom was a black person. we can do that with a voice recognition system. it could tell that she was in a particular age group. it could tell where she through
1:40 am
location tracking. and if you can tell where my mom lives, you know, her income level. so you know all about my mom and all she's told you was, yes, this is mrs. lawrence. and the goal of this application on the other end a utility company that wanted to my mom a bundle service so i'm here and my mom on the other end and because is what do i got the and i know that one way with these with chat that we build by the way if you hit zero it typically takes to a human person and so i got the phone, i hit zero and took it and it took me to a human person to person, on the other hand, did not know the person who had been speaking with my mom a chat box. they didn't even know. but i knew because i've built the systems, built systems like that i thought that was
1:41 am
interesting was kind of a inflection, if you will, because this is my mom could have probably lost her house, certainly couldn't. the system that they was trying to her and i asked a lady on the other end, you know that my mom that was a chat box. now my you got to know my mom. she would never tell you she's black. she would never tell you her age. she would never tell you. she just wouldn't do that. but the person was sponsoring the app needed to know in order to identify what bundle she was most likely to use. so since like that, i talk about differences. so i talk about a friend of mine who will educate it. he went to morehouse college, atlanta. his wife went to spelman college, atlanta. they moved new jersey, had a house in jersey or condo in new
1:42 am
jersey. and moved down south and was going to buy a house. and they they kind of knew, you know, we know how to get a house. got to have good credit, got have a low debt income ratio. you to have a job you got to be able to thereof. so they knew it was already pre-approved and they had a closing date and right before the closing date they got a call saying that there a problem with the underwriter. long story short they were denied last minute. well, the said well it was the algorithm well my friends know i'm in computers so got the first call what's on calvin. well i know exactly what's going on because i know mortgage companies to replace underwriters. that's what i does. i told you earlier, i teaches judges, learns, mimic human behavior so you can replace
1:43 am
people and become more efficient and make millions, millions and sometimes billions by doing that. but the algorithm, what is capable of, not just looking at those four elements, it can look at all of stuff. so it can pull data from sources that you would not know. as an example, if you're a bank loan in money is what you care about. so if someone has a felon that's living in their house, if you can get a hold of that information that matters and the risk for, well, that's kind of what happened to my i didn't know what was going on but i did and having conversations i told them why didn't they approve you. it was some bogus reason she hadn't been she was a contractor, hadn't been on the job long enough. she had been on the job for two years. that's long enough. right. so i told go back and ask them
1:44 am
why that the algorithm so there's no transparency none of these examples i gave you. nobody has broken the law. there's no law passed anti-discrimination lives doesn't exist when it to this stuff we're talking about they were actually derived prior the internet they hadn't been changed so that was that guilt question. so i realized then i have another story that became personal. i don't have enough time to say that story, but it was a medical situation that happened to me and it was an algorithm that my issue. so i went back and forth on it because i'm like i'm a first generation generation college student, first person in my family to go to college, i need a job and everybody knows i need a job. so the question, do i speak out on this topic? do i actually write book and i
1:45 am
you know what? i'm going to write a book and i'm going tell specifics of things i've seen and examples and these distinguished colleagues here, i know read their writings and i know what they're saying is true because i've built lots of these systems, ones that i didn't built i know exac ugly how they're built. i'm seen examples after examples where this have come in come up and i know many times it's about profit it's about this whole moral of profit versus social response, stability. and how do you win if you remember in the nineties i mentioned nobody was building these systems because that a profit model the minute become profitable to do so you kind of see and hear what you have today. i'm excited to be here thank you calvin. so, guys, we talked little bit about this. we wanted it to be a very conversational panel and like to
1:46 am
move into that right now and just go to your immediate and issues with we should be paying attention to right now. i mean everybody i think you referred to it as a terminator. everybody loves that big thing. will they replace us? well, they need us, but and is, you know, a way the future can go but this call in the open letter to say that we needed to have people this and who's managing it and who's control doing it. i think this is the conversation. love to hear you. yeah. you all talk right now from the loops. but but just you know, what should we all be paying attention to right now from all of your what's main thing that each of you would say? i mean, i'll i'll throw mine in and i'm close here, what you guys would say. but mine, you know, mine is there's a huge amount of. there's a huge amount of lobbying money spent on behalf of the biggest tech companies. now to get in front of lawmakers to, try and guide their thinking
1:47 am
about how a.i. should thought about and regulated and the there was a forum that chuck schumer hosted recently which the top tech ceos of the world essentially were brought and a single academic deb raji from uc berkeley had to be there and basically speak for all academic vcs in the face of these dudes all dudes you know they're and what they she told later that what they wanted to talk about was agi this idea that artificial general intelligence, this terminator scenario is going to enslave us all in the future if we're not careful, we don't get out in front of it. and what she and so many academics over the years has been articulated me is that is maybe a theoretical down the road but what it really is is the problem that tech ceos are most comfortable about. they would much rather talk about that than talk about the kinds of harms that carbon is describing or talk about the profit motive and the of out of control quality of companies that david has chronicled so
1:48 am
well. and so for me, i it is not getting distracted by that debate and instead trying to force the conversation back to some really big, i think, fundamental questions about who owns what data, has it been collected, what's going on. so like right now the writers, as you know, has filed suit on behalf of 17 authors and to try to get openai stop sampling their work and to come up with some kind of agreement for paying them for their intellectual property, which they say has been violated by chatbots, because it can summarize work of these authors john grisham and george r.r. martin. george r.r. martin and these folks. right. so if you look back at the way that google books won its fair use case a while it it argued before supreme court that it should be allowed to exert out a
1:49 am
certain amount of writing without it being the whole thing and that that in some way counted under fair use and and meant that you weren't replacing the commercial value of these books. now that same of book information in is enough without having to have whole book to train these a.i. models to basically imitate the work of those people and ask a george r.r. martin asked chatty betty to write a new george r.r. martin book paragraph by paragraph, and it'll do it for me. there is this continuum. how these companies have built these and huge amounts of data from typically social media and search those same companies own all of that are now the companies that have these big foundational models google met a microsoft with open i and what i
1:50 am
think is so is i was sitting in front of i was sitting eric schmidt interviewing the ceo, former of google and his argument is he he much more wanted to talk about agi than anything else and he said that the reason that he wanted to do it was that because it was because that search engines sorry, that social media he said i was years ago i was tremendously naive about social and what it might do to our elections and our electorate. the state of democracy and the rest of it. but at the same time he argued in the same conversation that this time you can trust us. this time we can relax. we got it. this time we're on top of it. and that it is the technical expertise of the people who build systems that is going to teach us the most appropriate way to regulate these systems. and so for me right now, there was a piece of regulatory capture on on behalf of these companies to to steer the conversation directions that they want to go. for me, that's real problem. so that's one thing i think about is don't get distracted by
1:51 am
agi. let's focus on these short term harms. and then the second part is how are people going to succeed in lawsuits? again against these companies? because while i do agree with your fundamental thesis that there is this of out of control quality and forgive me if i butcher your thesis here but but know that there's this there's a mega net is taking form right that sort of is supra corporate supranational in a way. at the same time, i think that that is all driven by incentives and those financial incentives think can be shaped and typically have been shaped in past. we've seen it with cigarets. we're seeing it increasingly with guns by lawsuits. and i think that basically, if you know, it used to be like for about three or four years ago, there was a survey went around that asked ctos at these big companies do know how the ai models you are using and the vast majority of them had no idea. and then it also asked do you care and and an even greater
1:52 am
number of them did not care and that is a signal because they don't have to it didn't cost them anything to care at that time. now i'm speaking to i was just speaking to a young trio of lawyers who start a new firm and their whole business is to go out to fortune 500 companies to explain to them, here's you're going to get into very serious trouble when a chat bot calls carmen's and blows up her life. and it turns out that it was on the basis of some sort of protected legal class like it always is. and i think that's the challenge, right? it is. and i mean, i the fundamental approach in my is i coined this as it relates to air culture, systemic racism, racism, the institutional component of our society. and that's why took the time to give that definition because people forget what is all of the hoopla of the noise and what a i, i that's what it does it mimics i behaviors.
1:53 am
and if we agree that there's this thing called bias, there's this thing called discrimination against not people of color, but all types, women of types of if we agree that that is the case, then obviously the systems who are being designed and developed by us giving them data will have at least have the of doing it. so to your question, i don't get up so much on this whole apocalyptic kind of viewpoint because situations we're talking about are happening right now, like i'm coaching in sandy applications that we're talking about being designed and developed. that's not 2080. i mean in 2000. you know it's not a generation from now it's right now these are the things that are happening so jake said right i mean it's kind of a bait and
1:54 am
switch in my mind because you want people want control and own the narrative because in the reality it is that this is a very difficult issue to fix and you kind of look at who's getting hurt. i remind people at my company i remind people at the end of our lives, no, it's like real people. it's like real people on the end of this. so people are actually hurt. they're being further disenfranchize and further, by applications. i'm a huge fan of i mean, why wouldn't that be? is still the industry that i'm in the ideal is that that computers i'm not afraid computers i'm afraid of the people who are sponsoring and i've seen it with my own teams. you know again i personally have never met a racist.
1:55 am
they might exist, but i just never ran in to them. it's like you don't know what you don't know. you just don't know if somebody is giving you $5 to build an app and you have to build it in five days. you build app in five days, whereas the person who's giving you $5 should have given you $25 to build it correctly. they won't do that cost. is this conversation of profit versus social responsibility. so even though i do believe that, things could get worse, it always does. when you disenfranchize one group of people, it comes back at your front door. sooner or later you're at a victim, right? so i do believe. in these scare tactics, i think going to happen. but right now will only happen if we sit back. we don't do anything, you know, in regards to like the legalities of is that there have been plenty of cases these examples where they people have
1:56 am
sued they've gone to court each and every time a person loses and black we won't take it to court because. we don't know what just happened to us and we don't have the money to adjudicate it. so therefore we just go on to next loan officer. we go on to the next situation. that's just the kind of way works in a community that i'm a part of. right. so, you know, you have two legal concepts. one is kind of disparate impact. the one is disparate treatment disparity. impact is some algorithm did something you but nobody knew they didn't know what it being done disparate treatment is they know that it was being done every. one of the cases thus far has been at minimum at best through disparate impact i mean it happened to you but nobody it well in some cases they do now many of these cases we talking that we talk about they do now. i mean there's a there's a great very vivid hypothetical that
1:57 am
calvin presents in book about the sort falls out from what we've been talking about, about let's say we have an automated driver, a like an uber automated car, air driven and has to choose what to you know, what what to crash into and. it is less to identify a black person as being human than a not than a white person. and thus makes that choice. and yeah, that seems to is that that's yeah there's disparate disparate impact for you and and it is seems likely that we would say uber did not in hand that outcome. i don't think that they sat in a room you know engineering that particular outcome. but as you get these systems multiplying and being conditioned and trained in opaque ways, that just happens
1:58 am
that, that pops out of it and. it is a very it is a very difficult, time consuming, costly process to try to audit these systems to prevent outcomes like this. in fact, i think it becomes even just to enumerate all the possible disparate outcomes that we've just named one, we've just named a very a very vivid one. but one can imagine every other possible of comparison. and we could say, will an automated car pick this over this? what is it, to be fair. so the you go, the more questions you raise. okay. you can literally, you know, circle this endlessly and get into real like epistemological questions of what actually constitute objectivity, what would actually constitute being not biased. we know hold this up as an ideal we seem to realize that it doesn't exist, but it seems that
1:59 am
neither humans are capable of implementing it, nor have we been able to cause to create that, implement it. and i've had someone say in the same conversation, well, you know, humans are so biased, that's we need computers. computers are so biased. that's why we need you and. you get into this loop of, okay, well you know, who's, who's, who's who's who's watching the watchers and well the issue is it's i there aren't enough to watch it. and if it's humans, you know, we've been living with that for a while, but. well, i mean, i don't want to, i don't, you know, i want to make make the point that certainly, you know, the fact that these to reiterate the fact that these systems are somewhat out of control does not not in any way remove responsibility, get them off the hook. i think at the least, companies have an obligation. be more honest and be more transparent.
2:00 am
say, expose. you know what training data is being used on, these systems at least, so that it can be so important to note none of those companies are doing openai in spite of its name. not let anyone see what training data is. and this is part a more general issue that research has moved out of academia into corporation is because these systems rely on such a massive training data, such a mass of systems, they they gain access in particular gain their ability from sheer size that and that's sort of a new thing in in computer science. you know in the 20th century, people did not think of, well, if you just feed something enough, it will magic you'll magically get emergent properties and yet that is actually basis of so much of the technology today even before the age of deep learning ai from which we get such empty, you
2:01 am
know, even, you know, the google search engine gave many of its abilities because was so much data being put in and the peter norvig, the a.i. expert, was with that google and he wrote a paper called the unreasonable effectiveness of data which was that did seem that if you at a certain point you pour enough data into something think there's a there's a rule that you need about 10 million examples before a machine learning before for machine learning supervised learning. yeah ai to sort of start to perform. and the fact that there's that threshold itself is just this bizarre thing. it suggests that these a's are actually making this movement towards being more and more brain like when i say brain, like i don't mean human or cognitive, i just mean that that they are becoming more like black boxes, but the example that you'd want to bring up to
2:02 am
to both compare and contrast was india has deployed a national program called aadhaar and i just want to say, how many people have even heard of adha not? too many. okay. adha is really paying attention to because i think it is it is where things are going. think bolivia has tried to deploy something similar. india was in the situation having a of going from very low tech to very high tech in a short time so they didn't have as legacy systems to replace but adha is your single government identify it's your driver's license it's your social security number it's it's your benefits card it's all of those things in one single number and it's you know, this was pitched for highest of reasons by the by the indian government was not pitched as being a profit motive. it was pitched as being the benefit and efficiency of of the
2:03 am
nation. and it was pitched also as being completely voluntary. however, 20 years later, it is all but mandatory. you are, you know, you can think of it as like, well, yes you can be offline right now, but definitely going to make your life inconvenient. well, even more so with adha, getting a bank account or cell phone is, more or less impossible if you have not signed for an adha number number. and what has been happening is the simple interval of all of these systems, both government and non-government, because all these private companies are now piggybacking of the adha number is leading these unintentional consequences of, you know, at the very least identity theft, an unprecedented scale because when we of identity theft. well one aspect of your identity stolen but you know even if you get my social security number that doesn't necessary give you access to many other aspects of the ad our number of that is if you can crack the ad hard
2:04 am
biometric id which has been cracked. as a matter of fact, you now have that person's entire identity not to mention that if there is some mistake in that amorphous corpus of data that swirls around because it's not centralized it's across many databases. the actual aadhaar database is minimal and it was designed that way for privacy purposes just backfired because what then is that as long as you've got these linkages now, you have these other systems with the same number that you can now tie to them. and this is actually one of those things that happens with how mom was identified, that all of this data i think there's a there's a frank pasquale talks of he calls it runaway data, data, promiscuity, things like that, that this data is just running all over the place and even it to you guys, right. i mean, we give data freely. yeah. and you're calling it a large language for a reason. it's a lost language model. it's different. like people get caught up. i think sometimes we again another bait and switch that i
2:05 am
and somehow gpt is same thing it's not gpt is a large language type of a.i. it's not like so these many of these example that i talk about or that we're talking about is not large language models at all, like language models skills, just problem. because in the real world of air, if you look at 98% of the applications out there, they're not large. they are very kind of supervised type language languages, if you will, whereas gpt is unsupervised. and what that means is that we're data and we're not really labeling all of the data in the real world these biases that we're seeing is like they're supervised. people are just labeling it incorrectly based upon who they are and should be called semi supervised enough, right? so imagine getting scrubbing the internet for lots of data because you got to have lots of data to do large language
2:06 am
models. you can't do it with like one little group or one little small data set. it doesn't work. so you got scrub the internet, you got to scrub stuff that. people maybe don't even. i give you an example like zoom. we've all use zoom. like if you notice, one of the policy is zoom. just put out there and they took it down. they say, hey, if you're going to use zoom, you got to freely give us your data, meaning that all of your conversations, all of your and photos you have to give it to us now seriously. it did that and they need that because in order for it to do large language models, they need lots of data, lots unstructured data. so we quickly it to them, we say, hey, i'll if you i'll give you insurance companies say, hey, i'll give you 25% discount if you take this idevice and put it in your car and let me track
2:07 am
you into office, like i'll raise hand for that. give me the 25%. so we readily give us the data because they need it in order to do the deep analytics they want to do, they got to have data. so they sent us many times unknowingly give them to data. i know when we build a systems, i go to a and first thing i want in a way is your data you can't do air without data. the more of it, the better we can do, the more accuracy that we have. it's like i use this example, the book because air is out about probability t like imagine having a a bag of jelly beans and you got you got a hundred jellybeans in a bag and you got 90 white jellybeans and you got ten black jellybeans and you shake them up and you close your eyes and you reach in a bag and get a jelly out. what's the chances of you getting a white? because this is all about
2:08 am
probability, that's all about probability. chances of you getting a white jelly bean is 90%. you purposely put 90 white jelly beans in the bag. that was your data set. you didn't use any type of logic in regards to how many beans that went into the bad. 90 white went in, ten black went in chances. you pulling a black jelly bean? the bag is 10%. so you bring the police out and jelly beans are your your police cars and you put all 90 cars in a black neighborhood. probably don't get a lot of black people some crime. the air is going to predict that for you. oh, so that's just the reality of how this systems work. you know, i'm going interject just because unfortunately, we our time is almost up and i just
2:09 am
got started. i know i know this and you'll have to read the books because there's just so, so much in them. and i just want if everybody could take like just 30 seconds and say one thing about that you love about what i could do for us because should temper it's here you know it's it's that person it's that other one that's living in our homes with us and so besides controlling it and having human agency understanding things. david, one thing you'd say that i do love god or if you want to say, i love it, but i'll let you know. mean no, i will say i mean the you know if i look at church it is amazing it opens up vistas of you know possibility for human society and organ asia organization that are literally unprecedented, that we have never seen before, that is fascinate. it's also terrifying because. you're amplifying human
2:10 am
tendencies unless you have the most pollyanna ish of views of human nature are very terrifying. but you we are looking at, i think, ways of people joining joining together. and i will say that, you know, one conditional positive is to the extent that do believe that the sheer complexity and opacity these systems i and even a i related to everything that i put under the the domain of mega net. to the extent that they are removing control from any centralized source perhaps. that's good if nobody has control everybody has a tiny bit of non decisive control. well, i mean it may be chaotic, but in actuality, you know, i mean i don't know is it you've got the robber of the 19th century versus, elon musk and mark zuckerberg actually having a cage fight today?
2:11 am
is it better that they are they want to have the cage fight instead of? you know, hiring peons do the cage fight them, which they certainly want to. i don't know. it is so, so there's always ups and downs to it. i will say also that, you know, applications of ai in terms protein folding, in terms of pattern, in contexts that sort of aren't as related to the socio political context we talking about today are thinking, good thing you've got a way to process and pattern recognizing data you've never had before. it's a little cheap to say that. that's something i love ai because it's like saying, oh, okay, you do something else, something else with a i that we been talking about. i like it, but i will definitely say that, you know, a.i. techniques certainly hold tremendous amounts of promise in in other contexts in which some of the issues we've been talking about today, you know, don't don't, don't really. that's you know, so i've encountered in my job shown a huge number of applications of ai that stand to make an
2:12 am
enormous amount of money without actually solving any problems. and then ai and then i bump into a, you know, like, let's replace screenwriter as writer, let's replace actors. and then on the other side i'll see huge amounts applications that could solve enormous problems, finding social services for people who are living unhoused, you know, connecting, lost of the archeological trail to connect etruscan pottery to roman. and you throw enough of that and can tell you what should be in between like cool stuff. we're not making money on that stuff, right? so for me, i'm a big fan of that stuff. i love the nonprofit of this stuff. and there's a couple of sort of heroes in my book who are who people as a matter of value as won't apply what they know to this stuff and will only work for social service and state health departments and so forth. and so i'm really dig a lot of those applications there is in the middle a little bit of problem solving that is also profitable and that's, i think, medical protein folding, cancer research.
2:13 am
there's a lot of and interesting stuff that's coming that i think doesn't a lot of that. one of the biggest problems, however, with that is that that data is white people data over, you know, by and large and there is such a gap in the of of the rest of humanity when it comes to that kind of data and especially as we get into genetics research, which is taking off and i think we'll all be sitting here talking about that in two years, you know that data is also highly skewed. but for me really love the simple nonprofit altruistic problem solving that i can make possible and can make quite affordable. unfortunately, the foundational models and the companies that control them are very much about making money and solving problems that i think don't actually have to be solved. that's why matt galvin, if close us out with a i love everything about yeah i do seriously i mean i've seen i know we caught up on the bad examples. i know i talk about them
2:14 am
specifically, but i also talk about the great examples. right, for every one bad one there's non life saving, life altering, life changing application for a i. i don't think the problem with the technology i think is conversations to me is a good about a.i. because it reminds us who we are as people like all of these issues. we've talked about are not technology issues people issue. every example that we talk about i kind of heard earlier a couple of days ago of a friend of mine said that there wasn't issue with splitting the atom bomb. the issue what are splitting the atom the issue was building a bomb right so for me i don't think that they're like i don't want to sit on this stage and say, hey, down with i know i got to keep moving. we just need regulatory
2:15 am
requirements, need people, we need consciousness, we need a mindset, we need folk to start looking and spending money and taking the time to ensure that the good that we do. i is not pepper with the bad right and i think we can do that. i'm being positive. i'm very optimistic that we as people we will find a way to to do i mean i does old things better and it help us do things we didn't know. we can give all types great examples of how i can help and i think it will. so again our time is up and i want to thank jacob ward david auerbach calvin lawrence and they will be out signing books. thank youi'm absolutely delighte here with these three

40 Views

info Stream Only

Uploaded by TV Archive on