I could only assume that this is a world without money, really, where money is just symbolic. There is this question about the timeline for me. Developments in robotics right now, they're not as impactful. Robots are not as skilled as I think Elon Musk would wish them to be. And they're definitely not replacing the majority of work that humans are performing today. The impact on productivity is there, but modest. With the speed at which AI is progressing, how close are we to losing our jobs? Actually, this is the theme of our discussion. And what's next with our productivity with Alexandra Shagalinska. She is the Associate Research Associate at Harvard Law School and a futurist and expert on artificial intelligence. Welcome back to the show, Alexandra. Good to see you again. Thank you so much for the invitation. Happy to be back. Take a look at what Elon Musk wrote on X just a couple of weeks ago, a couple of days ago, universal high income via checks issued by the federal government is the best way to deal with unemployment caused by ai ai robotics will produce goods and services far in excess of the increase in the money supply so there would not be inflation okay a lot of claims and assumptions he's making in this statement the notion that ubi will be given now because we're all going to be laid off by ai this is not a new idea how serious is this claim now well what is quite interesting also is that here we're not dealing anymore with UBI, so universal basic income. We're actually dealing with high income. This is something that Elon Musk is proposing, and that I think is fairly new. We have heard about UBI, and actually we've been hearing about it for the past 10 years or so, because of the looming massive unemployment caused by AI and automation. And I think it's been one of the most favored responses to that threat by the Silicon Valley. So whenever you heard, oh, okay, AI is going to take over blue-collar work or white-collar work, the most, I would say, frequent response was usually, well, let's do universal basic income, which, by the way, is a policy that's been discussed for at least 300 years now and never introduced, never implemented, just maybe except for a couple of small pilot projects like the one from Finland which took two years and was in a way not really UBI because people were given money throughout the time and they could spend it on whatever they wanted they decided to educate themselves with that money and also spend it on subsistence but that was just a very short pilot project so it's not really telling you about how people would behave in the long run if they received such cash payments So I would say we are talking here, even when we just focus on UBI, on a policy that's never been tested in the real life. It's just a hypothesis. We do have cash payments in different forms in many different countries that are supportive. They're not your main source of income. And we do have universal basic services in countries like Scandinavia, for instance. We have something that we could maybe call a universal basic asset. That would be something that Singapore is currently trying to introduce. So they're giving everybody a voucher to spend on AI related education so that they can upskill themselves. So you have a couple of different variations, but these are usually just a supplement to what we usually make in our real life. So UBI is cryptic enough, but Elon decides to go a bit further and he says, well, UBI is not enough. And this is actually something I agree with, because UBI, if you talk about a claim by, I think, Sam Altman three years ago, he mentioned that OpenAI would distribute among all the US citizens $3,000 per month. OK, that's a great bonus, I would say, but that's definitely not something that can support you on a monthly basis. So Elon Musk here says, well, that type of UBI is certainly not enough. We have to go for universal high income. And only then we can talk about a situation where that scenario of us really not working at all and choosing what we want to do with our lives is fulfilled. Right. So that's sort of the background story here. But I see sort of a pushback from many economists already because he came up with a fiscal solution to a problem that is mostly economic and societal. and he kind of talks about the government really spending that money, whereas the whole, I would say, digital dividend from AI is now in the hands of companies. So, well, I guess the government would have to tax all these companies heavily and then try and distribute that high income amongst the citizens, which in a way, frankly speaking, is a bit of a socialist idea, unlikely for libertarians. He also claimed that AI robotics will produce goods and services far in excess of the increase in the money supply. So there will be no inflation. That was my first question because somebody actually commented on this. If everybody has high income, then nobody has high income because there will be inflation. But now he's suggesting AI will fix that problem because the output will be so much greater. Can you just walk us through the thought process here? our what is yeah yeah how will we get from where we are now to ai producing goods far in excess of the growth of the money supply well i frankly speaking cannot imagine that either i could only assume that this is a world without money really where money is just symbolic in a way because if you really take it that far what i understand he means is that uh ai robotics and automation creates such abundance of goods and services that we just don't have to use money at all, which is something he, by the way, mentioned in another interview. The other thing is, you know, there's this question about the timeline for me. He talks usually about timelines that are fairly quick, happening fast, you know, in the next couple of years or so. And this is another question mark because frankly speaking, even when you look at the developments in robotics right now, they're not as impactful. Obviously, robotics is developing. It's a trending topic these days, but robots are not as skilled as I think Elon Musk would wish them to be. And they're definitely not replacing the majority of work that humans are performing today. So I just don't feel that this is a realistic timeline. And also the question of productivity is a big question. We've been using Gen.ai and Agentec.ai recently, you know, and the impact on productivity is there, but it's modest. It's around 16, 20 percent of a spike of productivity. So it not as much as he would wish I guess to create that type of abundance that he would want to see Right So I just feel that he throwing in another concept for other people to worry about It's just a way to mediate the fears surrounding massive unemployment. And I sort of understand why he does it, because in all honesty, other tech players are talking a lot about how people will not be needed at what they're doing right now, that AI will be far better at knowledge work, analytical work, research work, creative work as well. And they're not offering any alternatives. So they're not really saying what's going to happen. They just said, well, this is what it's going to be like. And then the economic part is something that others should take care of or worry about. And since nobody else is addressing it, and I wouldn't say that the government is addressing it by creating some sort of guardrails or ideas around the future of employment and then the job market. I guess Elon decided to step in with his highly utopian, beautiful view of luxury unemployment. than you actually think. That's why I recommend today's sponsor, Delete Me. It takes a few minutes to set up and they handle the entire process. They find where your information appears, verify those listings and submit removal requests to hundreds of data broker websites. They also continue monitoring over time since this data can resurface and reappear. I've been using it for over a year and they have reviewed over 325 listings for my information. The reports show exactly what was found and what was removed, which makes the process transparent. Check it out now. Scan the QR code on the screen or go to joindeliteme.com slash DavidLynn. Link in the description down below and use my code DavidLynn for 20% off. Take control of your privacy today. Now back to the video. Okay, before we talk about everybody being laid off, let's talk about layoffs happening in the tech sector first. So this is a news item that came in in the last couple of days. 20,000 job cuts across Meta, Microsoft, raise concerns that AI-driven labor crisis is here. So Meta is announcing that they're cutting 10% of the workforce. Microsoft is announcing a voluntary retirement plan for up to 7% of the workforce, mostly for senior level directors. But these layoffs are concerning for a lot of investors. More than 20,000 potential job cuts across Meta, Microsoft. and this comes after Amazon announced its most widespread layoffs ever. And Nike is also laying people off, but I don't think that's part of the same group. But anyway, this announcement is troubling. These companies are blaming AI and automation, saying they're getting leaner. Other people say, well, it's because they overhired during COVID and this is an excuse. How are you interpreting it? I think both things can be true at the same time. So one thing is really that during COVID, there has been that process of overhiring. And since COVID ended, we see how many teams are just being cut short. Metaverse, I think, project by Meta is one of the best examples, one of the most prominent examples where you actually see how, you know, it just didn't verify itself as a project that is viable. and therefore the whole team got released. They did not decide to transfer that team to new tasks. But the other thing that I think is the part of it that needs to be addressed is that for the past four months, we've seen rapid growth in the so-called agentic AI. And that is much more than just simple chat GPT or our usual experience with Anthropics, Claude or any other model. It's actually a new development in artificial intelligence that allows to create agents that are going to fulfill certain tasks end to end, right? So they can make payments, they can run different loops, they can query databases, come up with results, send emails, write reports. So they can do many different things that people have been doing and it's particularly visible in IT. So wherever you have the creation of software, these agents that are actually forming certain tribes or, if you will, communities, even societies, right? They have their own hierarchies. So you have an agent that is an orchestrator and that orchestrator is orchestrating the work of other smaller agents that are doing other things, reporting back. It resembles a big organization or even a society. And it's an incredibly powerful tool. So one coder can now use those tribes of different agents to perform multiple tasks. And here, actually, I think it's the only example where productivity spike is really impressive. So in IT, as opposed to other sectors where AI has been implemented, it's spectacular and I would say significant because the spike in productivity is, I think, incredible because of those agents mostly. And that's been happening for the past four months. And I think with that, you can clearly see that you can have four or five IT specialists who are doing the same job as 50 people have been doing before. And therefore, that remaining 45 people are just not needed anymore. You don't need them. You just need those five skilled people that can work with AI agents very, very effectively. So I think in the tech sector specifically, this change that we're observing, this shift towards like hiring freeze or just simply firing people is related to both of these phenomena, but mostly, I think now to agentic AI. So here is the source, layoffs.fyi, that actually CNBC was quoting 92,000 tech employees laid off in 2026. So far, there's a list of companies and how many people have been laid off so far. Now, you commented on why these layoffs are happening within the tech sector. So who in the tech sector is safe? Do we still need coders, by the way, in the future? I think we need orchestrators. So human orchestrators that will orchestrate the work of AI orchestrators. I'm not sure if that really makes sense, but that's the truth. So that would be a supervisor of the work of AI agents. And that person is obviously needed. That person will have that goal and the intention to do something. And they will, to a large extent, still supervise the work of those AI agents. And they will be responsible for that work. So somebody has to really review what those agents have done. They're obviously not flawless. They do make mistakes still. And there has to be some human quality assurance. So I would say that there are jobs in let say this area specifically that would be a meta type of orchestration or if you will supervision of artificial intelligence But coding per se by hand I don think is a likely thing to happen in the future I would still say people will try and learn it just to understand what AI is doing just to have a basic understanding of what's happening in that black box. And it's also obviously very good for us to learn how to code because its structures are thinking. But nonetheless, I think human coders in the future will become obsolete. I wonder how a computer science curriculum at, let's say, Harvard or MIT is going to change in 10 years. Yeah, that's a big question that everybody's asking, believe me. Also at Harvard and MIT, I would know. But I think what these big, well, academic institutions are trying to understand better is how to strike a good balance between still learning people those necessary skills and also the advancements in AI that make even students question the sole fact of having to learn to code. Why would they, right? Is that really necessary? I speak to my students as well and quite often they're a bit frustrated that they have to learn how to code. But in all honesty, if you really want to work in big tech still and if you want to develop frontier AI models, which will be a thing that people will be doing for the next years. And I think there will be big investments in that. Or if you want to develop robotics or physical AI, obviously you do need that skill set. It's just the amount of people who are going to work there will not be as big as it is right now. And I think we will just simply need less people, but they will have to be highly skilled. So yeah, I think just the institutions like Harvard are trying to figure out what's the best way to retain the good aspects of the curriculum in computer science that prepares people to really work with AI on that abstract level and come up with new AIs as well. But then also just teach them how to use AI responsibly and effectively as they are, for instance, by coding some new solutions, new tools, and working with those agents in different dynamics. it's not also computer science courses like i can imagine other courses in other departments like working with excel for example becoming obsolete if a spreadsheet absolutely absolutely all of it all of it um you know ux um creating apps um and everything that is i would say some sort of mix of tech and creativity that can be done with ai very effectively and i'm sure people see them and see that and experience it on a daily basis because they can create, let's say, a beautiful PowerPoint presentation in no time with a tool like Claude. It will take them like five minutes and it's going to be beautifully structured. And I think this is just exactly the type of experience that IT has, but just on an even larger scale. Speaking of Claude, so according to Reuters, we have Claude and OpenAI battling it out. the revenue gap is shrinking. So at the start of 2025, OpenAI led Anthropic in annualized revenue by $6 billion versus $1 billion. Now, by early 2026, that gap high widened to $20 billion for OpenAI versus Anthropic's $9 billion. So you can see that even though the absolute value has widened in percentage terms, Anthropic is catching up. What is Claw doing that is taking revenue away from other AI competitors like OpenAI? I think that the most interesting aspect of their current activity, which is also the one that's most threatening to the job market, by the way, is creating very specialized models. So Anthropic has decided to build different, let's say, tech stacks for the financial services, different ones for legal services, and yet different ones for research. So they're not just offering a fairly generic tool that is good at everything, but on a surface level. And I would say ChatGPT, to a large extent, has been something like that. It's that type of, you know, encyclopedia that you use, but it has quite often shallow knowledge of various things. And it doesn't have very specific use cases. It can be used by anyone for anything. And that's just like, you know, a bit blurry. And I think Entropic here has decided to place bets elsewhere and they want to establish cooperation in the B2B area. So they want to service business. They don't want to service individual customers. And I think they've been developing that branch very, very dynamically over the course of this past year. And because of that, they're very successful in B2B and they're trusted as well. I would say even more so after the whole situation with the Pentagon, you know, obviously some users decided to follow them because of certain values or principles. But also, you know, they seem to be a company with a particular trajectory, mission, vision. And I think to some businesses and organizations, that's also persuasive. OpenAI, I think, you know, is really trying hard to come up with new products that are also specialized. Now they have released a new image model and they want to replace Canva or Figma. That's clearly the ambition. So we don't know what's going to happen next, but certainly they have not been doing a great job at B2B and Anthropic has. Claude Mythos is launching. It's highly advanced, specialized AI. And walk us through, please, what it actually does. Supposedly, it exploits zero-day software vulnerabilities. What does that mean? and is the entire software index at risk. Just for reference sake, the IGV, which is the iShares software ETF, that's down about 21% from the beginning of the year to now, year to date. What is cloth mythos going to do to the rest of the sector? We honestly don't know quite yet because the project itself is quite cryptic and it's very hard to gain access to it. This is on purpose. Anthropic is trying to protect its model. But what we know is that it's a frontier AI model that sits above Cloud Opus, which is their main model, that is very good at spotting certain bugs in software. And we do know that Mythos has been released to Microsoft and a couple of other giant players just to show how good it is at spotting bugs in their core systems, which is obviously a major vulnerability for such important companies. And it's done an amazing job at it. So we can clearly see how it attracted a lot of attention also from potentially malicious players that would like to use it, for instance, to disrupt other types of services and disrupt currently fairly secure banking systems, for instance. So getting something like Mythos is obviously for a malicious player adversarial player of any kind a great gift And in that sense there are plenty of concerns around it um but yeah the information is scarce and i think it scarce on purpose what we do know is that apparently it's incredibly good and uh anthropic is currently releasing it only to vetted um organizations and and governments as well so yeah we'll see who would benefit from this tool I think governments could potentially benefit from this tool in ways that I'm not sure if I want to imagine. But mostly really, if it's in the hands of organizations that just want to protect themselves or protect their customers' data or protect their customers' assets, this is clearly a great shield against various types of adversarial attacks. And notice there are many attacks currently. There are attacks on various cryptocurrency platforms. There are attacks on OpenAI. So these are very much intensified. They have intensified over the past couple of months. And I think Claw is just a tool that exposes those vulnerabilities. So like I said, it can be a protective shield, but it can also be a very, very massive weapon. Okay. I want to show you this clip. uh before we go let's talk about humanoid robots so at china's um inaugural robot actually it was inaugural i think they did it before but humanoid versus human robot uh sorry humanoid robot versus human marathon the robot won okay this is what this is what the robot looks like uh just incredible event and i i i i don't know if all the runners were just slow i don't think so but um they did this last year and the robot time for this year tripled yeah and it's not just impressive in the sense that it finished running uh but you know it speaks to the battery it speaks to autonomous systems it speaks to like yeah cooling systems so huge advancements in all aspects humanoid robotics is a part of the five-year plan from the chinese government where is this headed well china clearly is very much interested in physical artificial intelligence and robotics. They are very clear about it. They are not so much interested in Gen AI and I would say even agentic AI. But their claim is that they want to build artificial intelligence that's going to help them in changing their physical infrastructure, right? Changing their cities, building new cities, building bridges, etc. So they're trying to build systems that are ergonomic, that are going to work well, and also will have a good battery life. And I think they're very much succeeding at it. I would also add that their society seems to be very optimistic about robotics. Maybe it's part of their educational process and how robotics is portrayed. But certainly I don't experience or I would not experience as much enthusiasm towards, you know, a physical robot racing humans in Western civilization. I think many people would feel either offended or they would just question the whole situation where the robot had a separate lane and then humans are running on their own lane. What's the purpose of such a race even? And I don't think we're so well adjusted to having real life interactions with these systems. They often feel scary to us. But somehow in China, you have this growing positive sentiment towards robotics. Everybody wants to build them. Everybody wants to own a robot at home with the hope that it's going to support in daily chores, etc. So I feel like there is a narrative that goes hand in hand with the technological progress. And China is definitely betting on physical AI. And maybe they're also hoping that, you know, certain limitations that we've experienced with language models, because language models are out there in the virtual space. They have no physical intuitions. They don't know real life. They don't have experiences of our world that has its own limitations. But I think what China hopes for is that if they develop robots that have this good perceptive system and are able to be around us in the physical world with so many barriers, with so many different hurdles for them, then that means that they're going to reach that next threshold of AI's development per se. So I think this ambition is really speaking to that effort. Final question. I was reading, actually watching a video. There was a company in the UAE that was developing a jet engine and it did it on its own using AI. So AI designed the engine and then manufactured it itself on a span of two weeks. How close are we to AI designing its own products, manufacturing it using automated factories with robots, and then selling it via its own e-commerce distribution models that it deems most efficient, basically running an entire production line and company from A to Z. Well, I think whenever we're dealing with something that is highly formalized, where you have a process that needs to be streamlined in a particular way, and there is an order of things that are clear and not too fuzzy, artificial intelligence can definitely step in. And currently, we can bridge certain gaps that we had before. So we can have a robot. But inside that robot, there will be a generative AI system that will be able to retrieve information from the outside world, react to them. So all of that can be sort of, I would say, built together. And I would absolutely see it as an option that's happening. you know the sales aspect can be done currently with agentic artificial intelligence we know it we're not sure whether that would be that perfect sales strategy but certainly a good one and then you know autonomous work in factories is something that China has been doing you know a lot of in the last years think about the so-called dark factories right where you don't even need light you just have those robots working all the time without any light needed because light is for humans and there are no humans inside. So I think, you know, there is that attempt right now to bridge these two things, that physical component with that capability of generative AI and agents. And when that is actually colliding, you might see, well, incredible change, I think. So it's not unlikely for me, and it doesn't seem implausible for something like that to happen in the next years. Alexandra, good to see you again. Welcome back to the show. Where can we follow your work? I invite you all to my LinkedIn, I guess, or Twitter if you want to. We can leave the handles, I guess, somewhere. So happy to continue this conversation. Absolutely. We'll speak again soon. So please do follow Alexandra in the links down below. And we'll have her back again to address any other topics. Feel free to write to me if you have any topics that you want to ask Alexandra next time. Take care for now. Thank you. Thank you. Bye-bye. Thank you for watching. Don't forget to like and subscribe.
