Time Is King
After selling the AI job apocalypse with certainty, its brokers are quietly editing their own script.
In February, I wrote my most liked (and hated) article: I’m Sorry to Burst Your Bubble: You Are Being Fooled About AI, and You Will Soon Feel Really Stupid. Most people understood my point (so overall, the Substack community does not disappoint). But many people didn't. They got stuck on the “stupid” word and reacted as if I was attacking their entire family generations. Others, in a very lazy way, concluded I was just another “anti-AI guy”. The point was never that AI was useless, but that a serious engineering achievement had been insanely conflated with an avalanche of fundraising language, theatrical warnings, and predictions designed to make the general public feel anxious about the “unknown”.
We are surprisingly living in times where skepticism is treated as something negative by default, which is extremely concerning. Treating caution as denial and acting as if asking for evidence is a personal offense is a move against progress itself. That was always revealing. When a person becomes angry because you ask how a prediction can be measured, you are usually not dealing with science anymore but with belief. I have been vocal about this in virtually all my articles.
UOL, a Brazilian news portal, just published a useful summary of what has become increasingly obvious: after years of dramatic warnings about a labor-market “bloodbath,” some of the most prominent AI executives are softening the language around mass job loss. What a shock! Who would know? The story, all of a sudden, is no longer quite so apocalyptic. It is now more carefully packaged as productivity, transition, reskilling, augmentation, and economic complexity.
Confident predictions of sudden, sweeping white-collar elimination? Who said that? The message is now being revised, qualified, softened, and placed inside a more institutionally acceptable frame.
AI Was Never The Problem
I work with AI every day. I use it, test it, deploy it, break it, benchmark it, and build around it. I have no interest in pretending these systems are trivial. They are powerful tools. In some workflows, they are trully transformative, as I have stated over and over again in previous posts.
But I can't stand the noise, the false prophets, and all the confusion generated, in part by ignorance, in part by design.
The public was told, again and again, that entire professions were about to disappear, junior roles were finished, coding was over, agents would soon replace teams, and companies would become mostly autonomous. On the same note, we heard that AGI was near, here, there, or inevitable by next quarter, depending on the podcast, the valuation round, or the conference stage.
In another article, the AI Prediction Audit, I tried to bring some discipline to that circus. I looked back at major AI claims, scored them, and asked a simple question most people in the hype economy annoyingly avoid: did the claim actually age well? The result was not flattering. The boldest near-term predictions, especially the ones involving rapid replacement, AGI in months, or massive societal transformation, tended to perform poorly. And all I did was to pay attention to facts.
The signs were everywhere: vague definitions, moving timelines, unfalsifiable claims, incentives tied to attention, and executives speaking with a level of certainty that reality rarely rewards.
Now the Language Is Changing
Axios reported that some prominent AI evangelists are backing away from earlier “worker bloodbath” rhetoric, with the shift occurring as public skepticism toward AI grows and major AI companies such as OpenAI and Anthropic move closer to public-market scrutiny.
When the audience was venture capital, breathless predictions were useful. A technology that might merely improve productivity is interesting. But A technology that might eliminate half of white-collar work is a civilization-level story is a conversation with much worse, serious repercussions, the type of narrative VC funds like to hear.
But when the audience becomes public investors, regulators, enterprise buyers, employees, and the broader labor market, that very same same language becomes a liability. What was a superpower is now weakness, a point of failure.
Reuters reported that Anthropic confidentially filed for a U.S. IPO on June 1, 2026, and that OpenAI followed with its own confidential U.S. IPO filing a week later. Reuters had previously reported that OpenAI was laying groundwork for a potential IPO at a valuation of up to $1 trillion.
While a private company can capitalize on their own self-promoted circus of horrors, a public-market candidate must sell durability. It must look less like a social risk, human elimination, and more like an operating business. It must convince investors that it can grow without triggering political backlash, labor panic, and regulatory hostility.
“Wipe Out” and “Change” Are Not the Same Thing
Matt Garman, CEO of Amazon Web Services, pushed back on predictions of a white-collar job wipeout, arguing that AI may change many jobs without making them disappear. He also pointed to Amazon’s plan to hire 11,000 software engineering interns and early-career employees globally in 2026 as evidence that the company still sees long-term value in developing junior technical talent.
Again, the idea that jobs would disappear, AI would replace humans, and companies would be fully automated was sold with certainty. And again, that kind of certainty is absolutely irresponsible. Labor markets are not demo videos. They are complex systems involving budgets, incentives, training pipelines, regulation, customer demand, organizational friction, risk tolerance, and human behavior.
The Data Is Mixed, Which Is Exactly the Point
TechCrunch covered a Ramp and Revelio Labs report finding that companies with high AI spending grew headcount faster than peers, including in entry-level roles. The report found that “high-intensity adopters” increased headcount by 10.2%, while entry-level headcount rose by 12%. However, these companies tend to be tech-forward, fast-growing firms, which makes it difficult to claim that AI alone caused the hiring growth.
This is a completely different story. Notice how the data is clean. The fact being reported is useful, quantifiabl, within context, and clearly limited to a very particular scope. In other words, the type of claim that can be challenged.
When a large company cut thousands of jobs and blames it on AI, that's an attribution, not a fact. The overall understanding among researchers observing this phenomenom is that many executives found in AI the perfect scape goat to justify bad hiring.
What we broadily see about AI is a surge of people and companies building things really fast, not necessarily making money from all of this building. We live in a time where we have an agent for everything and yet, work is not necessarily more relevant. We generate more things faster. That's not value in and on itself.
Even AI companies themselves are hiring more and paying more now than before. Think about what this means when we have the most powerful models, agents, workflows, etc.
Time is Undefeated
There is a phrase I like: time is king. I use it all the time. I am reminded of it all the time. It applies to absolutely everything.
Time has no need for arguments nor cares about replies, comments, applause, or social pressure. It does not get intimidated by consensus. It simply passes. And when it does, it reveals the truth in every single circumstance.
This is what is happening now. The script is being edited. All of that irresponsible certainty is being quietly softened. The job apocalypse is being reframed by something more operational, almost natural. The industry is discovering nuance at the precise moment nuance became financially useful.
I continue to believe that AI is too important to be left to people selling theatrical certainty, frivolous excitment, fear, and chaos. It deserves serious discussion, serious testing, serious economic analysis, and serious public education.
That was my point then and it remains my point now.
And for those who attacked the warning instead of examining the evidence, time has now done what time always does.
Time is king!


Nicely done, David, useful. I’m sharing. It is not easy to quantify technology, and much harder still to think about what it will come to mean — one of the themes of a recent book of mine about security. You think it’s a plane; I think it’s a missile. Anyway, nice job.
Exactly correct.