Emerging advances into artificial intelligence raise questions about the future of humanity.

TRANSCRIPT.

hello again I’m Chuck Todd and welcome to another edition of Meet the Press reports today we’re going to take a deep dive into the topic of AI artificial intelligence and with that in mind I should clarify that yes I’m the real Chuck Todd not some AI generated Avatar although we have those for you this episode by all indications artificial intelligence is getting exponentially more intelligent and powerful and it has the potential to do good like revolutionizing cancer detection but it also poses great perils both in the long term and right now as it’s rolled across Society this is the fledgling and deeply controversial state of AI in 2023 it’s a world of possibility and optimism if you believe the people who are making it and fear and dread even from some people inside the industry and some of that fear has to do with how social media went we were pretty optimistic about that and look where it led us so joining me now is Jacob Wharton he’s been covering the emergence of AI and now Jacob it’s the be-all end-all this was sort of a pet project of yours I know here we are it was speculative when I began and now it is no longer that now for this installment we wanted to ask this question I mean as we let this technology creep into our lives we want to look into the people behind it why do they seem so convinced that we should treat this technology like some sort of God that we should ham some of our most important decisions to it is it a way to save Humanity from itself or are we risking the downfall of society by experimenting with it so quickly AI has burst into our imagination our culture and our daily lives because after Decades of trying companies have learned to train it very quickly on huge amounts of our data and now it can do battle with customer service agents a hundred dollars we went from a 20 discount to 100 discount in seconds even as the creator of the technology sometimes I’m surprised create digital remixes of human creativity the Godlike power to commission any artist in history yeah that’s the fun part and even catch a glimpse of our thoughts wow so as long as I have seen it and you know the patterns of my brain then the AI will read that out of my brain yeah yes yes yes exactly AI Advocates like former Google CEO Eric Schmidt says it all has tremendous potential getting better materials solving climate change managing our Energy Systems better there’s a lot of reason to think that this technology can do all of that and already it does miraculous things at Cincinnati Children’s Hospital Dr John pestian and Dr Tracy glauser are training their own AI to head off suicide in children it can give a pediatrician an at a glance sense of which kids need immediate intervention an output will say this is a patient at high risk this is a patient at low risk where we just don’t have enough data right now roughly 30 to 40 percent of mental illness adult mental illness started as a as a child and so you want to be able to pick that up early they spent Millions on building and training it by pouring electronic medical records records of doctors visits even suicide notes through it what had most value in prediction and classification is the ratio of nouns to pronouns because those that aren’t suicidal are those that will use the word I more often than those who are suicidal and they don’t refer to themselves so if in the language of the suicide side note they refer to things other than themselves more often they’re more likely to commit suicide exactly now that project is pretty amazing a custom made expert built potentially life-saving artificial intelligence but that is the exception it’s not the norm the AI that you are most likely to encounter in your life is built by a handful of for-profit companies using data you never realized they were collecting and increasingly it makes decisions about things that are really important to you whether you know it or not life and death decisions are being made by decision making system and they’re determining people’s act to government support for housing food Medical Care University of Baltimore law professor Michelle Gilman and her students help people fight for benefits that have been wrongly denied them and increasingly she finds out that algorithms bought from a secretive company are at fault I’ve been in hearings where no one in the room can describe for me how does the algorithm work what factors does it weigh how does it weigh those factors and so you’re really left unable to make a case for your client in those circumstances you’re in this sort of kafka-esque Loop in which there is nobody making a decision and no one who is in theory in charge of making decisions has any idea how the decision was made right but yet at the same time many of the decision makers the judges give undue deference to the automated system because they think it’s a computer it’s mathematical it must be right Gilman says that once upon a time when people could come into an office like this one to get things sorted there were problems sure the potential for harm is just much greater when these flaws are embedded in algorithmic systems the invisible power structures we are assembling by accident and on purpose have the potential to change not just individual lives but the very future of humanity this isn’t intelligence this is basically a sort of warped mirror of what’s on the internet for the last 20 years Meredith Whitaker was an early AI researcher at Google she left after leading walkouts over military contracts and Google’s handling of executive misconduct Google made several changes to its products and policies after the outcry but Whitaker says it’s the issues around AI that worry her most what it really is is a probabilistic engine designed to spit out things that seem plausible based on a prediction of you know how how this should look that is based on massive massive amounts of you know effectively surveillance data that has been scraped from the web today she’s the president of the anonymous messaging app signal part of what she says is an effort to cut off the data fuel supply of AI she says that the same companies who collected data on all of us without any regulation for the last 20 years are using it to now build AI again without regulations let’s be real there are only a handful of companies in the world that have that convert combination of data and infrastructural power capable of creating what we’re calling AI from nose to tail what is the shortcoming of there being this very small number of models upon which a whole ecosystem is theoretically going to be built well I mean I think it’s the it’s the danger of concentrated power and it’s the danger of a concentrated power that has extraordinary ability to shape our social and political landscape we are now in a position where you know this I would say over hyped technology apology is being created distributed refined calibrated and ultimately shaped to serve the economic interests of these same handful of actors you know like Whitaker says there is nothing innately socially positive about a system sold as a replacement for human judgment and built for profit so the idea that this is going to sort of magically become a source of social good or that this is a you know kind of a natural substance that all of us have the ability to use equally and hey teachers will be using it and students will be using it and non-profits will be using it is simply not true that’s a fantasy used to Market these programs so who are these people making Ai and why are they doing it what do they believe about the future I mean if AI has the potential to make everything more efficient but also poses the risk of making it such that we can’t do things any other way why do they think that gamble is worth it the creator of chat GPT open AI CEO Sam Altman declined to speak with us but he has spoken openly about his desire to build not just AI but AGI or artificial general intelligence a theoretical future technology in which AI truly thinks for itself and surpasses human intelligence Altman ruminated openly on this podcast about his enormous optimism and ambition when it comes to AI I mean I hate to sound like utopic Tech Pro here but if you’ll excuse me for three seconds like the the the level of the increase in quality of life that AI can deliver is is extraordinary we can make the world amazing and his company now part of Microsoft says it wants to go far beyond chat Bots recently writing our mission is to ensure that artificial general intelligence AI systems that are generally smarter than humans benefits all of humanity some in the AI Community were appalled they think they are positioned to decide what benefits all of humanity wrote One prominent University of Washington Professor the top most people at AI companies tend to say they believe they are building something world changing I’m extremely excited about what’s going to happen because when these things are 10 and 100 times larger they’re going to be much more useful in terms of solving the problems of the world today climate change scientists new strategies new ideas they will write poetry in great ways they will inspire humans former Google CEO Eric Schmidt has been an advisor or boss to Altman and other top usai Builders do you feel like is the community of AI leaders that you’re part of also an agreement on what the potential is in the future AGI and the rest of it yeah in one I mean they differ on time Schmidt like Whitaker says that the number of companies capable of building true Standalone AI systems is very small I think eventually there’ll be 20 or 30 such groups spread around the world I don’t really know today there are roughly four or five depending on how you count the models that they’re training are 50 to 100 million dollars a copy literally the training and at the moment that small group enjoys little regulation or oversight and no meaningful public input as to whether this technology should be deployed the key issue from my perspective is how do we put guard rails on the worst behaviors and how do we get International agreement on what those things are but he says that with certain industry crafted guard rails in place the companies making these systems are the ones that should be trusted to figure it all out my concern with any kind of premature regulation especially from the government is it’s always written in a restrictive way what I’d much rather do is have an agreement among the key players that we will not have a race to the bottom you’ve described the need for guard rails and what I’ve heard from you is we should not put restrictive regulations from the outside certainly from policy makers who don’t understand it I have to say I don’t hear a lot of guard rails around the industry and that it really just as I’m understanding it from you comes down to what the industry decides for itself when this technology becomes more broadly available which it will and very quickly the problems are going to be much worse I would much rather have the current companies Define reasonable boundaries it shouldn’t be a regulatory framework it maybe shouldn’t even be a sort of a democratic vote it should be the expertise within the industry help you sort that out the industry will first do that because there’s no way a non-inderacy person can understand what is possible it’s just too new too hard there’s not the expertise there’s no one in the government who can get it right but the industry can roughly get it right and then the government can put a regulatory structure around it that was uh that’s a pretty it’s pretty arrogant answer there and I’m sorry I don’t I don’t know how else to react to that that I’m sorry nobody in government you won’t understand what we’re doing so you don’t have any input boy that is that is I know Eric Schmick a little bit I think if you push something you’d be like well I don’t mean it that way but it certainly comes across it like you stay out of our business you know I would say that what he said there is an Earnest uh belief that it really does boil down to expertise I think he does not believe that they have the expertise in government to really get involved in this it’s certainly at this stage but I will say that across the spectrum of experts that I have spoken to that is one of the major concerns that there is a a exclusion of other people from the decision-making process because only the people that can afford it 100 million dollars per copy technology should be in charge of a thing that’ll affect us all so thing that I have been working out in my own head is is this I think we’re all doing the skeptical approach to AI because we didn’t with social media I say collectively as a society uh how much of this are we over is there possible we people like you and I and the Skeptics are we over correcting because of oh my goodness we didn’t have any of that input um or is this the right reaction well so I what’s so interesting is that you know Eric Schmidt and speaking to me there he said it to me and he said it to many people before he says that he was naive about social media back in the day so how do we trust him now and yet he believes that he you know has some sense of how we should proceed what all of the people that I’ve been speaking to agree on whether they are for or against the AI moment we are in is that it is the unfettered access to data that fuels all of this and part of what that is going to do is create the same little cotery of very powerful companies that collected data on us for the last 20 years they’re going to be in charge of AI for the foreseeable future the internet didn’t have this it wasn’t it didn’t there wasn’t a feeling that only one company controlled the internet well that’s right because and I think with AI it’s so much different it does feel as if it’s a handful of companies that have this control the internet felt like we all had a piece of it well that is because the internet wasn’t fact created for the benefit at the time it was a military invention but it was built inside the government and the people who refined it were academics here we’re talking about something that was once upon a time at least in theory academic but it has really hit the road only when for-profit companies put billions of dollars behind I would argue that it’s one of the first times in human history that something that’s going to affect us as deeply as anything ever has is entirely in the hands of the people who are poised to make money off it an important point to end on there Jacob Ward I know you’ve been passionately following this uh for almost a decade now here’s the moment let’s see if we survive that I really appreciate it thanks Joe thank you and I want to note uh that Jake has been covering this issue for some time he’s written a book on it should take a look at it it’s called The Loop how technology is creating a world without choices and how to fight back when we come back a I is offering the possibility of letting us live forever at least in Hollywood and a new movie starring an AI generated James Dean could be coming to a theater near you and sooner than you think is is this a good idea stay with us [Music] welcome back immortalizing our icons isn’t new Madame Tussaud’s Wax Museum and holographic concerts or proof of that but what if we could actually talk to these people well joining me now are two people working on just that Travis Cloyd he’s CEO and co-founder of worldwide XR is a global futurist focused on Technologies around AI machine learning and more he’s currently working on the film Back to Eden which will feature a digital human named James Dean that James Dean also joining me is Kathleen Haas former USC uh from the USC Institute of Creative Technologies vision and Graphics lab where she’s focusing on AI generated virtual humans and her work by the way is primarily funded by the Department of Defense we will ask her about that so Travis and Kathleen welcome to Meet the Press reports and Kathleen let me start with a little bit of a vocabulary lesson so there’s you know before AI became such a buzzword in the last six months deep fakes with something is there a what’s the difference between a deep fake and an AI generated human well a deep fake is a 2d um uh it’s a 2d technique and uh it is embedded in video and uh it’s very effective we’ve seen a number of them they look very cool but the AI based approach to face replacement is a different process it is a CG face and in our lab we do a sort of a hybrid approach where we take we have a morphable model that’s developed from numerous light stage scans and that gives us all the reflectance data which means we can relight this the skin and then we marry that with the highest resolution 4D video that you can find and you can do a little encoding and decoding over here and a little machine learning encoding decoding over there Secret Sauce mix it together and then you output this pretty fantastic looking photorealistic aai CG face and and it’s animatable complete lately I can animate it by texture excuse me by text to speech or or audio or anything so it’s not a deep fake is like using white out and writing over it it sounds like right it’s like cut and paste a little bit um but the set a i generated obviously you said it’s a little bit more now you we put it we gave you a photo of me and you did an AI generation and you did it in 24 hours yeah now I wish you’d given me a little more hair so literally so this we gave you the high score we did not give you a 4K resolution for what it’s worth right um right so how would you advise somebody to be able to figure out well what’s fake there I mean obviously I know it’s fake with the hairline and all of that well it uh it it’s still up in the air we you know as far as how you detect whether that’s a deep fake or a fake are you putting watermarks in are you guys thinking about this yet or not we’re thinking about it for sure I mean we’re a research lab so we we’re now just we’re developing you know computer vision research lab at USC uh funded by the Army mainly and um our um our main focus is research and Publications so we’re sort of you know for the greater good generating knowledge in the topic and we’re putting together these really amazing uh you know capabilities um all the other parts of those is not really our departments and how long did that take uh yeah it took um by the time they got it it took a couple of hours actually so literally in a couple of hours you could generate you know with the with the right type of equipment you can generate essentially a fake human yeah it’s pretty good Travis you’re about to make a movie with a fake human yes James Dean um look you have the rights to this and all of that um why do you think this is a good idea you know um you know we’re a digital Rights Management agency so we work with a lot of historical iconic Estates and in working with the James Dean estate who my business partner Mark Rossler from CMG worldwide has represented the family for 40 years it’s been the desire on behalf of the family to extend the legacy of James Dean uh Inspire future Generations essentially and so in those conversations with the estate and with the family um we had the desire to to work with them to extend that Legacy through different mediums whether it’s film or virtual reality or experiential content for the future and uh over the course of time we’ve had a number of film studios that have reached out to us about doing James Dean feature films and it was always a conversation of hiring a big celebrity talent to play that part sure and it’s only been recently with all the you know Innovation Evolution this technology that we felt like creating the digital human of James Dean was the ultimate path in doing that yeah so how are you doing it is it just all of his every on-screen role is put into sort of one data set a data pool if you will and your AI James Dean pulls from that pool to some extent it’s a lot of the copyright images and video assets that we have the family has and controls um call it the source material so it starts there it starts with Aggregate and all that Source material running it through machine learning to create that base model and then you have a lot of other types of innovation that you would apply to it so if you but this is James Dean whatever role he played I mean what’s interesting here to me and this is sort of where I’m sort of stuck on AI humans is that you’re not getting the essence of who he actually was you you don’t have the thoughts that were going through his head then do you no we don’t right no we don’t but we have the family which we try to make it as authentic as possible um you know we do everything in an ethical and responsible way as much as we can in working with the family what does that mean leveraging books leveraging a lot of the copyrighted material just the family’s consent working with them to make it is as solid as we possibly can and it’s a combination like I said of different types of Innovations motion capture if we’re worrying about the body voice synthesis if it’s worried about the voice if we find people that have different because James Dean passed away 67 years ago so we’re limited to assets that existed a long time just like you getting your deep break or your digital human created with a video that was in 4k but you do have 4K videos that Kathleen could have used we’re very limited with the material The Source material that we have what’s interesting here is I want to put up this quote from Keanu Reeves he is clearly going to try to he doesn’t want people to use his image for this he said what’s frustrating about uh that is you lose your agency when you give a performance in a film you know you’re going to be edited but you’re participating in that if you go into deep fake land it has none of your points of view that’s scary and then he added it’s also a Fascination it seems for us the animals on the planet like how do we defeat death and you said Robin Williams also made actually put in his will he did not want to have his image used after death so more actors are thinking about this though yeah more actors are thinking about it Robin Williams put in his will 25 years rights of publicity nobody could use his rights of publicity for since the date of death so essentially he’s protected himself through including this language in his will are more do we need a if you don’t specifically put it there are other dead actors just sort of up for grabs not necessarily I still think you need to go through the right channels to secure the rights of publicity agreements and Licensing deals with those Estates to leverage a lot of the copyright material that it takes to create base modeling and to some extent yes there are the bad apples out there that could leverage a lot of the existing IP that’s on the internet to create a digital human of somebody else Kathleen what are your ethical guardrails well um mine in the lab you know we uh you know we have uh um we have to adhere to our sponsors and and uh deliver uh very specifically milestones and tasks uh your sponsor in this case being the Department of Defense uh um yep uh so you know ethics though it certainly comes up and it’s it’s a concern um along with all the wonderful amazing cool things that it can do uh I think to help you you have to have a full conversation about it our lab uh doesn’t um you know we don’t make constraints based on ethics we also don’t um transition it out into the world like um you know currently uh you know a project we will I’ve had some people who are like really upset at Microsoft blaming them for opening you know for the first for putting it to the public before everybody was ready yeah I’ve heard a lot about that too absolutely but I will say this about the law about the um the content itself and I don’t think you know James Dean is maybe the candidate but there there’s there’s such a digital footprint that we all have now and you know future you know I don’t know how many years everything’s moving so exponentially quickly it’s very possible that you could gather a lot of data on any in individual person with all the the photos and the contracts and the letters and the emails and everything you’d in an AI based uh uh system would actually be able to to pull out a personality from you really believe they could huh oh yeah I do I mean uh Ray Kurzweil did that with his father to a certain extent is a test about six years think of John Adams do you think if we put all the letters that John Adams wrote yeah created a terrific book yes that we you could create AI John Adams I think so I think you could have a so you could you could you could uh you know you could yukushin and it could be pretty immersive you gave me another rationale that said hey better that the good guys are first than the bad guys are first and on one hand I get that on the other hand is having an AI arms race a good idea I think we’re I think the cat’s out of the bag it’s just an AI arms race no censoring in your hands anymore uh well I think that uh that uh smart people uh need to get together uh from all walks uh whether it be government or Scholars or others and come up with uh you know constraints and Regulation and and some something yes I do Travis are you convinced the public wants to see dead actors I do I feel like um I wouldn’t call them dead actors but I think they’re icons of the past reimagined in a new frontier I wouldn’t necessarily use terms like dead actors essentially but I think there’s a desire to as I mentioned earlier just Inspire the future Generations by leveraging this technology I have to tell you it’s scary and exciting yeah I’ll not I’m not gonna lie I’m intrigued and I’m not sure I should be oh it’s wonderful I appreciate both of you Travis Kathleen thank you so for more on all the ethics of all this new AI technology take a listen to the latest episode of the Chuck podcast I I spoke with sociologist Chasey Reese anthis co-founder of the sentence Institute and next week join us back here for our Deep dive into the U.S and China race to Mars I’ll see you then and of course this Sunday on Meet the Press. thank you.