AI Geopolitics, Token Costs and Public Benefit

Welcome to the Tuesday edition of Techstrong Gang. I'm not Alan Schimmel, nor am I Mike Azard. I'm Jon Swartz, senior content writer at Techstrong Group, where I'm based in Silicon Valley.

And hey, the NBA season starts today, for anyone interested. But I thought I'd just pass that on, because I know Alan would've mentioned the Knicks are going to defend their title. But nonetheless, it's Tuesday.

So if it's Tuesday, it must be Bedlam, because that's our news cycle. And here we are to sort through it all with our panel again. It's a regular Tuesday group.

This is a very, very solid group that shows up every Tuesday, so kudos to them. And we're going to start off by going around the horn and welcoming everyone. First, we're going to start with Kate Scarcella.

Hello, Kate. Hello, happy to be here. Are you in the Northeast?

I am. I'm in Maine, and I can walk to New Hampshire. Oh my gosh, that's so nice.

Well, welcome. We're happy to have you again. Sid Nag.

Sid, you're in New York? Yeah. Good morning.

Good morning. It's good to see you. And finally, last but not least- Please.

Chris Blask, brother. That didn't come out well. Least but not last.

Least but not last, yeah. Hello, hello. Hello.

Sir, you're here in Canada again? Yeah, specifically the Niagara Peninsula, about 40 kilometers from the Welland Canal. Mark that name.

Look at him using the metric system. Well, we're going to have an international tinge to the first topic. We're going to start on the international stage, because we've got a lot of things happening in Russia, China, US, what have you.

So I'm going to try to summarize this before I hand it off to Chris. There's a lot to digest, as there normally is, so I'll try to be as succinct as I possibly can. Russia and the US are discussing plans to jointly develop data centers, potentially powered by small nuclear reactors.

Oh, boy. So we've got Vladimir Putin's investment envoy disclosed the negotiations recently in a conference in Moscow. They follow a recent agreement between President Trump and Putin to resume large-scale Russian diesel shipments to US and global markets.

At the same time, Russia and Ukraine have escalated military attacks on each other's data centers, because data centers are now basically being targeted alongside power plants, oil refineries, and other critical facilities. And then finally, we've got China, where the country plans to restrict or ban the export of 40 critical technologies, including AI, semiconductor design, and renewable energy, in a hardening of its tech governance strategy. The policy signals a decisive shift in Beijing's stance from managing domestic AI use to establishing outward controls on Chinese software and intellectual property in its tech rivalry with Western nations.

Chris, this is kind of overwhelming. We can start with the data centers and the Russia and the US, but I'm going to leave it to you, because you are our international expert as far as I'm concerned. So, look, for years we've talked about the cloud as if it were something floating above our heads, right?

But the cloud is buildings, electrical substations, cooling systems, chips, cables, people, and political jurisdictions. Right now, we're discovering all those things are strategic assets. So the interesting question isn't anymore just who owns the data center.

It's who controls the relationships that make it possible for that data center to keep operating to the first of several of the points that you raised. And as you say, I've had a lot of exposure to this sort of thing. Before the pandemic, about seven years ago in Colombia, I got the folks responsible for the national electrical infrastructure, keeping the lights on, to sit down with a major cloud hosting provider.

And I have these conversations like, "Yeah, that's nice, but physically, where is it? " And for OT people, this is always a little bit easier to think through than for IT people, because they're dependencies, right? And the five nines of IT are laughable, right?

So, we built the internet, right? We connected the world, right? And this is, not to dig too far into history, but post-World War II order, right?

This interdependencies of global trade and international norms and so forth, giving rise to things like what we're using to talk right now, right? But again, it's not a nebulous thing. It exists in places.

And with being the era of AI, where your data sits, who owns it, who controls it, what can you see, what can't you see, right? I've said enough on these Tuesday shows on all that. But these are real things, and we have to think through them.

So how this plays out for any individual party, nation, state, and so forth, will have a lot to do with how well they think through their actual dependencies. I want to say, can you actually write this? Who would've thought...

Can you imagine how you get on this storyline? Yes. Come on.

What's happened? Where did these alignments come from, and how are we not more upset about this? I understand that things are not in the cloud, and I understand the physical data structures that we have throughout the world, but it's not okay.

We already know, even as a person who deals with critical infrastructure and the North American Electric Reliability Corporation, the NERC requirements, having that data, it was a big deal to have a server in a data center and have the electronic security perimeter. And everybody understands electronic security perimeter, at the end of the day, is made up of vulnerable applications that you can actually easily get into another environment. If we see anything about AI, it's how it takes advantage of the software vulnerabilities that we have today easily, and it does it quickly.

Yeah. Now you're talking about having our adversaries- Mm-hmm ... to be a part of this?

Mm-hmm. Yeah. It's crazy.

Yeah, go ahead, Jon. Oh, no, I was just going to say, I'll get out of your way, Sid. But Kate, I was thinking about this.

Cold War, salt talks, nuclear warfare, and we're working with these people. So I'm sorry. Go ahead, Sid.

No, I was going to say, is this even a real thing? We've got sanctions- Yes. against Russia, right, officially, from a US government geopolitics perspective.

Then it seems like some sort of a side deal is being made between two individuals here. I think the individuals are named in the story, so I don't need to repeat that. And then you've got these data centers that are being built.

It's not clear what are the mechanics of the data centers, right? A typical data center investment, it seems, uses access to electricity, equipment, financing, software, technical support. Geopolitical changes can disrupt any of these things in a moment's notice, right?

So, where does it actually land is not clear to me. There's no disclosed locations, no funding commitments, right? There's talk about using nuclear energy.

Okay, fine, that might just be a thing to do in the context of the shortage of energy that we're witnessing in the world today. It sounds interesting, but I don't even know how it actually is going to play out. Can the operator actually replace things like accelerators, maintain cooling equipment, all the security issues, ensure the facility?

There's so many things that go into- There's so many ... a data center. It's not like you're building a pawn shop next door in Maine.

Yeah, you point out the absurdity of these accords. Most of what the Trump administration does, there is no detail. Maybe it's a one paragraph, one sheet, one page- Yeah, it's like a memorandum ...

nothing behind it. But still, I wanted to ask you, you brought something up, Sid, that I wanted to ask anyone who wants to answer this How might this accord impact global technology alliances and energy security, specifically given geopolitical tensions, if this were ever to amount to anything? Well, it has, because we're talking about it, right?

So, I try to get really cold-blooded about this, advisory and analytic services, right? We need to be able to stand and look at you, whoever you are, we're helping with your protective practices and think through it. Do I personally have opinions on this?

Oh, boy. Yeah. Right.

And those are pretty public, and Alan's not here, so I can channel him if we feel like it, but- Oh, go ahead ... it's not very productive, right? So if we look at this as, I'm an information security person.

That's not just cyber, that's information. Right? We're living in a world where narrative payloads have impact, right?

So a one-page document put out by a significant public figure can get hundreds of millions of hours of discussion and consumption among folks like us, right? That's something that if you were just trying to market a product is a huge goal, a huge target, and we're all very familiar with the post-Cambridge Analytica world and yada, yada, yada. So there's that level.

I'm not going to disagree with anybody here, do I think these data centers are going to get built? No. I think it's a provocative action to get to your point, Alan, which is policies, alliances, and so forth, to push groups and people towards and away from each other.

No, it doesn't make any political sense whatsoever for the Russian and US governments to co-join their... No. I mean, it's crazy.

The China piece is also very interesting. I don't know the Yeah. No, we're going to get to that.

It's just like we're getting carpet bombed with this stuff. Exactly. Okay.

I'm going to shield everything that I have inside my border, be it models, be it software, be it silicon, be it whatever else, and the only way you can access all of this is through APIs. Yeah. Who's going to write those APIs?

What is going to be captured via those APIs? And obviously China's retaliating because we are doing the same thing on our side with denial of access to Nvidia chips to China and all of that. So that's pretty understandable.

" I mean, this insanity. Go ahead, Kate. I think I interrupted you.

Yeah. For these Russian data centers. No, that's the ridiculousness of it, though, right?

Yeah. Yeah, and this whole thing with China, it just... Oh my goodness.

That's what I'm saying. If there's a writer out there, I'm telling you, you have the best Hollywood script, just like with this headline. This is like, come on, Paramount.

Come on. Yeah, they're going to reboot the Bond series of movies. There you go.

This could fall. Data centers, in a sense, am I crazy? Are data centers political pawns now?

Oh, yeah. They're used within the midterm elections by both sides. They're being carpet bombed- ...

in Ukraine, and now they are going to be used between the US and Russia ostensibly in this idea. Are they just like the new hot commodity to channel our anger and frustrations to? Well, they're carpet bombing Ukraine, but on the flip side, Ukraine is carpet bombing Yandex data centers in Russia.

Yes. Where all the bots are sitting, allegedly all the bots are trying to influence the midterm elections, so go figure. Yeah.

Like becoming a battle of spiders web. I tell you, this is a phenomenal script. I love it.

This is like, come on, what else is next? Okay. Think of a classic villain for the Bond movie.

If they reboot the series, you have an Elon Musk type character. Or maybe just call him and just have someone who looks like him play the role of the new Blofeld, right? Yeah.

I will disagree with Kate. Don't think we've ever done that on air. It's okay.

I think the script is entirely predictable. I think it's inevitable. Right?

We've followed the narrative for the last, call it 10 years, year and a half, however you want to look at it. " It's like, yes, that's sort of the obvious next step, right? And- I didn't see this coming.

I really didn't But here's the thing, right? These data centers are obviously not going to be used by governmental agencies, right? Because there's no way any government is going to use another data center that's sitting in an adversary's country, right?

So that's out. So the people that are actually going to end up using these things are allegedly enterprises, right? So think of you being an enterprise located in the Western world.

Are you ever going to run your banking from Jeep North Chase? Am I ever going to use a data center sitting in Russia? Who's actually going to use these data centers?

What is the use case? No one's talking about it. " But where's the use case?

Where's the demand? I don't even see that. So it's- The logic of the physical thing, I think we all agree on that.

I don't think that'll ever happen, right? The script is being written every day. We're living in a reality TV show that's well-scripted, right?

Yes. And as Americans, we're familiar with the Soviet Union or the Chinese government doing that. And look, you can pick on all of Western liberal democracy, the US since World War II, and there's righteous digs on how honest our public conversation has been.

But I don't think there's a... Who would deny that right now, things like this, they follow the narrative arc, right? Specifically, no, it's not going to actually happen, but the great piece of outrage to get the right and to push the alliances and so forth, and it worked perfectly.

All the technical things they talk about don't ever happen. They don't need to happen. They're not part of the script.

The Russia-US accord or whatever this agreement is sounds to me more like a fever dream. And in a sense of what we're doing, what we're going through is you're right. These are just concepts.

These are ideas that will never come to fruition. They're just to agitate, to distract us, whatever, what have you. And they also play within this obsession with our administration creating as many data centers as we possibly can, not just in the US, but everywhere.

That's just my opinion. But we'll see. We'll see until the next outrage, right, Kate?

Yeah. That'll be tomorrow. There's one every day.

Exactly right. What time is it? Yeah.

So we'll move on from that to something that I think is a little bit more real. So Accenture has introduced a new funding policy restricting employee access to Claude, capping central funding at $29 per worker per month to curb soaring enterprise AI costs. The policy shift, which took place effective October 1st, highlights this growing industry-wide challenge of managing operational expenses of consumption-based AI services, which we all know is tokenomics.

Some Accenture employees, particularly developers using coding assistance like Claude Code, are burning through 500 to $1,000 worth of AI usage monthly. So Sid, we've heard about rumblings of this, not just through Accenture, but other places. I think Uber was another one.

I'm wondering where you think this is leading to. Even more companies adopting similar track strategies? Yeah, I don't know how you actually make this work.

$29 seems like a very low number to give to a developer. Yeah. I was thinking the same thing.

I can just exhaust my token quota in a real New York minute. I don't even know how it's going to work. My view is that this moves accountability closer to the work, where project owners to justify consumption against delivery value.

So if I can show value in what I'm doing, be it if I'm a developer or whoever else, these sort of hard limits is going to insanity. So that's one point of view that I have. The other thing I think is the agentic AI costs depend on actually workflow architecture rather than model pricing.

So, the article talks about employees, it was said, $500,000 through coding assistance. My assessment is that repeated rethink, repeated reasoning, expanding context, tool calls, testing, all the nitty-gritty that's involved in writing code using agentic AI or AI agents is going to be expensive. The real trade-off that one needs to consider as opposed to looking at these things in isolation is if I were to hire a set of programmers to do the work from scratch, how much would it cost me?

It's going to be thousands of dollars. I think that's the real tokenomics that we want to think about rather than tokenomics in isolation of the new world orders in the world of AI, that people are thinking in a very isolated, in a vision world manner. I also think that AI changes the economics of consulting contracts and delivery margins.

The article actually talks about this. The tension between fixed price engagements and hourly billing. So if I'm a third party or a service, like IT service provider or a GSI that's writing, developing code for an enterprise as a outsourced entity, then how does that play into the economics of the pricing model for this GSI?

So that's a whole different ballgame. Are they going to get a different tokenomics quota or versus the internal employee? Those are the questions that come to mind for me.

Yeah. I just have a brief follow-up question to you, Sid. So if I'm an enterprise, how do I balance this cost containment measures of consumption-based AI tools against potential drops in developer productivity and innovation?

I think you would have to think about-- I'm assuming this quota that they're handing of $29 applies, it talks about per employee, it doesn't say which department. So in other words, you can't have a $29 quota for somebody who's doing a task other than writing code. The quotas are going to need more tokens, in my opinion, because of the nature of the work, versus somebody who's doing a well-defined and less entropy-oriented tasks.

So I think that's the way to think about it. I don't think you can slap on a mandate saying every employee will have $29 worth of tokens going forward. I think that's the way to think about it.

Okay. I was going to ask maybe Kate or Chris, whoever chooses to weigh in. So when this tokenomics-- I have a hard time saying tokenomics.

It's such a weird word. Anyway, as they become a central operational challenge for large organizations, what long-term pricing models or resource allocation frameworks can AI vendors and enterprises adopt to make enterprise AI use predictable and sustainable? Chris, you want to go with that?

Not really. I was going to take a different break. Okay.

But- I'm happy to. Yeah. Go ahead.

Yeah. I think it's almost like speaking to a lawyer. The cost of the AI is going to go along with the cost of consulting.

I think, whenever you're dealing with a lawyer, you have your paralegals, you have that structure where you have, the lawyer costs this much, the paralegal will cost this much, and the admin costs this much. That's how I think it would be structured going forward, where the cost is not absorbed by the GSI, but it's going to be absorbed by the company that you're consulting. So, and if I were the company that I was receiving that consultation, I would put a limit on how they're using their AI.

And so it's-- It wouldn't just be open-ended for me. It's just like with a lawyer. More than a year ago, I know it was on this show because as we talked about in the green room, this is my weekly Tuesday tick mark.

And I just started this company. I made some statement along the lines, I think that, the power usage, with AI will go down by 90%, and that maybe we can build things that way. And this particular metric for me, I'm having a lot of fun as a technical founder.

I have not had this hands-on technology in decades. I produce just ridiculous amounts of code every day, every week, every month, just myself, much less folks around us. I don't spend any money.

And it's not just because I have my own tools, but I don't spend any money on coding tools at all. And it's always sort of fascinating. My very good friends and peers who are better engineers than I am, who have to stop and wait can only spend so much per month and so forth.

And I think it really does come, we have this brute force capability. You can take a large language model capable of saying anything, writing any line of code that's in there, and just beat it until it submits. Or you can actually look at the workflow where you're trying to achieve an outcome.

And, I'll remain humble until we can prove this everywhere, but this metric itself, I'm not spending anything on coding. Yeah. Regular tools focused properly are amazingly productive.

But most people I think are just-- And it's not anybody's fault as such. No, I agree with you, Chris. And I think moving forward, you will see more localized language models.

Would that be a fair assessment of Quiet Wire? Is it- Well, yes and no. The platforms exist.

Look, we use most of the AI platforms out there and Google and GitHub, Micro VL. These platforms are handy. Yes, and we also will do it on a laptop, Raspberry Pi, completely all fine.

It all depends. But I think it has more to do with just narrowing the focus. If you have an AI system, you have a very precise focus, and you can produce, in this case, lots of code without running through quotas and burning 10,000 tokens per second to get a line that says if then that.

And I think that your line of business will grow. I think as it gets out, it will grow because of that. I would say yes, but never believe startup founders.

Check everything twice. Yeah. So you mentioned AI coding, Chris.

So, how might Accenture's spend capping policy influence how software engineering teams choose and optimize AI coding assistance? Well, that gets us into this, the mismatch in time. Because if I'm a big company like Accenture and I'm doing planning for my budget for the next 12, 24 months and so forth, I have to look out over the horizon in the times where, in reality, the tools that people are using in 6 months and 12 months may be so much different than they are from 6 months and 12 months ago, or more so.

So I think most organizations will over-budget, I would think. I think the efficiencies we're counting right now are so terrible. It's just the capabilities are so large that we're getting outcomes, even though the energy and dollars and human time put in it far exceeds what's necessary.

Okay. With that, Sid, do you have any other thoughts on this? If not, we'll move on to the next topic No, I think in enterprise, it's really back to the enterprise.

Enterprise should measure cost per accepted outcome. It boils down to using AI for business outcomes is not a sort of discussion around what that one employee does versus the other employee. If I can use AI effectively for delivering an outcome that is dramatically different than what I'm used to, these things, I think they're focusing on the wrong metrics, these things, Accenture and the rest.

Right? $2,900 per employee versus what does it do to the bottom line of Accenture in terms of driving business is completely irrelevant, right? Yeah.

That's the thing about it. Yeah. That's a good way to end it.

Yeah. And we're going to move on to topic C, and it's interesting. So Anthropic on Sunday said it plans to allocate 5% of its total shares to establish a new public benefit fund ahead of its IPO, which could take place as early as next month.

This proposed fund would operate as a wholly-owned limited liability company designed to support large-scale philanthropic donations, direct disbursements to affected populations, and targeted initiatives mitigating the labor market disruptions caused by AI. Anthropic noted that while it already has committed $200 million towards studying the economic impacts of AI, these bolder interventions are becoming necessary given the tech's accelerating capabilities and its impact on society as a whole. Kate, what did you think about this gesture by Anthropic?

To me, so I have, I think two thoughts. One is when we're talking about 5% liquidity, it's not 5% of annual revenue or immediately- Mm-hmm ... available cash.

So the practical value depends on the share valuation, liquidity, and how the fund is structured, right? And so it seems like, oh, wow, this is really great. But, at the end of the day, we don't even understand how this is really going to impact us.

What is 5%? And so I know it seems really great, but for me, I think that how do we understand the people who are going to be affected by this transformation and then who decides what constitutes a public benefit of it? Who does this?

Is it the company, the shareholders, the communities whose lives and jobs are changing? There's a lot of questions, I think, around this. Totally agree.

There's so many-- Sorry, John, go ahead. No, go ahead. No, I totally echo what Kate just said.

From my perspective, the questions are multifold, right? Mm-hmm. When does funding become available?

How is distribution determined? Will the support remain dependable if the company valuation falls? And if I am giving it for some sort of a charitable institution, right, where does the money go?

It's not very clear. Yeah. Yeah.

So if I'm building a charitable institution relying on this source of funding, right, Anthropic's IPO is the big question right now. We don't even know when that's going to land, right? Yeah.

What does 5% of Anthropic share cap table even mean, right? Those are the things that I would ask. Communities facing disruption need assistance on a practical timetable.

You can't be sort of throw a number, 5%, right? Yeah. The other thing is around the whole credibility issue, right?

The fund's governance will determine its credibility. The proposed funds would operate through some sort of, allegedly, through some sort of a wholly owned LLC with further governance details that have not even been laid out or outlined or delineated. If I'm the recipient of this fund, for me to make a decision-making authority based on me as a beneficiary on this fund's resources are multifold, right?

It'll be conflicts of interest, right? Those are the several questions I would ask, right? And then public benefit funding should also complement, in my opinion, accountability, how well AI is developed and deployed.

Anthropic is one of the few companies that is talking about facing the frontier, so it's good in that sense. But what does it do to sort of this position around mitigating labor market disruptions? Financial contributions should be sitting alongside with concrete responsibilities for product safety, and all of those need to be well-articulated before I, as an institution, start to take advantage of something like this.

" And that's a very arbitrary comment. So at least Dario is doing something meaningful for once. Yeah.

I think it was several months ago, OpenAI talked about some sort of public benefits that it wanted to pass along. In a sense, to put my cynic's hat on, what Anthropic is doing, is this a ceremonial gesture, or maybe is it like a covert marketing move to put Anthropic and investors in the public's good graces? Look, so to be clear, I and Quietwire are a mouse to these folks at Elephant, and so our intentions can be what they are, and we'll see how it plays out, but I commend them for taking the stand.

I agree with everybody's comments. What does that specifically mean is what all my questions are. But I took a similar position when I started this company, that there should be a greater good to it other than just corporate welfare.

But at the same time, I've been involved with nonprofits funded by the corporate entities that I'm part of and volunteer efforts in standards development on behalf of the companies that I've been part of all my career, and there's always a related reason, right? And I will let the future judge me and my company in how we execute on this, but I think it's structural all the way down into how we build AI systems and information systems. I think it's always been part of how we build business and channels and relationships, right?

I don't think that our corporate culture has been good at being honest about that and capturing that. And you can see the difference when individuals and groups inside very large corporations really mean it and don't, right? So I'm trying to tie these things together, right?

So I say every single day on this show and every other place that we have to build our systems to include the rules, like literally embed the rules. Like not just read a standards document or a policy and then have a bunch of people try to... Like that policy document goes in the system, right?

And you have to negotiate every single step of every action a system takes and every company to company and company to government to government to government. It has to be written down. Otherwise, we can never have nice things.

So if this is just a gesture or something, then it should not work out. If they're serious, then it should. But it has to answer all of the questions, Kate, and Sydney, you bring up.

Exactly what does that mean, precisely, right? In the same way that I, as a vendor, give a software bill of material downstream to who, why, specifically, at every step. If you can't enunciate every step, then it's not real.

Yeah. Well, in a sense, I'm going to give Anthropic the benefit of the doubt because I do think they, in general, are trying to do the right thing. Maybe I'm just naive, but I actually do think that's their intention, and I'm wondering if their IPO, and from all indications, it's going to be super successful.

But I'm wondering, in this age of AI, could we expect future IPOs to kind of follow along the same lines and balance shareholder profit with societal risk mitigation? I mean, this is a model for maybe other companies to follow in the near term? I think- You've got to be careful.

If you've ever taken philosophy in college, you face these sort of issues, right? We want companies to be moral, but then we also want companies to define what moral is, right? Right.

And that's a circular problem. I believe that good business is moral, ethical, and transparent and so forth. I think it's better, faster, cheaper than hiding and lying and so forth, and lower risk.

Maybe I'm wrong, maybe I'm a Pollyanna, but I'm an old Pollyanna that, again, is unbelievable for the business based on the idea that it actually matters what your company and what you and what your team says, and the commitments you make, and the commitments you don't keep. And there's a lot of old-school sociology to this pragmatic, hard-nosed business world. Yeah.

A part of me is skeptical about this. Because why is an AI company doing this all of a sudden? Corporate America hasn't done this, and we see IPOs happen all the time.

So what is motivating them to do this? Obviously, there's a lot of backlash against AI ethics and AI responsibility and all of that. So that's one part of me sort of questions this whole motive.

But maybe I'm being too skeptical. On the other hand, I do know that, as I said earlier in the conversation, Dario Amodei is very concerned about the spread of AI and the ethics piece associated with it, and the philosophy piece associated with it. So maybe he means something here.

And if he does mean something, that's great. More companies should follow, not just AI outfits, but others too, that should be creating a fund for public benefit. Because again, it's not clear to me what the word public benefit means.

It's in capitals. I saw that in the article. So, what does public benefit mean?

Is public benefit going to be building some sort of a thing that becomes another corporate greed profitable entity, or is it going to be something honestly helping, maybe uplifting society? Those are the questions that one has to ask. So I think there are a lot of open questions.

There's a skeptic in me that questions it, but then I do understand how such a thing could benefit a society at large. So, yeah. And I come from the-- As I talk often on the show about the industrial revolutions.

So, during Carnegie and Mellon and all these... Hey, I grew up in Cleveland. Our art museum is free, and it's free because of the steel barons.

Mm-hmm. The libraries and the Rockefellers, they invested a lot. Although, in the '70s, of course, our rivers were catching on fire.

So , if you balance out what they put into the cleanup, I think we're looking at that same type of idea. 0. So what I would say is that if you're looking at that, that I think one of two things.

One is that who would-- If you were an NGO, if you were like this, or if you were a nonprofit group or something, and you were going to be getting this money from-- How do you build a business on getting money that you don't know how much money you're going to get? I don't understand the public benefit from that point of view. So it would be, for me, it would be almost one-time investments- Mm ...

across the board, so that it's not a continuous type of something that you could build a business on. And from that point of view, they talked about the study of impacts, and I think it would be even beneficial to really start to look at-- Because there was a friend of mine who sent me, he's a snarky Gen Xer, and he sent me something recently about all the tech bros, about what does the new job look like. " And it was funny, but there should really be a serious study on what does our future look like.

Because there are a lot of benefits. We're foolish to think that AI is not going to benefit us. And at the same time, it's foolish for us to think that it's not going to have an impact, a negative impact.

And knowing that, we should already be able to understand, or at least be able to hedge what do these benefits, negative and positive, look like, so that we can already start to prepare for both trajectories, the positive and the negative that every industrial revolution brings us. That's well put, Kate. I'm sorry, go ahead.

I was just going to say, it's many years past my libertarian era. But to be clear, all of us on the screen probably all agree that Dario Amodei and the Anthropic team mean well. But you can take other philosophical stances.

The whole libertarian sort of independent approach has, and Kate, you were talking about them. So I can pick on the robber baron era all day long, and we didn't live an alternate past, though. So all we have is that evidence.

If FDR had come earlier, if the robber baron era hadn't happened, maybe the world would be better, maybe it wouldn't. That's really, really hard to determine. So I'm not going to sit here and try to get inside the minds of the CEO at xAI.

But maybe, I understand the logic. Maybe the idea of running down the path as fast as possible ends up being a better net good. Maybe that's actually the intention.

That assigning negative attention to actors is complicated. In my experience, most people believe that whatever they're doing is actually the right thing to do. Mm-hmm.

In the, we'll say, left-handed world, we like to talk about greed and so forth. But wealth is a great thing. Poverty is awful, and done decently, there is rising boats and increasing human assets.

All right. Well, that's a good way to end it, Chris. That's a very nice optimistic way to end it, because I think you are right.

Fundamentally, you are right. People mean to do and intend to do the right thing. Hey, thank you for another assimilating Tuesday show.

You guys are great. It's the same Tuesday crew, Sid, Kate, and Chris. I'm Jon Swartz filling in for Mike and Alan.

I think they'll be back tomorrow. I know Mike will. But in any event, thank you for watching today's show, today's episode of Techstrong Gang, and we hope to see you tomorrow.

Have a great day, everyone.