AI pioneer Geoffrey Hinton leaves Google, citing "profound risks to humanity" • FRANCE 24 English

Chris O'Brien is the editor of the French Tech Journal thank you for speaking with us on france24 great to be here uh the quote we're hearing of Jeffrey Hinton from The New York Times is that he says artificial intelligence takes away The Drudge work but might take away more than that what does he mean well particularly in this case when we're talking about the technology known as chat GPT this is AI that goes from just analyzing a lot of data to actually as the name suggests generative AI it generates content and so it's able to generate texts videos sound images that in some very basic forms can replicate what people like me do people do any kind of writing people creating any kind of imagery so some of that on a very basic level that's churned out at high volume it seems like chat GPT could be a replacement for that do you think your job could be in jeopardy uh I lose a little sleepover yeah I mean if you've ever played with these tools and I really suggested everyone uh tune in a little bit and and take them for a spin uh you'd be surprised and I was surprised if someone even who's been covering tech for 25 years just how powerful some of them can be so there's this seminal moment when um he worked for Google but the Rival Microsoft is the backer of uh of that chat bot that you mentioned chat GPT and when it starts appearing on their Bing search engine that really scared him yeah and specifically in this case uh you know his concern there you mentioned Google Google has their version as well Google bard and so he's basically saying he's fearing that this turns into a bit of an arms race because Bing's search engine has been a distant number two for a long time uh the company's tried to change that not had a lot of success now it's injected this AI into it and it seems to have supercharged it Google's a little freaked out and so his fear is well all the guard rails all the caution are going to be thrown to the wind and it will become this arms race that uh as both these companies sort of battle it out now a lot of comparisons have been made with Albert Einstein the father of the atomic bomb Robert Oppenheimer expressing their remorse over uh their uh work that helps create these weapons of mass destruction uh is this the same case is he talking after the fact the way they were well to agree yeah and I think it's even a comparison with Robert Oppenheimer that he invoked a little bit you know he's that notion that as a developers and Engineers a scientist you get so caught up in the possibility of making something happen of making this break for you breakthrough you don't take a breath to think should I be doing this you you just get caught up in that sort of frenzied uh attempt to realize a dream and then afterwards you sort of see what you've created obviously in the case of Robert I'm Robin Robert Oppenheimer that was quite dramatic but you sort of see it and you think oh my gosh what have I done you know in that New York Times uh story that we that you referenced on air uh what really struck me was his quote saying he's seeing things today that these tools can do that he thought we were 30 to 50 years away from seeing you know that really tells me that even someone who's been as fundamental to their creation himself has been surprised by this acceleration we've seen over the past year now Google made an announcements they had nothing to do with with um Jeffrey Hinton uh alongside Apple had said it submitted a uh the proposed industry specifications to help combat the misuse of Bluetooth location tracking devices for unwanted tracking are we going to see more and more of these announcements and do they amount to well self-regulation if you will well I think it's sort of a classic strategy by companies when they kind of see the writing on the wall they know there's an anxiety out there they want to get ahead of it and kind of shape these rules uh I think you'll see that more and more with AI we saw Google and then Facebook kind of try to do that in the last few years saying hey we believe in regulation two let's have a conversation but suddenly trying to shape what that looks like but at the same time I think they also recognize and policy makers are probably having this anxiety right now it's very difficult to really shape policy around these things because they move so quickly let's talk about those policy makers Chris O'Brien as an American in the suburbs of Paris where does your faith lie does it lie with U.S Regulators European Regulators or is this just Mission Impossible well I would lean much more to the EU because I think they're prone to take a more proactive stance on these things uh so I think that's a check in the favor of Europe I think the U.S will talk about it a lot and not much will happen there but again there is a reference into your in your story about the legislation at the EU level about Ai and that's a good example worth discussing briefly they've been working on that AI uh regulation for two years and all of a sudden when the process is just nearing the end all of a sudden chat GPT becomes a big thing and they're scrambling to kind of adapt it to that because it wasn't on the horizon when they started that project and so I think there will be a little bit of element addressing that in the proposal but again by the time that gets debated adopted enacted even from the EU perspective who knows what the world is going to look like in a year or two years because what's the purpose of this uh legislation is it to uh to enact transparency or is it to to pick up on the quip you made at the beginning of our conversation to save your job well I think it's very much not an attempt to ban it I think it's really what you said about transparency so very this is a gross generalization but broadly speaking uh it sort of puts three labels around uh AI in terms of how much of an impact it makes and then sets out uh some transparency requirements around that in terms of sharing what elements are going into driving that AI creation so now again with chat GPT what they're trying to do is say okay uh you have to close disclose the the sore origin of any images or content uh to ensure any to avoid any copyright issues or reusing other people's work but again that's obviously a very limited uh guard rail against this type of Technology uh last week on the show we had a Democratic party strategist who was in Washington who uh Rude the fact that in his view looking ahead to their elections next year in the United States uh the Democrats were showing up uh to this fight with a butter knife and that the Republicans were way ahead of the game when it came to AI that immediately brings back the memory of 2016 when uh tech savvy people worked with the Republicans on micro targeting uh the more successfully than the Democrats did is do you see any evidence for that that there's one side that's better at it than the other well I would definitely say that conservatives have become much more effective at utilizing those types of tools in terms of creating memes creating virality across different types of content that spread very quickly across their media ecosystem um you know what's worrisome with this generative AI technology is it now really anybody can create this stuff it's as simple as typing a search into Google really you know you just need to come up with a few keywords and generate a photo or a video or text and it's very difficult to distinguish many of those things from the real deal and so I think I expect over the next year we're going to see an astonishing increase in the volume of that kind of misinformation disinformation uh for the for the next election and I think it'll be a shock to people and a real issue and you know again you get into that question of does do federal election officials try to get ahead of it are there rules that Congress could put in place to stop that it's it's hard to know but in terms of arming themselves I have to believe both sides are really have this on the radar and are trying to think how do we weaponize this against the other party Chris O'Brien the editor of the French Tech Journal thank you for speaking this year on France 24. pleasure to be here

Channel: FRANCE 24 English

Views: 25,710

Keywords: ai, artificial intelligence, debate, f24, france, france24, geoffrey hinton, google, info, infos, international, microsoft, news, tech, technology, us, usa

Id: PcPB6oEN7ms

Length: 9min 36sec (576 seconds)

Published: Tue May 02 2023