Comments on AI, financial bubbles, and career orientation

2026-09-16

The following are comments I made in a recent private talk. I am writing a summary here since the topic is of interest to a wider audience.

The technology case versus the business case

Machine learning, large language models, and agentic computing have reached a point where the technology case has strongly been made in favour of Artificial Intelligence.

AI is here to stay because it has already been proven as useful. From the technologist’s perspective, its flaws can be addressed, its shortcomings ameliorated, and its output refined. Whether it can become truly intelligent or sentient is beside the point: it has utility and will continue to do so, whatever the legal-institutional arrangements around it.

This does not mean that there is as clear of a business case in favour of AI. By this, I mean that we are still not sure if every business that relies on it has actually witnessed productivity gains that corresponds to some tangibly superior product or service. Maybe those exist, but the results as not as clear-cut as with the technology case.

Financial bubbles do not avert a paradigm shift

There is this notion that the AI bubble will eventually burst. What is implicit in that view is that things will then return back to normal, such as to how they were pre-COVID.

Reverting back to a supposed baseline makes no sense, given that the technology case has been made. What a potential financial crisis will do is realign capital to match the then better understood realities of the technology.

The notional bubble can be understood as a cascading effect of hyperbole on top of hyperbole. Every investor involved, no matter how clueless they are about AI’s actual capabilities, is putting money into what effectively is a promise.

Where there are promises, there are big words and lofty claims that cannot be proven wrong in time 1 but are shown to be unrealistic in time 2. The meantime is the present situation where seemingly everyone is racing to make a quick buck out of what is going on.

A potential crisis, then, will wipe out the accumulated malinvestments by exposing them for what they are. Those who are actually making something viable out of AI will remain in business. The rest will be gone.

This was the case with the dot-com bubble: technology kept advancing, but much of the hype around certain products disappeared. The financial downturn did not avert the rise of the Internet, nor did it prevent the transition to a largely digitalised world with just-in-time delivery of goods from online platforms.

Same idea for the 2008+ economic crisis: it exposed malinvestments on a colossal scale. For example, investors realised that Greek sovereign debt cannot have roughly the same in terms of credibility as its German equivalent because the shared currency of the two nations (the euro) does not, in and of itself, establish an optimal currency area; an area that the European Union still lacks.

As such, financial crises are moments of truth. They do not stop the paradigm shift that technology brings. All they do is clarify what is and what is not worth investing in. It is the economy recalibrating itself.

The problem with finance, and why the economy goes from one crisis to the next with periods of exuberance in-between, is that the corrective mechanism is not exact science.

What matters for our purposes here is that AI will stick around long-term as the cornerstone of present and future industrial activity.

Arms race between the AI superpowers

The United States of America and China are the world’s two superpowers in this new frontier. Both have their structural flaws and competitive advantages, so we cannot tell who will come out on top from the competition (notwithstanding an international order already in flux).

What matters is that in addition to the aforementioned technology case, we are also dealing with a compelling geopolotical argument which can be summed up as “whoever is better at AI is better overall”. This is due to the centrality of the technology in the emerging capital structures.

Whether AI is regulated or not, and what the modalities may be, will depend on this power play. It cannot be reduced to a purely national consideration because whichever country hits on the breaks first runs the risk of falling behind in the race.

To this end, we might experience the emergence of a new MAD doctrine where the two superpowers understand that continued acceleration only contributes to their mutual assured destruction. On the flip-side, both governments will consider it necessary to support and/or pick their national AI champions, in pursuit of wider geopolitical ends. Again, the technology is not going anywhere.

I do, nonetheless, think that these are still early days on the regulatory front and whatever fear mongering is spread by those with vested interest in AI companies has business motives rather than deep-seated humanitarian concerns. Even mentioning humanitarianism in the context of ruthless moneymen sounds preposterous.

Against this backdrop, other considerations such as the energy consumption of AI data centres, matters of copyright infringement, and ownership of AI output, will simply be subordinated to the logic of power politics.

Career orientation for knowledge workers

From the perspective of economic organisation, AI essentially is skilled workforce as a service. The providers offer tools which do the work that was once handled by a team of people. As the technology improves so will the demand for skilled labour change to more specialised areas of expertise.

Right now the junior programmer is a dying species. Whatever role that person was performing can now be handled by an AI agent or several such agents which are orchestrated by a single senior software engineer.

As such, the pipeline from secondary to higher education will be disrupted. A college degree used to be one’s ticket to a stable job or, at the very least, a valuable asset in their favour. The investment in time and treasure was thus considered worthwhile. Now this is no longer a given. A teenager today will not necessarily get a programming job at the end of several years of extra schooling, private tutoring, and further formal studies for roughly another decade of their life.

The business model of the modern university makes no sense anymore the way it did in the recent past. It costs too much and is an investment with no reliable outcomes. We have already seen hints of this inefficacy in highly saturated fields such as the academia, where getting a decent job is virtually impossible. I expect the generalisation of the trend to encompass any endeavour that can be described as white-collar or “knowledge work”.

In other words, AI will stay, it will compete for knowledge work with everyone and probably come out on top, and the major political players have no powerful incentive to stop this phenomenon.

Studying will always be worthwhile as an intellectual pursuit. The milieu of a university will still be a place for networking and dating. But the current business model of higher education will need to be redone.

There can still be new opportunities for employment and entrepreneurship for those who are willing to venture off the beaten path. What matters is that there are no guarantees anywhere.

What must a person do

My thinking for a few years now is that a person’s resilience is a function of their community’s coherence. By sharing resources and showing solidarity to each other, people in communities can make more with less. Communities are undone when their members starting behaving along the lines of “me, me, me”, and, conversely, they grow stronger when all work centres around the “we”.

Those who are individualistic, either in principle or by circumstance, will have to shift their mindset and attendant patterns of behaviour towards collectivism. Only then can they create or take opportunities to connect with others in networks of collaboration and shared living.

Put differently, there is a necessity of retraining oneself to remain relevant in the AI age. It involves technical acumen that will have to be continuously updated as well as strong social skills. One can no longer do what used to be normal in many sectors, namely, to act as the functional equivalent of a good robot.

Communities with their collectivist ethos also create the basis for political thinking along localist and socialist lines. Wealth is being concentrated in the hands of ever-fewer people. The middle class is disintegrating as many knowledge workers are being forced to take on menial jobs they would otherwise not even consider. By having the culture of sharing in mind and by setting aside their individualistic fancies, people are better prepared to campaign for policies that, at the very least, have a stronger pro-social character or are, anyhow, not supportive of those who already control virtually all of the planet’s wealth.

The personal story

I am directly affected by what is happening. My coaching work will probably not be viable in the near future. A few months ago I cut my hourly rate in half. It did not boost my coaching hours. I kept doing roughly the same number of meetings for half the income.

Additionally, AI has removed all the easy tasks, so I must only accept to tackle the most demanding problems. Those I would normally charge extra for, but I do not have that option anymore.

I have been doing other working-class jobs on the side and may need to switch full-time to one of them. My contributions to free software have already been reduced, even if they are still sizeable, because more work away from home means less time for my computer-related interests.

Furthermore, I am trying to the best of my abilities to strengthen my social network. My goal is to have people around for the long-term that I can help and be helped by.

I remain as confident as ever because I know I can adapt and enjoy the challenge. Though I recognise that these are extremely tough times. Those who have hitherto lived in relative comfort are in for a rude awakening.