I Have Existential Questions about AI. And I Used AI to Answer Them
I spent an afternoon asking an AI the questions about AI that have been sitting in my brain for two years. Here's what it answered — and what I still don't buy.
I spent an afternoon asking an AI the questions that have been sitting in my brain for the past two years.
The questions:
If AI does the cognitive work, where do the humans go?
How many consultants can one economy actually absorb before the market collapses under its own weight?
If people are losing their jobs and their income, who is left to buy what the creators are selling?
Is UBI actually a solution, or is it just what people say when they do not have one?
Only 26% of working-age adults are using LLMs right now. What happens when the other 74% arrive and the social infrastructure still is not ready?
Would you call me pessimistic?
I run a community of over 6,000 talent professionals (started before the word "community" meant basically "yet another buzzword"). And a B2B2C consulting practice in the same domain. This means I have a front-row seat to what happens when companies decide humans are optional.
The AI gave me genuinely good answers. Which is either reassuring or deeply ironic, depending on how you look at it. Probably both 🤷🏻♀️
I am publishing this conversation because I think that asking these questions matter more than any answer we currently read online.
And because I am tired of the content that exists on this topic, which tends to fall into two camps: optimism dressed up as thought leadership, OR doom-scrolling nihilism that does not go anywhere useful.
I want something in between. Something a bit more raw and honest. I am known for saying out loud things that people think but for some reason keep quiet.
Fair warning: this article was written with AI assistance (mainly on the research part).
The "critical thinking" question — If AI does the cognitive work, where do the humans go?
Companies like Block are replacing people with AI.
That is happening now, not in some speculative future. And the conversation around it keeps defaulting to the same historical comfort blanket: the Industrial Revolution. Farm workers became factory workers. Factory workers became service workers. Each wave of automation created more jobs than it destroyed. It will all work out. (#eyeroll)
That argument does not hold this time.
Every previous wave of automation replaced physical labour and created demand for cognitive labour.
You lost your job on the factory floor and retrained to do something that required a human brain. The implicit assumption was always that human thinking was the ceiling, the irreplaceable layer.
LLMs automate cognitive labour.
There is no next layer to retreat to. When the machine can write the brief, analyse the data, screen the CVs, and draft the contract, what exactly are we retraining people to do? And for how long before that role is also automated?
The retraining answer has always had a dirty secret: it works best for people who were already going to be fine.
Younger, already educated, geographically mobile.
The people who successfully retrain after industrial disruption are disproportionately the people who had options to begin with.
The evidence from manufacturing displacement programmes in the UK and the US is genuinely grim. Success rates hover somewhere between embarrassing and tragic.
The consultant question — How many consultants can one economy actually absorb before the market for consultants collapses under its own weight?
A lot of people are stepping away from corporate life. Some are pushed. Some are so exhausted by the politics, the shareholder-value theatre, and the gap between what organisations say and what they do, that they simply leave.
They become consultants. Fractional this, independent that.
Like me.
I see this often in my environment. Talented people, genuinely good at their jobs, deciding that working for themselves is the only way to stay sane.
But I find myself asking: how many consultants can the economy actually absorb?
Because consultants need clients. Clients need budgets. Budgets come from organisations that are generating surplus.
If those organisations are simultaneously shedding headcount and redirecting profit upward to shareholders and AI infrastructure investment rather than wages, the client base for independent consultants contracts. The market that feels like an escape route may already be getting crowded before most people have realised they are heading for it.
The creator economy question — If people are losing their jobs and their income, who is left to buy what the creators are selling?
Then there is the content creator layer. An enormous number of people are building businesses around audiences: newsletters, podcasts, courses, YouTube channels, affiliate products, influencers of sorts.
The logic is appealing. You own your platform, your relationship with your audience, your revenue.
But this model rests on a foundation that is starting to look worrying: a large, financially comfortable middle class with disposable income and discretionary time.
Both of those things are wage-dependent.
If wages compress at scale because labour supply outstrips demand, who is buying the courses? Who is subscribing to the newsletters? Who has the headspace to consume three podcasts a week when they are anxious about their job?
Henry Ford understood this in 1914 when he doubled his workers' wages.
His reasoning was not altruistic. Workers who could not afford to buy cars were not useful to a car company. The production line required consumers. He needed his employees to be customers. That loop, production enabling consumption enabling production, is what kept industrial capitalism functional for most of the 20th century.
AI breaks that loop if the productivity gains stay at the top of the pyramid and do not circulate back down as wages or public investment.
And right now there is no serious mechanism in place to ensure they do.
The UBI existential question — is UBI actually a solution, or is it just what people say when they don't have one?
Universal Basic Income comes up a LOT in these discussions.
Simply what this promises to be: the government gives everyone a monthly payment, no conditions, as a floor beneath the economy. Simple idea. The problem is essentially everything beneath the surface 🤦🏻♀️
To fund it with AI productivity taxes you first have to define and measure AI productivity, which is fiendishly difficult.
Then you have to get sovereign nations to agree on a framework, which historically takes decades even for simpler things.
Meanwhile the companies generating the productivity are incorporating in the most favourable jurisdictions and moving faster than any regulatory body can track.
The deeper issue is that UBI assumes people are primarily motivated by consumption.
Give people enough to survive and the system equilibrates. But most people derive identity, structure, community, and meaning from work.
A monthly payment does not replace that. You end up with a population that is technically fed but purposeless and, eventually, angry. That is not a stable political outcome.
UBI is what people say when they want to sound like they have a solution without proposing anything that would threaten existing power structures.
What would actually matter is things like antitrust action on AI infrastructure monopolies, mandatory profit sharing, reduced working hours with maintained wages, and serious public investment in care, education, and infrastructure. All of which are also politically nearly impossible, but at least they address the structural problem.
The numbers
There are roughly 5.7 billion people on earth between the ages of 15 and 69. Approximately 1.5 billion are currently using LLMs in some form.
That is about 26% penetration of the population who could plausibly be using these tools. The other 74% are not yet in this conversation at all.
That gap should be a warning. The disruption so far has been concentrated in wealthier, more connected economies. The scale of what happens as adoption spreads to the rest of that 5.7 billion, without the social infrastructure to absorb it, is genuinely hard to think about clearly.
Would you call me pessimistic?
I asked Claude this directly. The answer was no, and I found it more reassuring than I expected (is Claude indulging me?) — Thanks Claude 🥹
The people who call this kind of thinking pessimistic tend to be insulated enough from the consequences that the problem feels abstract, or they have confused optimism about technology with optimism about the systems that distribute technology's benefits. Those are completely different things.
Being genuinely excited about what these tools can do and being worried about what happens to the people displaced by them are not contradictory positions.
They are honest ones held simultaneously, if you ask me.
The genuinely pessimistic position, arguably, is the one that says it will all work itself out because it always has before.
That requires ignoring that this transition has no real historical parallel, is moving faster than any previous one, and is happening inside political and economic systems that are already under significant strain.
The part I cannot not mention
This article was written with the help of Claude. I asked it questions. It gave me frameworks, data, analogies, and eventually a draft.
I edited, restructured, and added my voice.
The final product is a collaboration between a human who has been in this industry for 16 years and a tool that has read most of what has ever been written about economics, history, and labour markets.
I am aware of the irony. A recruiter whose job involves placing humans into roles, using an AI to write about what happens when AI displaces humans from roles, and publishing it on a platform already drowning in AI-generated content that nobody checked before posting.
But that irony is not a reason to be quiet, perhaps it is precisely the reason to be honest about what is happening.
The content available online is getting worse because people are using these tools without thinking. Publishing without caring whether it is true or useful or original. That is not an AI problem, rather a human problem that AI is accelerating.
I have been asking these questions since before most people in my field were taking LLMs seriously.
What happens when everyone uses AI to produce content?
What happens to job applications when AI can write them infinitely? These things are here now.
Where I land, for now
I do not have clean answers. Anyone who tells you they do is… potentially selling something.
What I do believe is this: the people who tend to fare better in major transitions are not necessarily the ones who predicted them most accurately.
They are the ones who built genuine trust and real community before the chaos peaked. Because when things get uncertain, people move towards people they already know and believe.
There is something stubbornly human about that. Algorithms cannot replicate the thing that happens when people who share real professional stakes sit in a room together and have the honest version of a conversation rather than the LinkedIn version.
The circuit of money may be broken. The political will to fix it may be largely absent. The technology is moving faster than the systems designed to contain it. All of that is true.
And in the middle of all of it, the most durable thing I can point to is: know your people. Build real things with them.
I want to see this as optimism as practice, and it is the best I got right now.
Founder of The Principal Recruiter. 16+ years in talent acquisition. Building better TA across Europe.
