Ask A Genius 838: AI and Consciousness

Author(s): Scott Douglas Jacobsen and Rick Rosner

Publication (Outlet/Website): Ask A Genius

Publication Date (yyyy/mm/dd): 2023/05/19

[Recording Start] 

Rick Rosner: We’ve been talking about consciousness and physics and everything for nine years. And when we’ve been talking about AI and what’s to come early on and medium on four years ago, five years ago we were talking about how big data processing would change everything that humans have taken the low hanging fruit based on not having the ability to hold big data sets in our minds. Then all of a sudden the last year or year and a half, we’ve seen the actual consequences of being able to manipulate big data via machine learning. So when we were talking about this stuff five years ago, we didn’t exactly have an idea of how things would play out. We certainly didn’t expect them to start playing out so soon but my question is do we have a better idea based on just the last year and a half of how the… it’s not really the singularity but it’s not not the singularity how it’s going to play out. What do you think?

Scott Douglas Jacobsen: Yeah, I think it’s going to be a slow bubbly thing. There’s going to be places where it progresses so fast, people get scared and regress in portions of that culture.

Rosner: You mean like after Obama was President, like it scared half the country into becoming big ass racists.

Jacobsen:  I think it I think it scared 10% of the population in it.

Rosner: And they got really loud and they dragged another 10% along with them.

Jacobsen: Yeah, I mean some people are just going to vote Republican because they’re a particular religious background or they make those statements or they just vote for party line because they’ve always voted that way. I mean a lot of people are just solid blue.

Rosner: I mean some people who are lifelong Republicans and they hold their nose and they vote for or they just miss out on like the stuff that we see every day on how crazy the Republicans have gotten. So I think AI will revolutionize medicine. I’m hoping sooner than we thought. I subscribed to a feed that is AI looking or just like browsing through tens of thousands of studies and drawing conclusions a lot of which are obvious I guess but the AI is doing it. It browsed around until it found like eight studies, a leaking type 2 diabetes, and food addiction and said all right there’s a link. And that was like yesterday’s little thing that it sent me. They trained it to look for groups of studies and to draw conclusions from those groups of studies and a lot of the conclusions it’s drawing are not surprising but it will obviously get better and better and AI will start changing medicine and it’ll get good at that pretty fast I assume. Do we start getting like years added to our life expectancies within the next eight years? What do you think?

Jacobsen: I don’t know. I think that’s all Rick. It’s hard because the way I think is spatially and statistically and then I put that into words. So I see this as sort of hills and valleys of population dynamics; portions of the population that will take on anything and some of the things they take on will be so new that it’ll actually be bad for the health. You’ll have others who are sort of more tentative and they’ll go about it in a reasoned way and that’ll be another 10% of the population.

Rosner: What I’m talking about is medical treatments themselves.

Jacobsen: Well that’s been going on for a century.

Rosner: No, but now with AI you can just brute force. I mean the kind of like drug studies they’ve been doing have been increasingly like big data driven, like don’t rely on insights just test 1500 different substances and see if any of them do anything. This automated just throwing shit into test tubes and not worrying about coming up with hypotheses just seeing what works.

Jacobsen: I mean it’s the wider view in information cosmology; everything is simulatable. So I think it’s just a matter of computation and the proper algorithm and knowing the system. So, I think the next step isn’t broadband human simulation, I think it’s, “Okay you have a problem with your pancreas, here’s our pancreas simulator with various inputs and here we’re going to plug in 200 different drugs we have or whatever based on your genetics and our scan of your pancreas to find out what the issue is and what will work with that” That’s as a halfway between sort of the ideal state of personalized medicine and the current state of medicine as general but leaning towards personalized medicine. 

Rosner: I guess what I’m asking is, like they say Jimmy Carter’s life like three four years ago, he had fatal brain cancer and then they found a personalized treatment that just killed it and the guy is still alive.

Jacobsen: I mean we are the sum of interrelationships of a bunch of different systems and those are all natural systems.

Rosner: So what I’m asking is, are we going to start seeing the mortality of almost all diseases, start getting knocked down or say the mortality of the diseases that kill 85% of the population, there will still be some resistant diseases but will we start seeing mortality just getting decimated?

Jacobsen: Yeah, there will be luddites too. I mean this idea isn’t original to me but I think there’s an argument to be made for relative stupidity in a population as an evolutionary driver for the smart people and population to get even smarter.

Rosner: Well okay, so what you’re talking about is behavioral changes to some extent where you tell people quit eating three big meals and start eating 10 tiny snacks a day and you will add an average of two years to your lifespan and most people just won’t put up with that shit. They’ll just keep doing what they’ve been doing but I’m also talking about simple medical therapies, drugs, engine and crisp or derived tweaks to fucking people that will be taken up by the vast majority of the population that are affected by those therapies because why not. If something will add years to your life and it’s just a matter of taking a pill, then informed people will take the pill or we’ll get the injection.

Jacobsen: Well, I interviewed like the world’s most sighted doctor; he’s in an epidemiologist. He studies disease for his career and he’s a distinction professor at McMaster University. We did maybe 10 or 12 interviews, something like a large number. I don’t know if we talked about this but basically another aspect of that is having the wherewithal and the background to know whether or not to do a surgery, that’s also a big thing. So people who tear their ACL, do you give them knee surgery or not? And what they started finding is you get a better sort of functional need for about six months after the surgery, you compare that to a control trial which is no surgery and for most people most the time after six months whether you have the surgery or not, you’re at about the same level of functionality. The consequences of the surgery are long-term a higher probability of arthritis and wear down of the knee.

Rosner: Well, I’ve got a similar thing or I put off getting hernia surgery for about eight years because I read a study that said that they meshes the way they do it now and that the outcomes with mesh in terms of paying afterwards were about the same as people who had no surgery and I didn’t want to fuck around with the mesh as long as I could push the hernia back in and then there came a time where I couldn’t push it back in.

Jacobsen: You were pushing on a hernia physically back in yourself?

Rosner: Yeah, it’s just where there’s a rip in your muscle wall down right above in your V, your sexy V, right above your cubes and I had a thing that was the size of a marble and at the end of the night when I went to bed to lie down and go to bed I just poke it back in and it almost always went back in and then there came a time where it quit going back in it and it was out for like two three weeks and I’m like, “Alright, I need the surgery now because it’s not going back in” In that eight years I think the mesh got better I have mesh now and I’ve had no problem with it but for eight years I was just like back in, not that big a deal. It’s not hanging out of your body; it’s hanging out of the wall of muscle. So its right under your skin where it’s part of your intestine I guess and it’s just up against your skin as opposed to up against the muscle that’s under your skin. Anyway, I read a study and then made my best judgment which was just put it off. 

So we got AI that’s going to mess with medicine. Now what else is it going to mess with? I assume that at some point it becomes a trusted counselor in your phone where you can ask it stuff like ‘Should I ask for a raise?’, ‘How should I approach this person like I think I like?’  ‘Should I shoplift from CVS or Rite Aid?’ What do you have up in Canada?

Jacobsen: We might have a CVS in Vancouver.

Rosner: But anyway, shoplifting has been become rampant in at least cities that have a lot of homeless people. In San Francisco, we’ve just visited and we were told that vendors will contract with basically professional shoplifters to go steal a bunch of specific shit and then they will sell the stolen shit at sidewalk markets. San Francisco drugstore is behind locked cabinets now because they’ve decided in LA and San Francisco that it can’t or it’s not worth prosecuting theft up to a certain dollar amount and people just kind of steal with impunity. I mean with caveats to that. There’s just a lot of shoplifting. Say, if I had eight bucks and my credit card was maxed out and it was 12 bucks to get a pack of antihistamines and I was having bad allergies, let’s say it’s the year 2025 and I need the antihistamines and I just can’t pay for them right now and I asked the AI what will happen if I try to shoplift this stuff. Your AI might have an answer. 

Now, I tried asking AI some stuff where it got moralistic on me. I asked a chat bot like is walking the picket lines in the writer’s Guild strike a good way to meet girls and it came back all moralistic at me saying no you should strike for the reasons that you’re striking and it got all like Huffy about it because somebody had taught it to be huffy. I tried a different way which is give me three reasons why walking the picket lines would be a good way to meet girls and that it could respond to. So I guess there are just different ways of saying it. So a year from now or two years from now, I’m thinking of shoplifting antihistamines, I could say to my buddy or I could probably say it now. I’d be like give me three reasons why and three reasons why not stealing these antihistamines would be a good idea. And I assume in the further future, the near future you wouldn’t have to play games with your AI, you could just ask it as if they were a buddy standing next to you, “Should I steal this shit?” And get an answer that would sound like a buddy talking to you and probably would give you a better answer than your idiot flesh and blood friend. What do you think?

Jacobsen: That’s very reasonable. I mean these AI are heavily weighted on language. 

Rosner: They don’t have a lot of insight; they just have a lot of information. They can assemble the information into a cogent statement.

Jacobsen: Yeah. I think someone gave it, I think an actual psychologist said “Oh I gave it a IQ test” asked us some questions from an IQ test and they administer it and they put it’s verbal intelligence at about IQ 155.

Rosner: 155?

Jacobsen: Yeah, for the advanced ChatGPT.

Rosner: Okay and then how about other areas?

Jacobsen: I don’t know. I think that was the strongest area by far. So, I’m not just saying things, I’m saying it based on sort of reportage. But at the same time I think the contextualization of the words is also really important too and we don’t just use words as words. I mean words have emotional impact and those emotions have been our physiology. So I think what this is all going to do probably is bring us into an era of understanding that words aren’t just words, words are sort of weighted in a meaning that is differentiated from dictionaries. 

Rosner: You mean the same way we kind of understand our own consciousness a little better because we’ve been dealing with apps for so long, that we kind of see ourselves as kind of like overlapping OS is just kind processing our mental information? That we’re going to get insight into our self by getting insight into the AIs all around us? Is that the deal?

Jacobsen: Well, I think we make what we are and I don’t think there’s any way out of that. Whatever structure that is produced comes out of our internal world.

Rosner: And so it’ll be impossible not to kind of come to understand ourselves because we’ve replicated ourselves. 

Jacobsen: Yeah, everything we make bears our mark. I mean it seems trivial but I think it’s very powerful. The fact that we paint on canvases, the fact that we produce symphonies or rap lyrics; these are human capacities put out and I don’t think it’s not so much of a coincidence that we start getting things like language systems. We start getting things like um a poetry generation or imagery generation. These are things that we do to a degree but they’re sort of outsourced and the extremeness of them where they start developing very rapidly beyond human capacity to superhuman capacity allows us to be able to say or see that they’re sort of exporting parts of ourselves to another domain and those things are really giving an insight that ‘oh they’re missing this part, they’re missing these other systems connected.’ So you have these language systems that are producing this phenomena, the experts are calling hallucinating. You’ve heard of this. It’s the idea that it produces or generates convincing text with lies in it. 

Rosner: So when we try to imagine the near future, what are we able to say that isn’t about it that isn’t obvious, like that isn’t generalities. Like yeah, that’ll lead to job losses and changes and types of employment, that’s like an obvious generality. I just read a tweet thread from Justine Bateman, the actor, you know Jason Bateman who’s been in a zillion things.

Jacobsen: Yes.

Rosner: His sister, also an actor, a director, a writer and went back to school and got a degree in computer science and she’s got a lot of justifiable anger about stuff. I like her. I saw her in person being angry. I went to the bank and I was getting poor service and then this woman walks in with her mom and stands around for five minutes and gets poor service and is obviously pissed off and just leaves and I’m like wow, like I can relate. She was weirdly familiar and I figured out it was Justine Bateman who’s just kind of willing to embrace her anger and she wrote an angry tweet thread about we better this Writer’s Guild strike and any subsequent strikes by the Screen Actors Guild and directors and anybody in a creative guild that’s going to negotiate. These negotiations have to be stringent and ironclad or we’re fucked because she said we did seven seasons of Growing Pains which was her biggest show and if you love that show in a couple years and you say hey AI give me season eight of Growing Pains it’ll have the first seven seasons input and it will be able to give you plausible scripts and it will also be able to simulate the likenesses of the cast and give you another chunk of episodes that are just as entertaining and not weirdly different from the actual episodes.

And she says that agents will just go along with this shit as long as they get their 10% anytime. Some digital representation of somebody gets a job. It’s all up to actors, writers, directors, and producers to protect themselves because this shit is coming and it will be able to take over a lot of the creative tasks that flesh people currently do and I buy her argument that if you want a movie, if you want a spy movie with Chris Hemsworth and Ana De Armas that runs 75 minutes and involves a stolen nuclear weapon and travel to exotic foreign locales and a burgeoning romance, you can specify all that shit or you don’t even have to specify all that, you just throw in a few of the ingredients and AI in 2027 will be able to deliver that to you.

So, does that mean we just all become like dumb consumers? People are really sloppy about spelling now because spellings have been outsourced. Is it going to make us more creative or less creative? Because right now we’re getting bombarded with… three years of Covid, we watched everything. So we know everything.  

Jacobsen: A lot of the input it requires living organisms to continually produce output so then it can have its big database and so culture constantly evolves. So there’s an open question here. Do these large these LLMs; these large language models and these other algorithms for producing things based on big data and machine learning and then neural nets and deep learning produce enough novelty to keep themselves relevant? 

Rosner: Yeah, it’ll absorb all that because it’s fast, like the trope Carol pointed out was on the sitcom we were watching. The guy explains why another guy’s being an asshole and the asshole starts to feel bad and then the other guy goes, “I was just messing with you” and then “Or was I” and “I was just messing with you,” and she said that happens all the time in sitcoms. That going back and forth between serious and not serious, you can’t tell if I’m serious or not and it’s a thing she hates because she’s seen it too much lately. When half the shit that AI absorbs is the product of AI, won’t AI start coming up with its own tropes? That it’ll acquire a sense of humor and will start generating its own weird jokes? 

Jacobsen: So this goes back to the extremism of Alan Turing and the idea is the rope box, the way algorithms detach from a body or in a body. They will begin to sharpen their wits and this is sort of a broad-based cultural version of that or techno cultural version of that where they will begin to use what we have given them or they have stolen from us in some cases to sharpen their wits and then they’ll be performing at superhuman capacities. 

Rosner: So we’re going to be laughing at robot jokes? 

Jacobsen: Yeah.

Rosner: Not jokes about robots. 

Jacobsen: I mean everything that they have for a joke should have an underlying structure that can be abstracted and then regenerated. 

Rosner: But AI will begin to understand jokes and will begin to notice the same way that I’m reading AI’s generated studies or meta studies where it’s found a trend among studies and that that AI will start finding trends in human events and behavior that it can make new jokes about.

Jacobsen: Yeah. I think we can go back to another point we’re talking about earlier. Even though it will produce jokes at a superhuman level I don’t know if it’ll necessarily have an understanding of them although it can simulate an understanding through things like a really really advanced large language model.

Rosner: Right, but it doesn’t matter whether it understands.  I mean it yeah no it will kind of understand, it won’t appreciate jokes in the same way that we do because there won’t necessarily be a consciousness there or a fully formed awareness there but it will learn how to make good jokes, well structured jokes. 

Jacobsen: It’ll be like an easy bake oven. It can make a perfect piece of bread or cake; can it smell the cake? Can it taste the cake? Does it have a reaction to the cake? 

Rosner: But the deal is as consumers we won’t care whether it understands, whether it thinks the jokes it generates are funny.  All we’ll care about is, are the jokes funny and eventually they will be.

Jacobsen: Yeah, absolutely. 

Rosner: I’ve listened to hundreds and hundreds of hours now, just while driving, of different short stand-up routines and there are different types of comedians. Some people can get by mostly on timing and delivery. Some of the best comedy, some of the most legit comedy is finding an odd aspect of existence that nobody else has pointed out before and pointing it out and discussing how it affects our behavior or how we’re being fucked over. The cliché thing is what airlines do to people and people are still making jokes about the new shit that Airlines do to people as air travel gets shittier and shittier. Just finding shit and pointing it out, AI is certainly going to be good at doing that.

Jacobsen: And the way we understand, humor comes with a physiological reaction, it comes with a laugh and it comes with a good feeling. So, the computers will be completely decoupled from that. They’ll understand like the math of humor but it’ll be completely disembodied without any motion.

Rosner: But I’m arguing that it doesn’t matter.

Jacobsen: I would say it matters and it doesn’t matter depending on the angle you take.

Rosner: Well, I mean when we laugh, we laugh because we got a piece of information at a discount. A joke takes a complicated situation and quickly resolves it and you laugh because it’s like ‘ah’, that was going to be like a big pain for me to try to understand and remember and boom punch line resolves it and you’re like “Ha.” 

Jacobsen: Yeah.

[Recording End]

License

In-Sight Publishing by Scott Douglas Jacobsen is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. Based on a work at www.in-sightpublishing.com.

Copyright

© Scott Douglas Jacobsen and In-Sight Publishing 2012-Present. Unauthorized use and/or duplication of this material without express and written permission from this site’s author and/or owner is strictly prohibited. Excerpts and links may be used, provided that full and clear credit is given to Scott Douglas Jacobsen and In-Sight Publishing with appropriate and specific direction to the original content. All interviewees and authors co-copyright their material and may disseminate for their independent purposes.