The End of the Knowledge Worker?

I can’t share my thoughts with you without nudging you to carefully read four of the articles/essays that inspired them:

• The Adolescence of Technology, Confronting and Overcoming the Risks of Powerful AI — Written by Dario Amodei, co-founder & CEO of Anthropic (the creators of Claude AI), published in January 2026.
• Something Big Is Happening — Written by Matt Shumer, technology investor and co-founder & CEO of OthersideAI, published in February 2026.
• The Knowledge Economy Is Over. Welcome to the Allocation Economy. In the age of AI, every maker becomes a manager — Written by Dan Shipper, founder of Every, published in January 2024 and updated in February 2026.
• The Joy of Delegating to Competence — Written by Jason Fried, co-founder & CEO of 37signals (the creators of Basecamp and HEY), published in January 2026.

The Goal Isn’t to Cause Panic, But to Shake People’s Comfort a Little!

On one hand, Dario Amodei calls the current rush toward artificial intelligence (AI) a “technological adolescence,” a turbulent and inevitable rite of passage that will test the maturity of our social, political, and technological systems.

On the other hand, Matt Shumer says that this year, 2026, is the most important of our professional lives if we work in the knowledge economy.

💡I completely agree, and I think our window of opportunity (for us ordinary citizens with non-technical backgrounds) is shrinking more and more.

• Indeed, many of us don’t realize that what we need to learn goes far beyond simply writing better prompts.
• Of course, these are solid foundations to have, but we need much more than that to fully harness the power of today’s AI and prepare for the AI of tomorrow.

AI that, incidentally, could arrive overnight, just as abruptly as ChatGPT entered our lives.

So, if you feel that AI is just a tool for writing emails faster or an improved search engine, you’re making a judgment error that could prove dangerous for your career.

➡️Test your AI adoption to uncover 3 areas you should improve on 🤓

AI opens the door to capabilities limited only by the imagination of those who use it. 🤓

The question is whether we’ll be able to survive this transition without destroying ourselves or becoming obsolete. 🤔

The Rapid Evolution of AI Models

The idea that AI progresses in small increments is a myth ❌

In reality, behind the volatility of public opinion, there is a smooth and relentless increase in the cognitive capabilities of these models. The pace of evolution for AI models is exponential, meaning that each generation improves and becomes faster at an accelerating rate.

This phenomenon is easily explained: the more computing power and data we add, the more the systems predictably improve across nearly all measurable skills.

Well, I also can’t ignore the words of Yann LeCun (January 2026), one of the world’s leading experts in artificial intelligence and Meta’s chief AI scientist until November 2025:

[…] The technology industry will eventually hit a dead end in its A.I. development — after years of work and hundreds of billions of dollars spent [because] large language models, or L.L.M.s, the A.I. technology at the heart of popular products like ChatGPT, can get only so powerful [regardless of how much money and computing power we might throw at it].

Personal note: I totally see what he means because AI cannot evolve faster than the infrastructure that supports it ➡️ The components required for the computational power needed for a technological leap similar to that of the introduction of transformers do not yet exist.

We are no longer talking about simple software. We are approaching a level of AI described as “powerful”: models capable of outperforming a Nobel laureate in most fields (biology, programming, mathematics, writing) and acting autonomously on the internet or on computers.

Even more striking, the cycle of self-improvement has begun: AI now writes much of the code needed to create the next generation of AI. For instance, the GPT-5.3 Codex model was instrumental in its own creation, helping engineers debug its training and diagnose test results. This feedback loop means we may be just 1 or 2 years away from the point where AI will build its own future…completely autonomously. 😱

The Knowledge Economy is Dead; Long Live The Allocation Economy

Since the 1970s, we have lived in the knowledge economy. Our value rested on what we knew and our ability to apply that knowledge.

But what happens when the skill of “knowing and using the right knowledge at the right time” is performed faster and better by a computer? 🤯

We are currently shifting, quietly but surely, from the knowledge economy to the allocation economy.

• In this new paradigm, we will no longer be judged on what we do (the work of a “maker”), but on our ability to allocate and manage resources (the work of a “manager”).
• Every employee, even junior ones, will need to act as a “model manager.” Our role will consist of choosing the work to be done, deciding whether the result produced by AI is sufficient, and editing it if necessary.

This shift redefines the very concept of competence. While machines automate “low-level” thinking (i.e., which largely consists of synthesizing and summarizing information), human intelligence must now focus on direction, evaluation, and tradeoffs. This transition from execution to strategic orchestration is the only way to remain relevant.

Essentially, we will assume a role of arbitration in which critical thinking and systems thinking will be essential.

The Biases That Prevent us From Adapting to The Pace of AI

The greatest danger today is the immense gap between public perception and technological reality. Several biases, some mentioned by Matt Shumer, hinder adequate preparation:

• “I tried it in 2023 and it wasn’t that impressive”: Judging current AI based on your tests from 2023 or early 2024 is a serious mistake. In AI time, two years is an eternity. Current models are unrecognizable compared to their predecessors; they are no longer mere tools for assistance, they are beginning to demonstrate judgment and taste.
• Using the free version: Most people evaluate AI through free versions, which are often more than a year behind state-of-the-art models — It’s like assessing the potential of smartphones by using a flip phone.
• The denial of speed: Many rely on the idea that technological adoption will be slow in traditional companies. While this may provide temporary respite, “AI-native” startups risk directly disrupting established players much faster than expected.
• The illusion of protection through complexity: Some believe their profession is too nuanced or requires too much empathy to be automated. Yet partners at major law firms and doctors are already finding that AI sometimes outperforms their junior associates in contract analysis or medical diagnosis.

Skills at Risk: A Bottom-Up Shift

Unlike previous revolutions that replaced physical tasks, AI is a general substitute for cognitive work. It is climbing the “skills ladder” from the bottom up.

For example, in programming, it has gone from “mediocre coder” to “very good coder” in less than two years.

Dario Amodei predicts that AI could eliminate 50% of entry-level office jobs within 1 to 5 years. The most at-risk fields include:

• Law: Contract analysis, case law research, and drafting legal briefs.
• Finance: Creating financial models, drafting investment memos, and data analysis.
• Software engineering: AI now writes tens of thousands of lines of code, tests applications, and corrects its own errors.
• Writing and design: Marketing, technical journalism, and report writing.

The risk is the emergence of a “subclass” of workers whose intrinsic cognitive abilities would be inferior to those of AI, with no clear path to retraining since AI is improving simultaneously across all fields. Can you imagine that?! 😱

My Recommendations (And What I do Myself Every Day)

If your work takes place in front of a screen, you must and you CAN take action starting today. Here’s how:

1. Upgrade to the paid version and experiment seriously: Invest the $20 per month to access the full version at least 1 of the cutting-edge models (GPT-5, Claude Opus). Don’t use it like Google. Give it your most complex tasks, evaluate its responses, refine your instructions, and try again.

2. Spend ONE hour daily using the tool: Dedicate one hour a day to testing new tools or workflows. Don’t just read about them—use them. Whoever shows how to do in one hour what used to take three days will become the most valuable person in the organization.

3. Document your process verbally: Use transcription tools to describe your work routines to the AI. Trust me: it’s easier said than done! Then, ask it to turn your explanations into flowcharts or critique your method to identify blind spots.

4. Develop your “agent manager” skills: Learn to delegate. Project management is becoming the new programming language. If you know how to lead AI agents the way you would lead a team of human experts, you’ll gain phenomenal productivity leverage.

5. Focus on what’s hard to replace: Concentrate on human relationships, learn to build trust, and aim for roles requiring legal responsibility (signing documents, appearing in court) or a physical presence.

6. Prepare financially: Especially in the West, the assumption that your current income is guaranteed may lead you to take on new debt. Now is not the time! Instead, build up savings and give yourself options in case the transition is more abrupt than expected.

In Conclusion: it’s Time to Roll up Your Sleeves

AI is not a passing fad. On the contrary, we are on the cusp of a profound transformation of our civilization.

I often tell my friends: “The confusion and ambivalence surrounding AI must be similar to what previous generations experienced with the introduction and adoption of the Internet in the workplace.”

I feel privileged to be able to experience such a transformation—of course, with a healthy dose of anticipation and perplexity.

• Technology is improving exponentially, and the world’s largest institutions are pouring trillions of dollars into it.
• Our ability to adapt, to remain curious, and to learn continuously will be our only true, lasting advantage.

As Dario Amodei points out:

Above all, the sheer number of risks, including unknown ones, and the need to deal with all of them at once, creates an intimidating gauntlet that humanity must run.

I believe we are at a crossroads between these risks and the opportunities that will arise from them.

The success of our society will depend on our ability to face the situation head-on, without illusions and without an undue sense of superiority.

The antidote to fear is knowledge put into action. — Christopher Penn, co-founder and Chief Data Scientist at TrustInsights.ai

Don’t wait for AI to knock on your door to realize it has already transformed your neighbor. Dive in now, with urgency and curiosity. 🙋🏽‍♀️

And if you need a little help ➡️ my AI adoption test will give you 3 recommendations based on your current level to help you take it to the next level ^^ 🤓

TL;DR – The End of the Knowledge Worker?

• The knowledge economy — dominant since the 1970s — is being displaced by an allocation economy. AI no longer assists knowledge workers; it can replace their core output. Dario Amodei projects AI eliminating up to 50% of entry-level white-collar jobs within five years. The shift is already measurable in law, finance, software engineering, and content creation.
• The new competitive edge is not what you know but how effectively you direct AI agents — essentially managing AI outputs as a senior manager manages junior staff. Workers who evaluate, edit, and orchestrate AI outputs will outperform those who still produce from scratch.
• Treating current AI capabilities through the lens of 2023 test results, or via free-tier access, is a strategic miscalculation. Premium models are categorically different. The adaptation window for non-technical professionals is narrowing fast.