"Godfather of artificial intelligence" talks impact and potential of new AI

this week Google jumped into the AI arms race with the launch of Bard the chatbot will compete with openai's chat gbt which is funded by Microsoft in a few short months these chat Bots have wowed users with the ability to answer complicated questions draft emails and speeches plan custom vacations and much more but as it improves this technology offers immense promise and significant some say existential danger Brook Silva Braga traveled to Toronto one of ai's leading research hubs to meet some of the people building this Brave New World so yeah so this is a an experimental preview of a chatbot we've been working on powered by our generative language model okay Nick Frost is a co-founder of cohere like Google and openai the Toronto startup has trained massively powerful computers on trillions of words and ask them to talk back can it write call it a book report about War and Peace okay it can War and Peace is a historical novel by Russian author Leo Tolstoy in a major leap for computing these models can understand and create natural language it just wrote this yeah but it's based on all the stuff that it's seen the tech had been in research labs for years chat gpt's release last fall cheese or cheese so luscious and bold brought it to an unprepared public it's writing me a poem about cheese and it's pretty good in the old school programming we would write code to tell a computer or what to do we still do that lots of people still do that but another thing we do now is we write code to tell a computer how to learn what to do and the way it learns is being shown many examples of what it should do so in the context of large language models the way that works is we get a huge amount of text and then we show it a few words and we get it to predict the next word this simple technique turns out to give you something very useful and very powerful so powerful that openai's new gpt4 does better on the bar exam than 90 percent of aspiring lawyers these models write computer code in seconds and churn out blog posts on any topic that AI programmers haven't banned they're becoming the newest Arbiters of what ideas are out of bounds and while chat apps are the current rage other forms of AI now you can generate a video with nothing but words we'll do much much more this is the biggest technological advancement since I think is comparable in scale with the Industrial Revolution or electricity electricity or maybe the wheel or maybe the wheel yeah Jeffrey Hinton is known as a Godfather of artificial intelligence for the last 10 years he's helped Google create Ai and Mentor the industry's rising stars including open ai's Chief scientist Ilya siskova and Nick Frost over at cohere in fact Toronto is a global AI Hub today in large part because Hinton moved here 40 years ago when the Canadian government agreed to fund his unusual research I was kind of wig because I did this stuff everybody else thought was nonsense While others pursuing AI tried to program logic and reasoning into computers Hinton thought it was better to have them figure things out themselves the idea was to mimic the brain with lots of practice these virtual neural networks Illustrated here would make the right connections to solve a given task there were doubters the big issue was could you expect a big neural network that learns by just changing the strengths of the connections could you expect that to just look at data and with no kind of innate prior knowledge learn how to do things people in mainstream AI thought that was completely ridiculous it sounds a little ridiculous it is a little ridiculous but it works it's all the real only in the last decade or so have computers been powerful enough to prove Hinton was right his machine learning ideas applied to different kinds of training data now create all kinds of outputs this of course isn't Tom Cruise but a deep fake impersonating him I went down to the office because they're making a robot of me this isn't Andy warhol's voice in the recent Netflix documentary but a clone generated by resemble AI click record and then read the sentence out loud after a few minutes income between reading training text it'll go ahead and build us in the background founder zohab Ahmed had cloned my voice too Habitat for Humanity which helps homeowners build homes alongside volunteers yeah I definitely hear myself in there smile into the camera a company called Synthesia had me read a script in front of a green screen they used that video to create this digital version of me we paired it with resembles voice to create a TV reporter who will say anything you type the technology will only get better in the years ahead using this Tech to spread misinformation seems inevitable this is my very first day in zingwana's agency using it to replace journalists and lawyers and accountants and Radiologists and novelists and songwriters and Painters well that could happen too but it's going to take a whole ton of jobs I think it's going to make a whole lot of jobs easier and a whole lot of jobs faster I hear your answer in two ways one wow it's great if it turns out that way two maybe it's a little scary if the people making this technology haven't come to graphs with the potential downfall of what it could do yeah I think we try our best to think about what the true impact of this technology is in the true impact of these advancements is the subject of a high-stakes debate right now will AI quickly Zoom past human abilities and become what's thought of as artificial general intelligence or AGI last month open AI CEO Sam Altman wrote the risks could be extraordinary a misaligned super intelligent AGI could cause Grievous harm to the world but Frost and others say Terminator style worries are overblown large language models are just algorithms that write some smart sounding words they don't understand even basic truths what day is it it's Monday it's Monday it's just yes it's just guess yeah I don't think the technology we're building today naturally leads to artificial general intelligence I don't think we're close to that his mentor used to agree until quite recently I thought it was going to be like 20 to 50 years before we have general purpose AI now I think it may be 20 years or less some people think it could be like five I wouldn't completely rule that possibility out now and whereas a few years ago I would have said no way are we close to the computers coming up with their own ideas for improving themselves yes we might be and then it could just go fast that's an issue right we have to think hard about how to control that yeah can we we don't know we haven't been there yet but we can try okay that seems kind of concerning um yes what do you think the chances are of AI just wiping out Humanity it's not inconceivable okay that's all I'll say how Hinton wonders will we manage a technology that could give a handful of companies or governments such awesome power and so I think it's very reasonable for people to be worrying about those issues now even though it's not going to happen in the next year or two people should be thinking about those issues for CBS Saturday morning still writing his own script for now Brook silver Braga Toronto good for you Brooke yeah yeah right but two questions like why are you building it one and two why aren't you thinking about the possible impact it's the it's this big question mark and and and the rest of us have to deal with to try to make life easy I thought the bigger question was when Brooke asked what do you think of the possibility of AI wiping out exactly humanity and the guy says it's not inconceivable yeah exactly I mean my point exactly right I don't think we can answer that in 10 seconds

Channel: CBS Mornings

Views: 140,081

Keywords: CBS Saturday Morning, CBS News, video, geoffrey hinton, artificial intelligence, technology, chatgpt

Id: uG2SKIdLUj4

Length: 8min 17sec (497 seconds)

Published: Sat Mar 25 2023