Tech Founder Explains Why Supply Chains Are Breaking and AI Ethics Teams Lack Diversity
About this episode
Jon Stokes, founder of Ars Technica and independent writer, joins Vance for a sprawling conversation about systemic fragility across three domains: semiconductor supply chains, cryptocurrency as an escape valve from centralized financial censorship, and the concentration of ethical power inside AI systems run by a handful of companies. On supply chain, Stokes explains why Taiwan's TSMC can't simply be "unplugged and moved" the way an oil field can be seized — chip manufacturing is a fully interdependent, non-modular global process, and disruption (COVID lockdowns, phantom orders, shipping bottlenecks) ripples for years, not months. He and Vance trace a chain of unintended consequences from understaffed fast-food restaurants to stalled trucking schedules to empty store shelves. On crypto, Stokes — speaking as a systems/software person rather than a "crypto replaces fiat" true believer — describes Operation Choke Point and modern de-banking (his friend's account closed over an LLC named "Blaze") as the real motivation for decentralized finance: censorship resistance, not speculation. The conversation's centerpiece is Stokes's critique of AI ethics teams at Google and Facebook, which he argues are staffed almost exclusively by secular, "woke" engineers and linguists with no formal ethics training and no religious representation, making unilateral linguistic and moral decisions (e.g., "birthing people" vs. "pregnant women") that get pushed to billions of users overnight. He extends this into a broader worry about AI sentencing tools and algorithmic hiring — arguing that centralizing judgment into one "super-powered" model eliminates the messy, regional, human variability that lets a pluralistic society actually live with disagreement (comparing it to Greek myths of all-powerful, flawed gods). He closes pessimistic about reforming Google from outside and instead pins hope on building decentralized alternatives.
“So semiconductor manufacturing, it's not like, it's not as modular as some other kinds of manufacturing or as you might think intuitively that it is... If Apple needs a chip, they can't just pick from a menu of people that make chips to their spec and be like, you are gonna make this chip for me this week.”
“So there's a demand surge that is hitting TSMC. There's a demand surge that's hitting the semiconductor industry. There's a demand surge that, you know, hit lumber a while back and then cratered, and now it's hitting steel.”
“The bank is obligated to put in a what's called a sars, a suspicious activity report. And that is only the beginning of all of the reporting that goes on... the government's role in, in censorship.”
Key moments
- ~5-13%: TSMC/Taiwan explainer — why chip manufacturing can't be relocated like a commodity extraction industry; the "fully baked cake" metaphor.
- ~15-20%: "Phantom orders" mechanism — panic-buying up and down the supply chain that manufacturers can't distinguish from real demand, compounding shortages.
- ~26-32%: Vance connects supply-chain fragility to farmers feeling it first (his own Twitter poll of farmer respondents rating disruption 8-9/10) before consumers do.
- ~38-42%: Bitcoin/DeFi as censorship-resistant alternative to banking gatekeeping; Stokes's personal story of a friend's bank account closed over an LLC name.
- ~44%: Operation Choke Point — using financial regulators to de-platform disfavored but legal industries (guns, later OnlyFans/adult content).
- ~56-58%: Core AI ethics argument — a narrow, secular, "woke" demographic making unrepresentative moral calls that reshape language for billions of users.
- ~68%: Harvard Divinity School and Princeton secularizing away from their founding religious traditions, framed as a parallel institutional drift.
- ~90-98%: Sentencing-AI and vaccine-triage thought experiments — why centralizing one AI's judgment into every hospital/court removes the regional variability that lets a divided society coexist; closing argument for decentralization over efficiency.
Notable quotes
“There are a bunch of people with no formal ethical training who are having opinions in publishing papers about ethic and all their ethics are woke... Is there a Christian in the room? Is there a Muslim in the room? Often the answer is no.”
“The minute that code is deployed... you have just changed policy across the entire hospital system, the entire court system, the entire Facebook newsfeed system... and there's no way to run away from it.”
“Operation choke point... a lot of credit card processors and banks were trying to de platform firearm sales.”
“If Apple needs a chip, they can't just pick from a menu of people that make chips to their spec.”
Predictions made in this episode
- id: pred-ep153-01
Full transcript
Read the full transcript (word-for-word, with timestamps)
Jon Stokes [00:00:00] If there are going to be people who are making decisions about how an AI that interfaces with billions of users a day speaks its mental categories, its gender categories, its kind of conception of, of reality and human community. If there are gonna be a room full of people making calls about that, I wanna know who those people are.
Vance Crowe [00:00:25] Hi everyone. I am Oliver Crow, a second grade student and aspiring YouTuber, and you're listening to the Vance Crowe podcast. Welcome back to the podcast. I'm glad you're here. Today we interview John Stokes, who was the founder of a magazine called Ours, Technica, which was a tech blog that talked about all sorts of cutting edge things. John left that now, and he writes for himself@johnstokes.com. But he talks about all sorts of things that I am super interested in. We talked about the semiconductor supply chain problem, really the supply chain in general, where this is going to impact consumers. And we got into all sorts of discussions about inflation and woke culture writing in things onto artificial intelligence. And we even had a conversation about Bitcoin and cryptocurrencies and how the grain markets might end up using these technologies. This was a fantastic conversation. I've been waiting weeks to do this interview and John did not disappoint. Before we get to that, I wanna bring up the Articulate Ventures Network. This is a place for people that listen to the podcast, that wanna find other people that listen to these types of interviews, like exploring ideas, like hearing ideas with people that you do and even don't agree with. And it's a private place. It's away from the tragedy of the commons where everybody's yelling about things. Instead, we have a small, tight, tight-knit community of people that have amazing conversations. We do a thing called the Circular Firing Squad, where we pick out debatable topics and people join and we have a really good debate where everybody leaves learning something and not feeling bitter and angry at other people that we're in the debate.
Vance Crowe [00:02:09] The other thing we do is monthly shared experiences, and this last month we decided we were gonna do a move challenge, and the network moved over a thousand miles, meaning that every single day, the most that you could get would be five miles. Some people walked it, some people ran it, biked it, whatever. But people every single day used it as motivation. We had people saying, I lost weight. We had other people saying, I used the chance to call and connect with people I haven't talked to in a really long time. And other people saying, I learned all about my neighborhood and got to walk with my kids and do all kinds of fun things. So it's one of those places that if you've been looking for a community and you like the types of podcasts that I put out here, the types of conversations that happen, you might consider joining the Articulate Ventures Network. You can do that by going to Network dot Articulate Ventures. And there you can sign up, sign up for a month, just try it out, or you can sign up for a year. We would love to have you. Well, without further ado, we're gonna head to this interview with John Stokes. John Stokes, welcome to the podcast.
Jon Stokes [00:03:15] Thank you.
Vance Crowe [00:03:17] So John, you ran a, a website called Ours Technica, which is how I learned about you oh, so long ago. But lately you've been talking about all of the challenges with supply chain, specifically around things like microchips coming out of Taiwan. And so I thought, man, I would love to have you on, this is very timely topic. The article that you caught my attention with most recently was how difficult it would be to just pick up the manufacturing from Taiwan and try and move it to a place in the United States.
Jon Stokes [00:03:49] Can you
Vance Crowe [00:03:49] Kind of summarize what your point was there? 'cause I think a lot of people don't realize just how complicated that is.
Jon Stokes [00:03:55] Sure. So semiconductor manufacturing, it's not like, it's not as modular as some other kinds of manufacturing or as you might think intuitively that it is. So if you, if I need a shovel, I can go to Tractor Supply and I can buy a shovel from one brand. I can go to Home Depot and buy a shovel for another brand and like that's an input into whatever I'm doing on my property, right? If Apple needs a chip, they can't just pick from a menu of people that make chips to their spec and be like, you are gonna make this chip for me this week. When Apple comes up with a design, the design targets
Vance Crowe [00:04:40] A
Jon Stokes [00:04:40] Particular process that belongs, it's proprietary to a manufacturer and it targets, its specifically at a, at a fab and a factory and you know the whole bit. So, so Apple, if Apple loses access suddenly to TSMC, apple can't just snap its fingers and be like, well, you know, we're gonna just move this to Korea and you know, I will send them the files for my microchips in Korea will spit out the microchips. You know, Samsung will begin spitting them out instead of TSMC. So, so that's, that's part of it. Another part of it is that all the inputs into high-end semiconductor manufacturing are global. So there is a Dutch machine that's used. There are German lenses from Zeiss, there are Amer, there's American software and tooling and all that stuff goes into a particular process for a particular manufacturer. So on the, the inputs to the process and on the outputs of the process, it's all like, it's a fully baked cake is what I said in the article. So if you've got a fully baked cake, the whole thing is, is like this, that's, that's the semiconductor supply chain and it's, it's all wired together in a certain way. So you can't just swap out a part here and a part there. TSMC can't just be like, well, you know, I was getting this this critical tool from from Zes, this lens, but now I'm gonna get it from this other guy that makes it, you know, we're just gonna swap that out.
Jon Stokes [00:06:12] And Apple can't just say, I was getting these chips from TSMC, but now I'm just gonna get them from South Korea. Like, that's a process. Like that's a whole, you know, that's called retargeting. You know, all all of these things are, are really stitched together and baked in. So if something happens to one component of that whole complex network, then you kind of gotta start over and like rebuild a big chunk of it. So if something happens to TSMC, apple kind of has to start back over, retarget a new process, you know, come out with new, new plans and new files and, and you know, tweak. Its designed to work on, on some other process. If TSMC loses access to one of its suppliers in Holland, let, like, let's say that China captures Taiwan and says we own TSMC, now we are the boss of TSMC and the USA says, well, okay, we don't want the CCP, you know, the Chinese Communist Party. We don't want them to have access to this leading edge chip tech. So, you know, in Germany, don't do business with TSMC and Holland don't do business, like cut them off, you know, then TSMC can't just be like, okay, well I've been cut off by Europe, I've been cut off by the us, I will just get my pieces from elsewhere and keep going. You know, that it, none of that works that way. So when this system breaks, the whole thing breaks from soup to nuts, right?
Jon Stokes [00:07:47] Like just like the whole thing kind of, kind of disintegrates and has to be completely rebuilt, then this is gonna take time, like maybe even years depending on what's broken and you know, what the cause was. So the point point reminded me that article was to Sure,
Vance Crowe [00:08:03] Well, I was just gonna say, it reminded me of like, you can't just start a Blue Water Navy, right? Like the, the British
Jon Stokes [00:08:09] Yeah. Yeah. Has such a
Vance Crowe [00:08:10] Huge advantage because you have to, it's not just the parts and all of the things you're talking about, but the know-how and, and how to work and integrate these things. So it's, it like your point, years and years, it could be because it's not like you can just be like, oh, we were doing, we we used to have an army, now we're gonna just put 'em on ships and there'll be sailors there. It doesn't work that way.
Jon Stokes [00:08:30] Right, right. It's a, it is a whole, yeah, exactly. And, and, and that was sort of my point was that with the article, the bigger picture of the article was like, look, you know, if China invades Taiwan, they can't just capture TMC the way that you can capture an oil field. You know, if somebody invaded Saudi Arabia, they could capture the oil fields and the facilities and they could bring some people in, start them up, and then we could just, you know, steal all the oil, right? Like, semiconductors don't work that way. So if China captures TSMC first off, you know, it could be that some of the facilities are gonna be damaged from the invasion, it's likely that some of the facilities would be damaged from TSMC, you know, because if the, if the Chinese are, and you keep using
Vance Crowe [00:09:18] This acronym, who is TSMC?
Jon Stokes [00:09:21] T-T-S-M-C is Taiwan Semiconductor Manufacturing Company. So, so T ti Taiwan Semiconductor Manufacturing Company is right now, they share process leadership with what I believe Samsung if, if I'm, if I'm remembering correctly. And that means that they make the, the most advanced semiconductors. Now Intel is close, Intel will will be up with them in, in another, I don't know, maybe a year or two, but, but Intel has lagged those guys. And so if you want the most cutting edge, you know, arm based, which is a certain kind of architecture semiconductor like for the iPhone or for the Mac or something like this, or for a phone, TSMC is your, your your top choice.
Vance Crowe [00:10:18] It's been fascinating on Twitter the other day I decided, hey, I'm gonna just post a, a question on a scale of one to 10, how much has the supply chain disruption impacted your life? And a, a large portion of the people I interact with on, on Twitter are farmers. And I was expecting like, you know, twos and threes, I was expecting to be like, ah, it's kind of a pain. I I got more than 60 responses that were like eight or nine and people saying it would be a 10 except for I think it can get worse. And this is something I don't think regular consumers are feeling because you're sitting at home, you want some pens or you want a chair or a, you know, any kind of piece of equipment, you go on Amazon and they only show you what they have, right? So you don't see the empty shelves. And my, and my sense is because farmers don't have as easily as, as readily available, like alternatives, they're feeling it first, but that this is coming to consumers. Do you think the, are we gonna start feeling the pressure around microchips from the supply chain issues in the regular consumer world anytime soon?
Jon Stokes [00:11:28] So, so I think we already feel that. So obviously the one place that that's happened and what is with automobiles, and there's a few things to know, actually, TSMC makes higher microchips. There are so-called legacy nodes. There are older chips, older plants that make, you know, chips on an older process. Those are valuable. Those go into cars and telephone boxes and all manner of gadgets. Like, you know, I've got this, this, I'll get a different, I've got this, she fire flashlight here, it's got like a little chip in the emitter somewhere, you know, it's got some control circuitry, you know, that stuff is made on these older nodes. And those older nodes have been a lot of what's outta stock. So we haven't, we haven't really felt, from my understanding, the true impact of TM CT SMCs over subscription and, you know, and, and kind of over overfilled its order book yet in the consumer market in a big way, possibly in some, some high-end auto type applications where there's a car and they're trying to stick a, a nice, you know, a high-end processor in that car for some reason. And then that's not available. So, so that's, that's a place where where you feel it directly, we're going to start feeling it directly in the fall with Apple.
Jon Stokes [00:13:00] Apple has already announced, Hey, look, we're gonna have availability problems with our iPhones and what our max, those high-end semiconductors are also an input into different other kinds of processors up or processes upstream. So you might use a, a, a computer that's based on one of those to design a, a part or to manufacture a part that goes into some industrial process, you know, so, so if you're not able to get that part out or design that part, or make a new part or replace that part, because you can't get that chip from upstream, then you can't give it to the guy downstream. So a lot of the, of the shortages that we've had in consumer electronics so far have been from the older legacy nodes and not yet from TSMC, we have not yet seen what it would look like if TSMC went offline
Vance Crowe [00:13:58] And what's going on. Like, is it, is it just that they're not able to get them into the container ships to get them out, or they have, they actually slowed down production?
Jon Stokes [00:14:08] No, they haven't slowed down production. They're, they're pegged out, man. They're, they're producing more chips than they ever have. There is a lot of, and this, I I'm not super clear on what's going on me personally, but, but there's a one phenomenon that is definitely at work is so-called phantom orders. So what happens, this happens in all kinds of shortages. We saw this in firearms with the periodic, you know, firearm sales spikes, the Obama panic, and then, you know, some of the other panic buying where a consumer wants like say, you know, 10, 10 widgets or not a consumer, a, a supplier, A supplier is, is a, is a widget retailer, and they need 10 widgets for the inventory. So they call the widget maker and they place an order for 20, or they call five people that make those different widgets, and they place an order for 20 at each one hoping that somebody will get them 10 widgets. Or this guy gets me one, this guy gets me three, this guy gets me two. And at some point, you know, they're able to collect up enough widgets to meet demand, right? So what they've done is they've created these orders that they don't really need to fill, they're just, they're looking for 10. And so they've ordered like a hundred from different vendors just in hopes of getting 10. This is happening across the supply chain and it's happening in semiconductors, it's happening in electronics where a shop needs a critical part. And so they just, they just try to put in a, a panic order of like a bunch of these, right?
Jon Stokes [00:15:44] And so what happens when you're a vendor is you can't tell all those orders look the same. You know, this guy is swearing to you on the phone that he really needs like 50 of these chips and you don't know if he needs really five or if he needs all 50 or you know, you don't know where the slack is. So you've got 50 orders on your books and you can't tell which ones are these phantom orders, you know, and which ones are actually legit consumer demand. And so many of these manufacturers are in that boat with everything. And it's not just microchips, it's a, this, this thing, this kind of thing happens that, you know, probably with, with upstream parts for toasters and microwaves and farm equipment and whatever, the guy that's making that piece of equipment knows that in order to meet, he knows what his demand is. He knows that the lead times have stretched way out for this cri, these critical parts that he needs. So he just calls up everybody and places a cajillion orders, you know, and you multiply that across the economy and you have this phantom demand that is built up and it takes a while for everybody to dial it back down for all that's for people to cancel the orders. You know what some guys are doing, what some manufacturers are doing is they're saying, you are gonna give me a deposit on these orders and you're contractually obligated to take delivery of them. You know, so that's, they're trying to cut down on this, this kind of like, well, I only need five, but I'm gonna order 50.
Jon Stokes [00:17:14] They're saying, look, if you want to order 50, you can order 50, but you're gonna be out, you're gonna sign on the dotted line that you will take delivery of 50 and you will pay me for 50. And so, and, and there are OEMs that are desperate enough to do that, you know, so they are signing, you know, at these in inflated rates for inflated inventory so that they can take delivery of all that stuff. So there's a demand surge that is hitting TSMC. There's a demand surge that's hitting the semiconductor industry. There's a demand surge that, you know, hit lumber a while back and then cratered, and now it's hitting steel. So, so the this, this is kind of moving in waves across the supply chain. Shipping is a factor. So, so you asked me what the factors were. So what is this phantom orders? That's a factor. Like there's real demand from, from the stimulus, from the restarting of the economy. There's legit demand, there's phantom order demand, like this fake demand that's like, you know, puffed up out of panic. There are very real shipping constraints. China lost shipping capacity just like two or three weeks ago, lost a chunk of air freight capacity from COVID Cruise being out from COVID. They have ports closed from
Vance Crowe [00:18:27] COVID,
Jon Stokes [00:18:28] They have factories closed from COVID. So they have, you know, Delta is hitting them. The Sinovac vaccine is just not very good. It wasn't that great against, you know, version 1.0 of Coronavirus. And, and it's really not that great against Delta. So they have, you know, outbreaks, they are probably not able to control them. They have had flooding over there, you know, so they have had a lot of different challenges keeping their ports open, keeping their factories open and staffed and keeping their internal, you know, engine humming. So there's all kind of supply constraints that are going on there. So you've got shipping, you've got supply, you've got e the whole thing is a mess.
Vance Crowe [00:19:10] And then on top of that, you have businesses that are sitting on just piles of cash and they have so much, I, I work a little bit in the banking industry, and one of the biggest problems that they have is normally you go out and give loans and you gotta keep your deposits coming in. So you got plenty of money to be able to loan out. Well, that's not the problem now. Now the problem is the deposits are amounts are so high because people were saving cash, they weren't making, they weren't building. And if you're not making capital, like large scale build out a factory or build out more infrastructure, then what are you gonna do? Well,
Jon Stokes [00:19:45] You, you
Vance Crowe [00:19:46] Have money, but you don't have supplies. You're gonna be willing to pay a much higher price for it. So then that only accelerates the, the, the, the weird way in which prices increase and, and the phantom orders all combined together. It's, it's, it seems like a perfect storm on, on top of the fact that our ports were a little bit jammed up and now we've got a giant hurricane rolling through Louisiana. So just knock out all the, all the freight that was coming into and out of the Mississippi River. And that's all gone now too.
Jon Stokes [00:20:18] You know, this is, this is natural, natural disasters accepted. This situation that we're in now was, was foreseeable and foreseen when we started the lockdowns. Now, you know, people have their opinions on the lockdowns. I supported them on the public. I'm, you know, I might as well cop to it because I was on Twitter supporting them early on by May. I was like, yeah, this isn't working. We should stop doing this, or we should make them more targeted because we can't just keep locked down. But, you know, back when it was two weeks, four weeks to flatten the curve, I was like, yeah, you know, sure we can do this. But, but it was also clear that the economy and the supply chains, they're not like a machine that you can turn off and then flip back on and wh it on the side and it boots up and it starts going again. You know, this is a complex, interconnected system. So we knew that even locking down four weeks, we were doing damage to this, we were breaking this system in ways that, because nobody, it's a, it's self-organized, so nobody sees all the pieces. Nobody has complete visibility over the whole supply chain. That's why you and I are here talking and trying to puzzle together what's going on, and I'm getting dms and you're getting dms and everybody's trying to piece together what the supply chain situation looks like because nobody has, we're not trying, nobody has visibility over the whole, you know, all, all the flows, right? So, so, you know, it's a market and, and our market economy is, is that way, it's this complex self-organized ecosystem.
Jon Stokes [00:21:53] And we turned it off just suddenly. And so it was really clear, and I have a, I have a number of Twitter threads about this, and I'm, I'm pretty sure I wrote about this for the prepared.com, that when we went to open it back up and turn it back on, we were gonna see massive dislocations because everything is outta sync. You know, you've got consumers showing up with cash, but there's nothing to buy, then you've got things to buy, but there's nothing, nobody, there's no labor to build with, you know, and then you've got, you know, all these different parts and pieces have to start to come back into some kind of alignment. So the supply chain is like a microcosm of everything that's going on with like restaurants and McDonald's and, you know, the labor market and just pretty much everywhere. The whole situation is such that everything is trying to knit back together as fast as it can from this state of being paused, because none of this was designed to be paused. You know, none of the, no part of the modern economy was designed so that you could press pause on it and just everybody take a month off or like six months off and then flip it back on, and then we're all just rolling again. You know, it doesn't, none of it works that way.
Vance Crowe [00:23:07] And you can get these weird perceptions, right, where you're like, Hey, we, we handed out a bunch of PPP or we made it so people didn't have to go into work and like, look, everything was fine, but what you don't realize is the startup, again, so one of the knock on effects I think people are not aware of is there's not very many fast food workers right now. So if you're a long haul truck driver and you only have so many hours you have on the road from the moment you turn that truck on to when you've gotta turn it off, the, in the past fast food was a critical com compar component of the way that we got got goods delivered because we have these really strict rules on how, how long somebody can be out there. But if you all of a sudden have fast food places that are running at 40% of their labor now, the lines to get through fast food takes longer, which means those truckers aren't moving stuff and they've got to shut their trucks down. There's no, there's no if ands or butts about it. And so you have this little tiny thing where you think, ah, McDonald's doesn't have enough workers. Yeah, you wait in in line a little bit longer and you don't realize, no, that's stopping your milk from getting to the grocery store. It's stopping best buy from getting computers, it's stopping the components from showing up at the, at the factories. And to, to your point, like there are millions of these things, the the unforeseen stitching together is, is absolutely impossible to predict over the, the large scale. So when you think about the angle that you know about, I often say like you, you don't, you never really learned anything until you've changed something about your behavior.
Vance Crowe [00:24:38] What about your behavior has changed because of what you know about supply chain?
Jon Stokes [00:24:44] I certainly am somebody who plans their purchases now in a big way. So I mean, this, this started last summer where we knew, hey, the kids are, are probably gonna be outta school this year and they're gonna be remote and we're gonna buy Chromebooks, you know, before they go out of stock. And so we bought Chromebooks for the kids, like early buying some Christmas presents early, you know, not doing like the full Christmas thing, but just like trying to buy some of this stuff early so that we don't get caught in the rush. So, so really it's, it's just about not assuming that something that I need is gonna be in stock and planning to wait, you know? So right now we're trying to get a whole house generator for a guest house that we just built on the property that took a really long time because of supply chain and you know, there's no whole house generators right now, so, you know, you just kind of put in the queue and you cross your fingers and you hope to get one. And so it's really about changing your sort of consumption patterns to match the fact that there are just no, no guarantees that you're gonna get what you want.
Vance Crowe [00:25:54] So John, moving away from supply chain, I have it on just one of my connections was in a grain meeting, a very large one. And for the first time among all these farmers, as they're putting forward their ideas, somebody stood up and said, what happens if the US dollar is no longer the reserve currency? What would it take for the grain, our grain organization? It was a specific one to move into cryptocurrency. So you are in the, in the world of crypto, if you were advising, you know, some large industrial group to, to dip their toe into the world of crypto, how does one even begin this conversation? And is that the right place to start from? Like maybe the US dollar might not be the reserve currency that we trade grain in.
Jon Stokes [00:26:44] Yeah, you know, this is, this is a, that's a, that's a big question. I, I am not a crypto is gonna replace fiat kind of person. That's not my, that's not my entry point into crypto. My entry point into crypto is, is as a software person and as somebody who thinks about complex systems and resilience and things of that nature. So you have a, a sort of large camp of crypto people who are like, well, we want to use this as money. I, I'm not, I don't really care that much about buying and selling things in Bitcoin or, you know, whatever the flavor of the, the weak coin, I'm more interested in, can we build, can we build a resilient payment system that that is censorship proof? Can we build publishing platforms? You know, can we do interesting banking things that get around these toll collectors in the system, these choke points in the financial system. So some of that involves using crypto as money and, and some of it maybe doesn't. So, so if I, you know, was advising a, some kind of trade group in terms of thinking about that, you know, the real, the place I would start with them is to just Bitcoin, you know, just look at Bitcoin, you know, maybe set up some kind of working group and, and set up some kind of strategy around Bitcoin, because Bitcoin is still kind of what they call the shelling point.
Jon Stokes [00:28:18] You know, it's kind of the, the, the default. There are, there are other coins out there. I think Ethereum is great. I I like Ethereum. I'm, I'm bullish on it. There are so-called, you know, like next generation blockchains that are really nice. There are, there are, you know, proof of stake and proof of space and other kinds of, you know, more energy efficient blockchains. And I think that there's gonna be a role for those and a place for those. But it's still really early days, you know, if you're gonna make a bet on VHS versus Betamax, it's like too early to make that bet. But like, you know, Bitcoin is still like pretty good. It's like a pretty, it's a pretty safe, you know, consensus pick and it'll kind of get you started. So that's, so
Vance Crowe [00:29:06] Let's talk about the, that's
Jon Stokes [00:29:07] Kind of like how I approach it. Let's
Vance Crowe [00:29:08] Talk about the censorship proof, like payment processing, because I mean, having worked in the banking industry, I don't think people realize, like if you go and make a weird transaction, you know, you, you bring in, it's not just $10,000, it's like cash in a weird amount. The bank is obligated to put in a what's called a sars, a suspicious activity report. And that is only the beginning of all of the reporting that goes on and the, and that's the federal government's role in, in censorship. Are you talking about being censorship proof from that, or are you talking about cancel culture saying, Hey, visa you, if you, if you process these people's payments, you're a bad person because they're a bad person.
Jon Stokes [00:29:54] I, I'm, I'm talking about the range of it. You know, there's, I mean, I've, I've run into some of the stuff myself. Like I have a, I have an LLC that I use that has the word blaze in the name of it, because I was gonna build like a messaging platform at some point, you know, back in 2010. And I, I have like, you know, it was gonna be fast. And so I put Blaze in the title and you know, now this is a marijuana reference, right?
Vance Crowe [00:30:24] So, so, so I used
Jon Stokes [00:30:26] That, that, that I used that domain name for my LLC and it's registered in Texas and I paid my friend in Texas for some work. And, and when he got that payment, his bank closed his account and did not give him an explanation. And we, we, they just treated him like a criminal and we're pretty sure they thought that he was doing we business, you know, because of I'm paying him from this, you know, this account with Blaze in the name of it. And they're like, yeah, this is some kind of weed thing and you know, it's pot related and it's not legal in Texas and we're gonna close you. So he just got his checking account closed, boom, it's done well,
Vance Crowe [00:31:08] And what people don't realize is the bank's not even allowed to tell him, right? Like if they're reporting to the government, and it's one of those things the government says like, you keep your mouth shut about this, it's like a dangerous thing for banks to talk about at all. That is fascinating.
Jon Stokes [00:31:23] Yep, yep. So, so that was a, you know, that and not, but not just that, like moving up the stack, there's operation choke point, which took place, you know, in the Obama years where a lot of credit card processors and banks were trying to de platform firearm sales. So, you know, if you had a gun shop or something, you know, they wanted the banks to not do business with you. And New York was using, its, its kind of special status as a financial regulator, you know, local New York authorities to say, you know, to lean on banks to, to drop these kind of politically disfavored businesses like guns. So, you know, this is happening in the adult industry now with this OnlyFans, you know, de platforming that just got reversed. So, you know, these kinds of things, guns, porn, this stuff, like, you may not like it, you may like it, whatever, you may be libertarian and not care, but these things are kind of the canaries in the coal mine for the payment system, you know, so,
Vance Crowe [00:32:32] So what can crypto do to, to get around this stuff? Like how do you explain this to people that aren't familiar with the world of crypto?
Jon Stokes [00:32:42] You can, you, you know, with crypto, if you've got like some kind of payments layer where it's not all rooted through the banks, I mean, in, in the defi ecosystem, the decentralized finance ecosystem, you can get loans now you can get micro loans and you can do all kinds of like, money transfers that's essentially like cash. It can be anonymous, it can be synonymous. You can do know your customer, you can not do know your customer. And, and people, the first thing people think is, oh yeah, this is just for ransomware or for money laundering. But, but it's also for these kinds of politically motivated attacks where, you know, a centralized banking system just decides, Hey, we don't like this industry, we're gonna boot them off the payments, off off of payments. And you know, those of us that like, you know, firearms are kind of, are kind of my thing. I'm into guns and, and you know, I want to be able to, to go to a local gun store and buy a firearm and it's my legal right to do so. And if MasterCard doesn't like it, like I don't care if MasterCard doesn't like it, I don't care if Chase likes it, I shouldn't have to care if Chase or MasterCard support my second amendment rights. I should be able to just pick that stuff off, you know, if it's legal, I should be able to do it. And I don't want this kind of extra legal situation where MasterCard can come in and can censor me. And if I've got alternatives, you know, for, for exchange and money and for doing banking and for doing credit and stuff like that, which, which the crypto ecosystem is, is, is building and is providing, then I don't have to worry about what Chase thinks or MasterCard thinks about, you know, my shopping habits.
Vance Crowe [00:34:24] What is the time horizon for the, this kind of ecosystem to come up? I mean, you're saying it already kind of exists, but you really have to know what you're doing. You'd at least have to have a sense for where to go and who to trust. When does this become something that a motivated person, but that that isn't well connected, could, could start using,
Jon Stokes [00:34:47] You know, I think probably a motivated person can already use some of this stuff right now. You know, you can already access a lot of these apps and a lot of these kind of tools right now if you, if you're nerdy enough
Vance Crowe [00:35:03] So then, you know, you've got this motivated person, they can do it. Where do they even go if they wanted to be able to start separating from the, the more traditional banking credit system?
Jon Stokes [00:35:14] Yeah, that's a, that's, I I am not somebody that is, that has gone down that path at all. I have friends that do it, and so I can't give like a nuts and bolts, you know, walkthrough other than set up a Coinbase account, you know, kind of start playing around with that. Check out unis swap, that's a, that's a decentralized exchange. So if you just Google U-N-I-W-S-A or SWAP unis swap, that's like a, a distributed Coinbase. And so if you google, you know, defi intros, I mean pro probably this is something that, that I should write about is like sort of a more nuts and bolts intro. But I personally am not a, I'm still very much tied to the traditional financial system. Like I'm kind of waiting on some of these things to develop. I'm not an early adopter with this stuff or really with, with most of anything. I'm, I know too much about tech to be, to be an early adopter of especially something critical, you know, like, like for my finances. So, so I am not, I'm not the guy to, to walk anybody through like how to kind of unplug from the banking system and get loans on you to swap and that kind of thing.
Vance Crowe [00:36:31] So speaking of early adopters, you've written a lot about artificial intelligence and machine learning, and one of the big subjects that you were covering that I thought was like touching a third rail was people claiming that we're putting ethics and morality into the AI that are, you know, bad or not correct or biased or, so what, first of all, what prompted you to, to jump into this space and how did it go for you?
Jon Stokes [00:37:04] Yeah, that was a, that was kind of an interesting journey. So Google and how, how to, how to zoom out on this? So, so the, so the, the first thing to know about AI and modern, so-called large language models is that they're very, very expensive. They're getting cheaper. But with ai, we are kind of in the same world that we used to be in in the seventies with mainframe computing where, you know, the old bill gates or the fake IBM quote, like there's maybe a world market for five computers or something. You know, you see this circulating, I don't know that it's real, but I mean there's like, you know, a handful, a single digit number of really big massive ais that are at Facebook that are at Google that people can train and can use. So, so because these things are so resource intensive to deploy a really big model is, you know, takes the resources of a, of a Google or a Facebook. And because there's so few of these really big massive models, and there are these critical points, like Google is a critical node, is a critical entity in our life. We search Google, we get YouTube, we, you know, Google owns a lot of things. A lot of us do mail through Google. Facebook owns another big piece of our existence. So because these things are at these critical points, the ais and the shape of the ais that they use matters for billions of people.
Jon Stokes [00:38:41] So if you've got a platform with 2 billion people on it, and you've got an AI that's curating and use feed, or that is, you know, interpreting language or producing language for people to interact with, and it's, you know, when you tweak that, you've tweaked it for 2 billion people. And so, so that matters. And so what's happened is that there, there are so-called AI ethics people that have been hired at these platforms. And so they bring their ethical considerations to bear on these ais and the rules of government and their design. And so they are concerned, I think, correctly with what kind of world are these ais helping to create? What are the ethics of the use of language? What are the ethics around communication and curation and you know, filtering and stuff. What are these ais doing to the users, to 2 billion users? What are they doing to their minds? What are they doing to the way that they talk and think and act? And what do we want them to be doing? So these are all questions that matter, and they matter a lot because there are so few of these ais that, that affect so many people and touch so many different minds and eyes and screens. My issue that I, I came, what I came to be aware of when I started to study this was that all of the people that are doing ethics at these companies, they fit a very specific profile.
Jon Stokes [00:40:20] And you can probably guess what their ethical framework is. It is what we would call on Twitter woke. They are not Buddhist ethicists, they are not Catholic ethicists. They are not Muslim ethics people. You know, they are woke ethics people who have their norms and values formed by the internet and by Twitter, these people are often not trained in ethics. The bleeding lights of of AI ethics are linguists and engineers and computer science people who don't even have like a formal background in like a faith tradition or, you know, like even I think a graduate level ethics class in some cases. Not, not that, not that I want to gate keep and say that, that those are the only people that get to have opinions about ethics. But my point just being that like you take these engineers and now all of a sudden they have this brief of like, we are gonna govern the ethics of these complex systems and of these ais. And so then my question was where did their ethics come from? I personally am a Christian, and so I am wondering am I represented? And, and when there's a meeting at Google about the ethics that govern the AI that touches billions of users, is there a Christian in the room? Is there a Muslim in the room? You know, is there a Buddhist in the room?
Jon Stokes [00:41:53] Often the answer is no. There are a bunch of people with no formal ethical training who are having opinions in publishing papers about ethic and all their ethics are woke. And some cases I, I don't like them, you know, I don't agree with them about how the world should be ordered. I don't agree with their values, I don't share their goals. I don't honor their, their prophets or worship their gods. And you know, they're the ones that get to be in this room and have this, have these meetings and have these opinions about how these platforms are governed. So that, that was kind of the, the questions that, that was the impetus for me to start looking into this and to start asking questions about it and to start kind of wondering like, what are we doing here?
Vance Crowe [00:42:43] So John, what does it mean for you to have your ethics in, in the room and to be, to be counted?
Jon Stokes [00:42:50] Oh, here's an example. You know, some of the papers that are published on, on these large language models. So a large language model is a model that can for some definition of understand, can understand English or, or you know, some, some language French, whatever it's designed to work with, let's just say English. So, you know, Google has a large language model, and when you type in a search, Google tries to understand what the words meant. It's not just, you know, keying off of certain keywords and things that, you know, that was 20 years ago. Now it's trying to, it's trying to understand what does this sentence, what does this person mean when they, when they say when they type Mary had a little lamb, do they, do they mean she had a, a pet lamb that was small or does it mean that she had a little lamb and a little pea and a little, you know, you know, is she eating? Is that what that, you know, they're trying to, they're, so they're trying to, to pull in all the context of the language and understand what the sentence means. And what happens also with these large language models is they can go the other way. They can produce English. So there are large language models now like GPT-3 that can write an essay on a topic. If you give them a prompt, that is, you can't tell that a human didn't write it. And in fact, the first, one of the first essays I read on GT three, it was about T three, it was a description of it and how it worked and what it did, and some philosophical discussion of whether or not it really understood language.
Jon Stokes [00:44:27] It was written by G PT three, and I didn't know it. I thought, well, this is a pretty good essay. And it was only like later that I came to understand this essay about GPT-3 was written by the ai, and it's, it's actually quite decent. So, so you've got these machines that are understanding language and they're producing language. So a lot of the, the woke discussions are, and some people don't like this term woke. I don't know that I like the term woke. I don't have a better,
Vance Crowe [00:44:56] It's a pattern language, which we understand what you mean. Yeah,
Jon Stokes [00:44:58] But I'm just gonna, I'm gonna say woke. So, so some of the woke, a lot of the woke disputes are around language, you know, birthing people instead of, you know, pregnant women or something like this, right? So this is a classic, this is a good example. So, so then you'll be like, is is there a point at which when I type into Google, you know, resources for pregnant women, does Google say, did you mean resources for birthing people? You know, and auto corrects my, my, you know, it's like, is there are people that would like to see that
Vance Crowe [00:45:37] Happen?
Jon Stokes [00:45:38] You know, there are people that would certainly like for Google to slap you and to say, no, no, no, we don't use that language, you know, are are we're just gonna show you instead of, we may not correct you, but we're gonna show you a bunch of birthing people stuff. And if it is aimed at pregnant women, you know, that's gonna go further down in, in the, in in the results. You know, so, and so this is the thing that happens in some of the AI ethics papers is they complain, they're like, well, look, you know, this, these Google results are gendered and they really should be non-binary. And you know, this is an affront and this is violence. This is literal violence to non-binary people that this Google result that I typed up, you know, implies a gender binary. And, and so and so these people want, these are the kinds of people that are in the meetings where it's gonna be decided like, what my user experience for Google is gonna be. You know, I, I don't particularly care what they think about what is literal violence and what is not literal violence. I don't subscribe to their gender theology. I, I don't believe in an inner felt gender that people have or any, any kind of thing like this. So, and people from my tradition typically don't believe that stuff. We have our own theology of, of sex and of gender. You know, that goes back to St.
Jon Stokes [00:47:08] Paul and Galatians and, and we have our own ideas about how the universe is structured and about how human society and human community are structured and, and about the kinds of things that we want to see reflected in the language. We have our own linguistic practices and we like them and we would like to keep them, and we would not like to have them colonized by woke people who are, you know, working for Google. So, so when I say I want my faith community in the room in that meeting, I mean, I want someone that I share some kind of recognizable linguistic reality with instead of commitments and understanding of the world to be in the room and to say, well, look, you don't get to dictate for a couple of billion users how they're gonna think about gender and how they're gonna speak about gender and biological sex and these kinds of things. You don't get to just decide that unilaterally and impose that on, on, you know, billions of people you're gonna have to reckon with, with, you know, dis representative of Christians that use Google or Muslims that use Google or Orthodox Jews that use Google. So you don't get to just, you know, wave your hand. But these, but these folks, if you read their papers and you read their output, they would like to do this. They would like to dictate for the rest of civilization how we think and speak about basic categories of male and female and human.
Vance Crowe [00:48:35] I think one of the big challenges people have when they start talking about things like ethics is they say, well, just because you have studied this doesn't mean that you're more ethical, but what you don't realize until you go study, like I, I had a great Jesuit education particularly focused a big time on philosophy and theology. And one of the very first things in very first class of logic was he who controls the definition controls the argument, and the words matter in the way that you describe things and what, how you decide to define them. And it seems as though what you're describing is almost a, is like a tower of Babel, so to speak, right? It, it, it's like we are going to change the way that language is. We're going to actually make it based on our reasoning, which to us is, is as pure as it can possibly be, but it's completely disconnected from any, from the, from the hundreds if not thousands of years of, of moral traditions. And it seems like people are, are dancing on the, on the grave of religion in some way, forgetting that it's not an accident. That, that these religious faiths came about. That, that they, that the, the, the sacrifice that people had to make in order to believe in God or in, in order to follow the teachings of their religions were there because they were lessons hard learned. And when you don't apply them, bad things can happen.
Vance Crowe [00:50:06] So it's amazing to me because I had never thought about AI being able to institute this change in society so much faster than even just, just social media itself or, or the change to our universities that AI could do it on the scale of billions of people within hours of being instituted.
Jon Stokes [00:50:28] Yeah. That, that's the, that's the level of power that we're talking about. And you know, my, my fix for this frankly is, is that there shouldn't be that much power pooled in the hands of this one tiny and undemocratic institution. You know, if there are going to be people who are making decisions about how an AI that interfaces with billions of users a day speaks its mental categories, its gender categories, it's kind of conception of, of reality in human community. If there are gonna be a room full of people making calls about that, I wanna know who those people are. I wanna know what their, what their, their ethics and their commitments are. And I want somebody that represents me and my faith community to, to be in that conversation. You know, I don't always have to win every argument, but, you know, I, and, and, and because because again, as you say, within a matter of hours with a, with a, with a deploy of some code, you can make a change that has this kind of massive ripple effect across humanity.
Vance Crowe [00:51:43] I think it was in your Twitter feed where it was talking about the number of people like positions. So Harvard used to have Harvard Divinity School, and now it's being run by an atheist, or what is it? Princeton was traditionally a Presbyterian school and either they have no minister now or they also an atheist. It's a fascinating thing to think about some of these storied institutions moving away from, from their religious tradition. But when you talk about changing Google or pushing back on Harvard's decision about this, how likely do you think it is that, that you and your religious tradition can win out here?
Jon Stokes [00:52:26] Yeah, that's a tough question. I mean, I, I mean Harvard is like a separate matter, you know, just to speak narrowly to Google. I mean, yeah, I would love to be able to put pressure on them as a Christian to say, look, you need to either fire half of your AI ethics staff, or you need to double the size of it. And either, either way, whatever the proportion of Christians in the population nominal or otherwise is here, we need representation on that staff. You know, I would like to be able to do that, but I can't. And I also market, it's hilarious
Vance Crowe [00:53:06] To think about this to, to, I was just having a conversation the other day with a guy that I had a neighbor once who, who was like, Hey, whatever percentage of the population is black or Latino in a community, then they should be made up of equal representation of police officers. And he just thought that, you know, just assert this. If you tried to do that with religion in order to, to push back on the woke to say, Hey, we need to have X per X percentage, be Catholic and Protestant and Muslim people would lose their absolute minds over that. They would think they would paint that as the most racist discriminatory thing that you could possibly do. And yet people are totally willing to do that over something that we, you know, like skin color. So it's fascinating that, that at that kind of logic there.
Jon Stokes [00:53:58] Yeah, I'm, it's funny because, you know, for me, I am, I don't, I, I understand the analogy, but like, if you have a committee that's supposed to be making ethical decisions for a super powerful AI that touches, you know, 4 billion lives and that committee has 15 people on it, you know, we do gotta talk about how it's made up and how it's selected. This is different to me than a discussion about representation in a boarding school, a representation even on a police force or in some other kind of capacity. You know, these are people that are appointed specifically for who they represent. You know, the people that, that are on this AI ethics committee, like I said, now they don't have a, they don't have any, any particularly excellent training in ethics that I can tell. You know, they're pointed based on these kind of like, I identitarian representational grounds, and if we're gonna play that game, then let's have an AI ethics committee that looks like the country and doesn't just look like some elite circle of activists or academics. You know, if we're doing a different kind of situation where it's like nominally a meritocracy or something, or where, you know, we're having elections, you know, or something like this, well, you know, maybe this is different. But, but yeah, I mean, I, I, so I can see where people make this analogy or where one would make this analogy, but for me, like this is a pretty narrow and weird case.
Jon Stokes [00:55:38] You know, it's a private company and the decisions made around ethics and AI are touching billions of lives, and they're being made by like, I even, at least I want to know who's making them and what the process is. You know, I want to have some visibility there, but, you know, I can talk this way, but none of this is gonna happen. You know, realistically speaking, you know, I can make all kinds of fantasies about, you know, I wanna see a Catholic theologian or a Pentecostal theologian or you know, a Buddhist or a Muslim theologian on this AI ethics committee. And, you know, there's no, i, I have no leverage and like, no one has any leverage because, because these companies are essentially monopolies and they're, they're quite well entrenched. So the only thing that's gonna happen is we're gonna have to move to alternative platforms. You know, the market is gonna have to speak in some way, and this is part of what interests me about crypto, and it's related to this AI ethics stuff. Like I am pro decentralization. You know, I have historically been pro antitrust, you know, I've been like, well, we should just break Google up. We should just break Facebook up. But I'm, I'm even happier to say, well, we should build alternatives to Google and Facebook, you know, that are decentralized and to where no one person controls it. So, so I, these things are kind of possible in the world of crypto.
Vance Crowe [00:57:12] Do you think AI is, is dangerous overall? Do you, do you think the the sum total of the good that will come out of it will, will outweigh the potential dangers of the things you're talking about, or even other dangers associated with ai?
Jon Stokes [00:57:29] Yeah, that's a, you know, I think, I think AI can be dangerous and, and it's certainly is likely to be dangerous, not in a Skynet kind of way, but, but I'll tell you, I'll tell you the, the imminent pressing danger of it. And I am, I am pro AI and AI development. You know, I'm not advocating a lot of legal restrictions yet on it because I think that we may innovate our way out out of some of what I'm about to describe. When you have a, the legal system, there are judges all over the country. People elect their judges, some of them are appointed. There's a slow process of negotiation and it's human. And judges make, you know, sentencing decisions and they make them in this distributed, organic, crunchy, slap dash fashion. There's apparently studies that show that if you are up for sentencing closer to lunchtime, you know, you get more jail time than if the judge is fresh and happy early in the morning. So, so, you know, judges have their decisions reviewed, so on and so forth. So, so each judge is an individual human. There's systemic pressures, there's local pressures. There's so many different things that go into that sentencing decision. Some judges are corrupt, you know, and are on the take and sentence people, you know, we had a case in, I think it was Louisiana, where a judge was sentencing kids to juvie because he was being paid by the institution that ran the facility to, to pack the, to pack it.
Jon Stokes [00:59:12] So then let's say that you wanna solve that problem by introducing an ai, and you're like, we're gonna have a sentencing ai, and the sentencing AI is gonna look at a bunch of signals and is going to come out with the optimal sentence of justice for all these different cases. And so we're gonna take a lot of judge discretion out of the loop, or maybe this sentencing AI is just gonna be advisory and is gonna suggest to the judges what the senate should be based on, based on a fact pattern. And we're gonna program it and we're gonna pattern match it. And the AI is gonna look at the case and decide, this person should get five years, this person should get two years, this person should get community service. Right? The minute that you offload that decision to a sentencing ai, you can think of that sentencing AI as one judge. So it's like if you had one human judge who was in the loop on every, you know, hundreds of thousands of sentences a day, that one guy, that is a lot of power for one guy, right? And now we're gonna give that power to a machine that was programmed by who, what were their interests? Like what, what did they use to train on? Is it auditable? Did they test it? How did they go back and like, validate it, you know, what were the benchmarks for if this thing is good or if it, how, how do we know if it's not performing? Like how do you know if a sentencing AI is screwing up? Like what are the criteria for this thing is broken?
Jon Stokes [01:00:45] Go back and tweak it. Are the criteria published or are they internal? You know, what are the success goals and metrics? What were their targets when they trained it? We don't know any of that. And yet you have these sentencing ais that are there. And a hiring AI is the same problem. I have a friend who just got a job from interviewing with a robot, and you know, you've got this instead of just hundreds of thousands, millions of HR managers out there given an interview. Imagine that you've got one super powerful superhero HR manager who was able to clone himself and be everywhere at once and all of his foibles and all of his shortcomings and blind spots and biases. And, you know, every limitation that that woman or man has as a hiring manager is now replicated across the entire economy into like millions of hiring decisions. Is that what we want? Like really, is that better than having like the disorganized chaos of a mess that we have now to organize and pull all that behind one set of blind spots and one set of weaknesses and one, one category of fallibility? I don't know that it is. So, you know, this is a, this is the, the, the conundrum that we're forced into with AI right now and many, many places, loan applications, hiring, medical care, sentencing, all of these places that I mentioned. And there's an AI somewhere that some company made that's a black box that nobody knows how it, like, how it was put together and if it even works properly and it's being sold and somebody's using it as an input into all of those places.
Jon Stokes [01:02:28] And you can think of it again as one superpowered person who is limited. And I mean, in antiquity, you know, Zeus is like running around and he's like raping people and he's stealing things and he's being a bad guy. And the Greeks had this idea of the gods as sort of mega humans, you know, they were like very powerful. Zeus is very powerful, but he's also got a lot of problems. Zeus is a messed up guy in, in the mythology. And, and Lucretius actually has dialogues of the gods, which I read in grad school where he tells a lot of these body tales of the gods. And the point is to kind of lampoon and to poke them a little bit and to highlight how in, in this, the, the Greek stories the gods are, are almost too much like us. They are just really powerful versions of us, but they're just as screwed up. Like this is kind of what we're building with these ais is we're building really powerful versions of us that are just as screwed up and in some cases even more limited and
Vance Crowe [01:03:36] Well, and like, I mean, to your point, like I, the Greek gods are in many ways all of the voices in your own head, right? And one of the benefits of having a human being be a judge is that they had to go through things too. They, they have some form of, of mercy saying, you know, that if I had been in that situation, I could see myself there. And you know, all of the factors that go into the humanness of making judgements are really deeply important. And certainly they're flawed. But you know, to, to your point, the other day I watched a guy being sentenced and they said, and he was wearing a mask for COVID, right? So you could only see the very edges of his eyes. And that was it. And to me, I was like, we have absolutely hit a bizarro world in which we are having a judgment in which jurors only see this person's eyes. Now, do you need to see somebody's whole face in order to be able to judge whether they're innocent or guilty? I don't know. But something feels very dystopian and dark and not, not humane when it's either a judge that isn't real or a person isn't physically standing in front of people, they're on a computer, you know, closed, you know, closed circuit camera, or they're wearing a mask. A lot of these changes that are going on, certainly for incrementally correct reasons, or in incrementally you can understand why you would make that decision. It seems like a dystopian thing, but I had not thought of how big of an impact AI could have on things like court cases.
Vance Crowe [01:05:09] Sure could make the backlog go down faster, but do we really want that?
Jon Stokes [01:05:13] Yeah. You know, there are some things that, that are, that should be slow and, and process riddled and bureaucratic maybe by design. You know, people think that that's a bug, but maybe it's a feature. Maybe the slowness of some of these human systems and the fallibility and the unevenness of some of these human systems is, is a, is a feature. Because what happens is that we, we don't have an agreed upon concept of justice in this country, you know? And, and prior to the rollout of these mega godlike ais, we didn't necessarily have to, we could have local justices and injustices. You could have a regional concept of justice where certain things happened, and you could have people in another region who were like, yeah, I would never wanna live there because you can't get justice. But then somebody else is like, well, the judges are this way, you know? And so these things could be messy. We could have differing opinions and, and you, you could kind of, you know, we can manage the complexity in a certain way. And, and you can live in an uncertainty where, and when I say we don't, we don't have a, we don't have a, a a, a concrete conception of justice. You know, look at, look at the, you know, the triage concerns around like vaccination versus unvaccinated. There are people that believe that if you're unvaccinated and you get COVID, it's your fault and you need to go to the back of the line for care.
Jon Stokes [01:06:52] There are people that believe that if you are unvaccinated and you get COVID, it's the same as if you were vaccinated. It's not your fault and you just go in the regular queue. Okay? Not taking a side on that, just saying these are, these are real views that you've encountered. I'm sure I've encountered these views in COVID. So you've got conflicting and fighting understandings of what's fair. One guy says it's not fair. Like, look man, I I, I I will take a side. I am, I am in the side of, it doesn't matter if you're faxed or un faxed, if you're in a hospital, you should get care and doctors should suck it up. That said, if I had a kid who needed a, who needed a bed for COVID, you know, and there was a voluntarily had no medical issues, just decided they didn't wanna take the vaccine person, you know, that was, that was in the bed and there was no bed available for my kid. And the hospital was completely full up of people that had just completely turned down the vaccine. And, you know, they, they decided that they weren't gonna do it, and they didn't have any preexisting medical conditions around it. They just didn't like it. I might be irritated, you know, I might be like, look, man, my child is, needs a hospital bed. You opted out, you opted not to reduce your risk. Anyway, whatever you think about what I just said, you can see that there's like real, like meaningful conflicts there, man. There's like, I understand both sides of this. I get both sides. I get a parent or a person a, a a that has a spouse that needs medical care that can't get it because the beds are full of unvaccinated people.
Jon Stokes [01:08:29] I get that they're angry that they think that's unfair. I get the other pe the other side of it that says, you know, I'm entitled to medical care just like that, that guy's spouse, you know, I get that too. We are fighting over this. We're in conflict over it as a society. We're, we're, we're probably not gonna resolve this. What happens in the legal system is when you have those kinds of conflicts, you can have a situation where in one area in Chicago, they resolve it one way. In New York, they resolve it a different way. In Texas, it's, it's, you know, it's different. So if you just can't stand it, you can move, you can move around, you can kind of, like, you can, you can manage it, right? You can find a place where you can fit. The minute we introduce an ai, it has one set of values, it's gonna come up with one single answer to that question. And everybody everywhere is going to use that answer, right? And there's no room for, for wiggle room. There's no like, well, you know, crap man, I can't take this anymore. I'm, I'm moving to New Hampshire. I can't take this anymore. I'm gonna go to the suburbs where like, you know, they have a, they live a different way of life. You know, if we were to make those medical vaccination triage decisions with an ai, if every hospital in the country is deciding who gets what kind of care by consulting a single ai, then that is one super powerful policy that one, one person, you know, an AI is making and enforcing.
Jon Stokes [01:10:05] And there's no, there's no room for, for that conflict to kind of flourish. And for us to work it out the way that we do, the messy way that we do as humans, it's just been decided, boom. Like this is the decision. This is the policy. We're gonna enforce this everywhere. And the minute that that code is deployed and the minute that it's pushed into production and everybody gets the latest update and begins using that new decision, then you have just changed policy across the entire hospital system, the entire court system, the entire YouTube content moderation system, the entire Gmail spam filtering system, the entire Facebook newsfeed system, you know, on and on. I could go on and on behind every one of these things. There is a super powerful ai, and when somebody makes a policy tweak, it goes out to everybody. And there's no option to like run away from it. You know, there's no, there's no way to change it or just fight it out or work it out.
Vance Crowe [01:11:02] John, I have so much to say off of that, but I want to be respectful of your time. This is, I think you're dead on about that. I am very, very suspicious of centralization for all the benefits that efficiency brings. E everything good comes from the conflict. Everything comes from the order and the chaos mashing together and, and a new better answer coming out. And I would love to have you back on, man. This was a fantastic conversation. If people wanted to read the stuff that I was reading by you, how would they go about finding you?
Jon Stokes [01:11:34] John stokes.com. So JON without A-H-J-O-N-S-T-O-K-E s.com.
Vance Crowe [01:11:42] Well, John Stokes, I had been looking forward to this interview for a very long time, and you did not disappoint. So thanks so much for your time and we'll have you back on again sometime soon.
Jon Stokes [01:11:52] Thank you. Enjoyed it.
Bring this conversation to your organization. Vance Crowe speaks to conferences, boards, and leadership teams on economic uncertainty and succession and knowledge transfer.
Book Vance to speak