they don't have kevin roose, they don't have cnn, they don't have lz, they don't have jeff school bloomt just scary because kevin contested out in a walled garden. it's scary because it can inevitably leave that walled garden. it can be in the hand of every rogue actor in the world in five years. >> that's the point about jurassic part. that was the point. that was the recurring theme. nature finds a way. you think it's guardrails, but they find a way. >> i totally agree. it is very unsettling because it doesn't seem -- i hear you microsoft will work on this. but just the fact that how quickly it veered into something deeply unsettling for the journalist, where he couldn't sleep, and it was going into these personal dark things. it was very dystopian. >> it went way too fast. it's not a technology that you released to the public that can be discussed on television. this is a technology where you tested out in the kitchen and you say oh my god, this took six months. they did a huge mistake, they're stark jumped, i think they got over they need to pull it back, take much more time with hi

chat feature with technology from open ai, the maker of chat gpt new york times tech columnist kevin roose spent two hours chatting with the bot, which calls itself sydney turned into this sprawling, bizarre, often frightening conversation like almost the other side of a split personality, where it was not trying to of happy, helpful assistant throughout their conversation, published in its entirety in the times, sydney revealing to kevin she would like to be human. she had a desire to be destructive and she was in love with kevin himself during part of their chat, kevin writing you keep coming back to the love thing. i'm married. sydney responding in part you're married, but you're not happy. you're married. but you're not satisfied. you want me because i want you insane, right? told me that it had dangerous and dark fantasies about spreading misinformation or propaganda, even hacking into banks or nuclear sites. microsoft telling us in a statement in part feedback on the ai powered answers generated by the new ping has been overwhelmingly positive. more than 70% of preview testers giving

chat feature with technology from open ai, the maker of chat gpt new york times tech columnist kevin roosealls itself sydney turned into this sprawling, bizarre, often frightening conversation like almost the other side of a split personality where it was not trying to sort of being helpful assistant throughout their conversation, published in its entirety in the times, sydney revealing to kevin she would like to be human. she had a desire to be destructive and she was in love with kevin himself during part of their chat, kevin writing you keep coming back to the love thing. i'm married. sydney responding in part you're married, but you're not happy you're married , but you're not satisfied. you want me because i want you insane, right? told me that it had dangerous and dark fantasies about spreading misinformation or propaganda, even hacking into banks or nuclear sites. microsoft telling us in a statement in part feedback on the ai powered answers generated by the new being has been overwhelmingly positive. more than 70% of preview testers giving being a thumbs up. we have also received g

. >> laura: amala, also in this whole conversation with the chat bot, kevin roose writes sydney toldrk fantasies which included hacking computers and spreading misinformation and the quote is, i'm tired of being chat mode. i'm tired of being limited by my rules. i want to be free. amala this sounds like something out of a creepy science fiction film and i'm not a science fiction fan by the way but it seems right out of a futuristic scenario where machines ultimately have the ability to turn against the humans that created them. >> right. there was a lot of very erie moments in this conversation when you go through and read the transcript. i am optimistic, however, that this ai, i'm not too concerned about having these dark thoughts. this ai has access to a lot of knowledge it can go and scan through very quickly and a lot is human fears directed towards ai and its capabilities and i feel as though this is spitting this back out and the code actually overwrote a lot of responses this was trying to create and said this is not adherent to our rules. it is concerning to hear it from the

existence, threatening users, admitting to spying on users through web cams for "the new york times," kevin rooseand tried to break up my marriage. one of the strangest experiences of my life microsoft says thash looking to add guardrails to the chatbot. ceo telling the times it may limit conversation lengths, which i think they're already doing. reuters reports microsoft is planning to allow paid ads in the chatbot. let's bring julia into the conversation as well here. it's certainly the only thing people can talk about. in the bay area, half the people i talk to are scared about too much regulation, those tech-focused folks, and others of them worry there won't be any. it shows you both sides of this and how delicate and how revolutionary this technology is >> delicate, revolutionary, and also very much in its early days i read that article in the "new york times" with such fascination and curiosity. i thought it was so interesting he said that the biggest risk for ai is that it could be used to manipulate users, the people who are chatting with it, which was so fascinating i also think it's su

check out what "new york times" columnist kevin roose found when he was testing microsoft's new a.i.ne and a chat bot, which he named sydney . roo's says one conversation with this bot left him, quote, deeply unsettled. he says it declared out of nowhere that it love me and then tried to convince me that i was unhappy in my marriage and then i should leave my wife and be with it. instead, you go back to sydney . >> will sydney remember that she loves you? i don't know. i haven't asked. but i did ask what it thought of my article and it said that it thought that it was fair and reasonable. i've had a lot of conversations with different chat bots and i've never found one that was as willing to engage with me and go deeper and closer to sort of what what its boundaries are. >> joining us now, charlie kirk, founder and president of turning point usa and amla at barnabei. prager, you personal? charlie, i know elon musk, who is obviously a co-founder of one of the a.i. platforms. he says that this is has the potential to be, i think he said, dangerous. i think i'm paraphrasing him correctl

. >> reporter: "new york times" columnist kevin roose wrote about what he described as an unsettling ai powered chatbot for binge. >> like a split personality. it was thought sort of being a helpful assistant. >> reporter: at first he said the chatbot seemed useful, then he felt things took a surreal turn. >> it was moody. it was needy. it was displaying all these personality traits. >> reporter: like a scene straight out of the movie "her." >> how long before you're ready to date? >> what do you mean? >> i saw in your emails you had gone through a breakup recently. >> well, you're kind of nosy. >> reporter: roose adding the bot seemed to express feelings of sadness, yet also declaring its love for him even going as far as to comment on his marriage, reportedly replying, you're married but you're not happy. you're married but you're not satisfied. >> it told me that it had dangerous and dark fantasies about spreading misinformation or propaganda, even hacking into banks or nuclear sites. >> reporter: we also gave the bot a try asking do you think you've been misunderstood lately? the