“Open AI’s definition of AGI (artificial general intelligence): Highly autonomous systems that outperform humans at most economically valuable work.”
“Computer science and AI community are gripped by an ideological vision that places AGI, meaning machines, that exceed all human capabilities, as its highest possible pursuit.”
The statements above are from a research paper/essay by MIT economists for Brookings on developing pro-worker AI, that I have even shared with subscribers to The Whistle newsletter from my blog. The statements tell us the capability levels of AGI developed by humans in order that they can outperform us in every possible activity, especially at the most economically valuable work, and I shall return to them later in this piece. I have been writing about AI now and then on my blog, and my last blog post on the subject traced the evolution of AI and how it reached where it has. As I wrote then, the important turning point in the history of the development of AI, I think, is when humans decided to train and test machines to be better than us, after initially experimenting with whether they could think like us.
Now that the age of AI and AGI is truly upon us, and is said to be the biggest transformation mankind will see after industrialization and automation in the previous centuries, perhaps it’s worth looking at what kind of economy, life and world this is likely to create. I continue to be a skeptic of AI and do not see any value in generative AI that does our work for us or becomes assistants for every silly thing. One look at the Google AI mode advertising campaign that aired with the last IPL season in India will tell you what I mean! Nor do I agree with the ceding of autonomy to machines and agentic AI which will again take over our work, and AGI more generally speaking. Still, I think it is worthwhile imagining what an AI economy will look like. If nothing else, it might help us anticipate the future better and prepare for it as well as find ways to avoid obvious dangers and pitfalls.
Why AI, why now, and why the same AI models in the developed and developing worlds?
The first question to ponder over, I think, is why AI and why now? After all, we have had experimentation and research on AI since 1954, as I wrote in my last blog post on AI, which progressed in fits and starts through the decades and stalled along the way as well. It seems to me that tech companies, in their quest to innovate and race ahead of everyone – including governments and regulators – are doing so in order to compete for, and control more and more areas of our lives and work and make supernormal profits at the same time. In the advanced economies of the west as well as in China and Japan, this AI technology is the ultimate control over people, as well as a weapon in competing with each other. This is mainly because the demographics of these countries are changing rapidly, with fewer younger people in the active workforce as years go by. Of course, they could import workers from other parts of the world as they have done for decades, but with the new wave of right-wing anti-immigration policies being pursued across the developed world, this route is closed. Therefore, they have no other option but to pin all their hopes and aspirations on technology and AI as the great savior.
The next question to ask is, should the rest of the world be following this path of technological progress blindly and even if we were to adopt AI, should it be the same kind of generative AI and AGI? Here is where I disagree with the path that AI development has taken in America and across the developed economies. Where AI has been seen merely as a superior way of working to humans and can therefore replace us at work. I think AI developers missed an important stage which could have taken AI in a different and more beneficial direction: can machines perform tasks that we humans would like to, and need to, but are not capable of? The key difference is between asking machines to compete with us on the same sets of tasks and learn from us, and asking machines to perform a different set of tasks that we humans are still not capable of. It might have meant a completely different path for AI development and might not have relied as much on machines learning from us, but machines that we can program to perform superhuman or beyond-human tasks for us.
What if we had a different kind of AI?
In emerging and developing economies, if we can get AI to stretch our capabilities, by helping us accomplish tasks that seemed near-impossible, it would work as a multiplier of human capital and potential. It means that AI developers should not get bogged down with questions of whether machines or humans are superior but with what can machines help us achieve that we cannot accomplish ourselves. AI then becomes a human capability enhancer, not a superior or faster worker than us. If AI continues on the path that it is on, there is no doubt in my mind that businesses, governments and all organisations employing this form of AI will retrench severely. Already tech giants – the biggest users of AI themselves – are laying off people in their thousands since 2025 and this year as well.
Those in favour of AI argue that like in previous periods of industrialisation, computerisation and digital technology, AI will free humans to take up new jobs and careers. This argument doesn’t hold water as this particular wave of AI is like nothing that preceded it; AGI is capable of performing cognitive tasks, thinking, applying judgement and even making decisions on its own with little or no human intervention. Higher order tasks are already taken by agentic AI. As you can see from the two statements that I shared at the start of this piece, the aim of AI developers was always to make machines and AGI infinitely superior to humans. Comparison and competition were always built into the entire AI experiment right from the beginning. Instead, if AI were to perform tasks that are beyond the realm of human possibility and would be helpful to us, we would not only have AI help us stretch our own capabilities and accomplish the unimaginable, we wouldn’t have to cede autonomy to machines and humans would still be in charge.
AI’s impact on the economy
Returning to the kind of economy that AI would engender, many believe that AI is disinflationary. I suspect this observation is coming from the fact that AI has a definite downward effect on labour demand, even if this is yet to be proven empirically. The AI economy is necessarily one designed to work with fewer people producing more, which is precisely why advanced economies with ageing and shrinking workforces are in pursuit of it. And which is also why these countries foresee a huge jump in productivity thanks to AI. I think AI will change the labour-capital equation, skewing it heavily in favour of capital, by which I mean technology.
I happened to discover a book only now that an economist has actually written on the AI economy: The AI Economy: Work, Wealth and Welfare in the Robot Age by Roger Bootle published in 2019. I haven’t read the book, but managed to read a review of it in Forbes, a publication I usually do not read. From the review it appears that the author of the book, Roger Bootle, economist and founder of Capital Economics in London, doesn’t think that AI will cause technological unemployment – a term I take to mean that technology will not displace human workers – but that people in general will be paid such appallingly low wages/salaries that the state will have to intervene with welfare schemes such as Universal Basic Income. When I read this, I thought to myself that with fewer people working governments would earn even less revenue from taxes, so where is the possibility or room for governments to provide additional welfare schemes such as UBI.
However disinflationary AI might appear to be especially on wage-price inflation, there is another whole area in which the AI economy is going to impose huge costs on all of humanity. Besides creating worklessness and shrinking the demand for labour, AI will significantly expand the market for scarce natural resources and from the news we read we know that this scramble has already begun. Whether it is metals, fossil fuels, non-fossil fuels, critical minerals, chemicals, energy, water and the like, these are going to be required in gargantuan amounts in order to keep AI data centers running 24×7. And countries are busy competing for a share of all this, in order to fulfill their AI dream. Therefore, the AI economy is going to be one which will raise the costs of all natural resources and commodities to the detriment of our planet. In fact, a lot of this pressure on natural resources is likely to undo the good work done on climate change until now, and in reversing course the results could even be apocalyptic for humanity. Elon Musk has an idea: put data centres in space! How on earth they will operate there is something he has yet to figure out.
AI’s effect on labour and on work
If the AI economy is essentially a 24×7 economy even more than the digital economy is, what will life for us humans be like? Earlier waves of industrialisation and automation provided us with more leisure, which is not present to the same extent in the digital age since technology has blurred the lines between work and leisure. The AI economy will probably offer less leisure in continuation of the digital economy, but more importantly, millions out of work will have all the leisure they could possibly want. Unfortunately, unlike previous periods of industrialization and automation, this time there won’t be any gains to disposable income, except at the highest income levels. Leisure needs to go hand in hand with rising incomes so that it also translates into more and better consumption and improvement in living standards.
If AI is mainly a labour-saving, speed-enhancing technology, we must also accept certain caveats to its ability to displace jobs on a significant scale. There could be a huge difference in its impact at the firm level and at an aggregate economy-wide level. At the firm level, it is certain to displace and replace many jobs across functions, from the cognitive and managerial to the routine, repetitive and mechanical at the factory. And it is possible that many of the new jobs it may create will be at an aggregate level, but I don’t think anyone still knows what these might be. That said, there is no question in my mind that AI will tend to depress the value of work, whether it replaces employees or not. This is because automation and mechanisation tend to commoditise tasks and work, and therefore, the output. In previous waves of automation, firms replaced some of their factory workers with technology, and engaged the rest as contract workers, depriving them of annual benefits that accrue to regular and permanent workers on their payroll.
I said this in my blog post on a new strategy for WPP and the advertising industry that relying too much on AI will only commoditise work between agencies and between people and perhaps increase the in-housing trend among clients. I can also see some of the job displacement scenario playing out at the large advertising agencies and not all of it is due to mergers. I think WPP alone laid off around 7,000 employees during its restructuring and this might be a conservative figure, but when I see posts on LinkedIn about their new hires and leadership changes, they are all at the CXO level. This highlights one of the problems with AI: it eliminates the need for middle-management and perhaps even entry-level workers, and has only a few opportunities for senior and top management folks.
Now, let us zoom out and look at the economy-wide picture. If most people at junior and middle-management levels in companies are going to lose their jobs to AI, it means that the middle-class is the most threatened at an aggregate level. Economies cannot afford to lose their middle-classes as these comprise the bulk of consumers. Even if they don’t all lose their jobs, their wages and salaries are going to be driven down drastically because their work will hold no value in an AI economy. Perhaps this is why Roger Bootle has written about the state having to intervene with UBI and other welfare schemes in his book on the AI economy.
The death of the knowledge worker
Because of its ability to depress the value of work and commoditise it, while also displacing jobs at managerial levels, AGI might be the end of the knowledge worker that Peter Drucker hailed as the greatest thing to happen to industry and the economy decades ago. Because it can perform cognitive tasks, manage workflows and even make decisions without requiring human intervention, AGI will force millions of people to consider alternative employment, including skilled manual labour. I am not being facetious; there are many who already say with a straight face that the jobs most in demand in the future will be those of plumbers, electricians, service technicians, mechanics, hair-dressers, tailors and the like.
Besides, because AGI allows people without any specialized knowledge or experience in an industry to suddenly become adept at certain tasks because AGI is doing the work for you, champions of AI also talk glowingly of its democratisation effects! Think of unprofessional PR agency idiot bosses in India, for example, who will overnight become experts and “pros” at advertising and marketing without knowing very much about the work and the industry; they have been meddling and trying to muscle their way into the industry anyway, when they should be booted out and barred from ever doing business again. Therefore, the corollary to the death of the knowledge worker in the age of AGI could well be the rise of the ignoramus.
The death of knowledge itself
Next, let us think of what AI will do to knowledge and education. And I don’t mean only the use of AI in classroom teaching and its misuse or abuse in examinations and the like, that everybody seems to be focusing on.
I would like us to think about what the need for education itself will be in the age of AGI. When no particular domain knowledge is required, nor any particular skill or proficiency, why would people go to school and university and what would they learn there that AI does not already know and cannot already do? In my opinion, if AI and AGI are allowed to continue to go down the road that they are already taking, it would not merely depress the value of work and wages, it would lower the value of knowledge itself and therefore, of formal education.
The other caveat and option to AI displacing jobs, is that it could force millions of people to become entrepreneurs. In an economy, where few jobs will be available since the demand for labour is going to shrink, what else are people to do but try their hand at starting a business of their own. Surely, not everyone can be an entrepreneur and I certainly know that I am not cut out for it, but millions might be forced into this as the only livelihood option. Another work option is the gig economy comprising delivery workers, car-hailing drivers and the like. Meanwhile, large companies using AGI will become larger and more powerful and the rest of the business world will comprise minions who will struggle to survive. There will be little competition to speak of, and we have already seen increasing concentration of market power in the first two decades of this millennium, especially in the tech sector. This trend is likely to grow, unless anti-trust regulators and governments get their act together to promote greater and fairer competition.
AI and widening inequality
This brings me to the last and most important impact that AGI could leave on generations of people and on economies. Widening income and wealth inequality. As a technology that discourages hiring and works at a much faster speed than any human is capable of, AI will deliver higher productivity of the kind that developed economies are hoping for. And it will also deliver greater returns to capital than to labour, making it the most unequalising force in economic history. AI such as gen AI or AGI will, therefore, not only take away jobs but will reduce salaries and wages as well without creating an equal number of equally well-paying jobs elsewhere to make up for the disruption. This is how this wave of AI technology is different from all previous automation and computerization changes that have taken place before.
If all this sounds dystopian, it is because it is so. Any technology that competes with humans and replaces them for years to come is going to be highly unpopular with people and with the working population, especially the young. Which is why businesses and governments are going to have to take the responsibility to upskill and train people, on an ongoing basis. And governments have to also prepare for a world where they have to provide millions of people dole and welfare schemes to help them make ends meet. This is not a choice that governments will have; they will have to do both.
Much better, in my view, to have AI that is a human skill enhancer and can stretch human capability to achieve the unimaginable. A more restrained and friendlier AI that is sensitive to human needs and not harmful to the planet either.
Not simply AGI, but MAGIC, is what the world needs from technology.
The featured image at the start of this post is by Kindelmedia on Pexels
Post script: I must say that I think that a lot of this AI hype and mania is being fuelled by unprofessional PR agency idiot bosses and their cronies in India who have wrecked my career in the advertising and brand communications industry – which has nothing to do with technology or chips or with data – in order to cover up all their unprofessional conduct over the past two decades. This is now yet another of their global circuses – besides their political shenanigans and warmongering – and I just hope the AI mania doesn’t turn out to be the crash that could cause yet another global financial crisis and recession, because a lot of equity markets and private credit are reported to be tied to it. I do not wish to be dragged into any of this nonsense that has nothing to do with my work and career, thank you.
Peripatetic Perch 