Exclusive: Inside the mind behind ChatGPT | Nightline

tonight we take you inside the headquarters of a small company in San Francisco called open AI creators of chat gbt even if you haven't heard of it your life could soon be impacted by this powerful new technology built with artificial intelligence just this week open AI unveiling its newest version which can answer complex questions in seconds it can write speeches take tests we learned it can now even pass the bar exam to become a lawyer placing in the top 10 percent they believe it could one day help doctors spot disease that the human eye or mind might miss the CEO Sam Altman is just 37. gbt stands for generative pre-trained Transformer that just rolls right off right on the tongue yeah we really named that well you know how does it work by ingesting a huge amount of text significant fraction of the internet this AI system can learn the underlying representations of what these words mean how they relate do a little bit of something that is sort of like reasoning sort of like understanding what you as a user want some of the time not always not perfectly and try to help you what changes because of artificial intelligence part of the exciting thing here is we get continually surprised by the creative power of of all of society I think that word's surprise though it's both exhilarating as well as terrifying to people I think people should be happy that we're a little bit scared of this I think people should you're a little bit scared a little bit you personally I think if I said I were not you should either not trust me or be very unhappy I'm in this job concerned because they acknowledge even they don't know the full power of what they've created opening eyes Chief technology officer Mira morati shows us the new version out this week just listen as I ask a complicated s.a.t question Lisa gives her brother Sam a 15 second Head Start in a 300 meter race during the race Sam runs at an average speed of 5 meters per second Lisa runs at an average speed of 8 meters per second not including the Head Start since the last time Lisa started running which of the following best approximates the number of seconds that had passed when Lisa caught up to Sam within seconds it answers correctly the answer is B 25 seconds and so we pressed about the potential for cheating you have this this technology can beat most humans at the SATs the bar exam how should schools be integrating this technology in a way that doesn't increase cheating that doesn't increase laziness among students education is going to have to change um but it's happened many other times with technology when we got the calculator the way we taught math and what we tested students on that totally changed the the promise of this technology one of the ones that I'm most excited about is the ability to provide individual learning great individual learning for each student they argue the eventual benefits outweigh the negatives take for example your taxes we gave chat gbt the tax information for a typical family of three and chose a recent year giving it pages and pages of tax code I'm copying the tax law here it's a few pages long so you're copying tax law just multiple pages of tax law and within seconds the standard deduction gpd4 says it's twenty four thousand dollars which was correct and tonight a first look at what the technology could see now available for test use for selected blind people through the be my eyes app watch as we take a photo of what's inside this refrigerator the technology analyzes what's there I see that you have some bread some mozzarella cheese tomatoes and mayonnaise you can make a simple grilled cheese sandwich with these ingredients you can make a strawberry toast by spreading the raspberry fruit spread on the bread and topping it with sliced strawberries perfect providing a recipe but is this also a recipe for something else replacing jobs with technology you've said AI will likely eliminate millions of jobs many people are going to ask why on Earth did you create this technology I think it can do the opposite of all of those things too it is going to eliminate a lot of current jobs that's true we can make much better ones the reason to develop AI at all is that I I believe this will be uh in terms of impact on our lives and improving our lives and upside this will be the the greatest technology Humanity has yet developed so in the wrong human hands it could be a very different device it could be a very different power we do worry a lot about authoritarian governments developing this and using Russia China are you speaking to the government are you're in regular content regular contact more and more every day as the company continues to improve its technology they're extremely aware of its shortcomings the possibility of biases and spreading misinformation what should people not be using it for right now the thing that I try to caution people the most is what we call the hallucinations problem um the model will confidently state things as if they were facts that are entirely made up you call it hallucinations a lot of people would think of that as a lie or an untruth yeah does chat gbt does artificial intelligence create more truth in the world or more untruth in the world oh I think we're on a trajectory for it to create much more truth in the world we release gpt4 and we showed our progress with reducing hallucinations there so I think this is a problem that we will make tremendous progress on in the coming year it's a problem that marathi's team is consistently working on you're spending 24 7 with this technology you're one of the people who built this technology what is most concerning to you about safety this is a very general technology and whenever you have something so General it is hard to know up front all the capabilities as well as its downfalls and the limitations of it is there a possibility that it could do something in the future that we don't want it to do and there's no way to turn it back it's not out there without bounds we have some checks and balances in place including monitoring detection and monitoring of what's going on in the platform and taking action if we see things that are not in accordance with our usage guidelines or you know we see potential harmful applications of technology is there a kill switch a way to shut the whole thing down yeah so the the there's the Sci-Fi version of this which is you know you imagine someone like in a data center pushing a button and this bomb goes off or whatever um but what really happens is like any engineer can just say like we're going to disable this for now or we're going to deploy this new version of the model a human yeah 2024 the next major election in the United States might not be on everyone's mind but it certainly is on yours is this technology going to have the kind of impact that maybe social media has had on previous elections and how can you guarantee there won't be those kind of problems because of chat gbt we don't know is the honest answer we're monitoring very closely and and again we can take it back we can turn things off we can change the rules this doesn't have the same ability to sort of instantly go viral like a like a tweet on social media we can monitor what's happening we can stop things that are happening and we'll be doing that very closely 10 years from now how does this change Humanity I think that pushing the rate of progress on human knowledge will have a great impact on human well-being whether you know that's with climate or education or Health Care maybe you solve diseases that right now we can't even wrap our brains around hi everyone George Stephanopoulos here thanks for checking out the ABC News YouTube channel if you'd like to get more videos show highlights and watch live event coverage click on the right over here to subscribe to our Channel and don't forget to download the ABC News app for breaking news alerts thanks for watching

Channel: ABC News

Views: 24,547

Keywords: AI, Artificial, CEO, ChatGPT, Intelligence, OpenAI, abc, abcnews, p_cmsid=2494279, p_vid=news-97931966, technology

Id: Rty_HFuyG4I

Length: 8min 46sec (526 seconds)

Published: Fri Mar 17 2023