The Executive's Signature: We Don't Blame the Algorithm

The single most important sentence this desk says about the AI economy is a refusal: we do not blame the algorithm. The machine is not an actor. It did not convene a meeting and decide to dissolve the research department. It did not weigh the sales team against a subscription to a compute cluster and choose the cluster. It did not sign a multi-year, multi-billion-dollar contract for GPUs. Every one of those actions was taken by a person, with a name, a title, and a signature — and the entire rhetorical architecture of "AI-driven" layoffs exists to move the reader's attention away from that person and onto the tool. This piece is about moving it back.

The passive voice is doing the work

Listen to how these decisions are announced and you will notice the grammar. Roles "were eliminated." Functions "have been automated." The organization "is being reshaped by AI." The machine is the subject of every sentence; the executive is nowhere in the clause. This is not an accident of style. The passive construction is load-bearing — it converts a discretionary human choice into something that sounds like weather, an impersonal force the company is merely responding to. "AI made these roles redundant" lands very differently from "our CFO decided to cut these roles and reallocate the budget to compute, and our CEO announced it in language the market rewards." Both describe the same event. Only one has a defendant.

The scale of the naming problem is measurable. Challenger has AI cited in 101,743 US job-cut announcements so far this year — the top stated reason for four consecutive months. TechCrunch maintains a running, named list of the major tech layoffs that have specifically invoked AI. That list is useful precisely because the invocation is a choice: a company decides whether to attribute a layoff to AI, to "restructuring," to "culture," or to nothing at all. Our own Layoffs by AI ledger tracks that decision as its own data field — who claimed the machine did it: the company, the media, or no one. The attribution is not a fact about the technology. It is a communications decision, and it has an author.

The self-fulfilling prophecy

There is a darker mechanism hiding inside the "AI made them obsolete" claim, and it deserves to be named plainly because it inverts cause and effect. An executive cuts experienced staff and redirects the budget to AI tooling. The remaining workforce, now thinner and stripped of the people who held the context, becomes measurably less effective. Output quality drops; the expensive models produce confident work that no one senior is left to catch. And then — this is the move — the resulting decline is presented as further evidence that humans were the problem and more automation is the answer. The prophecy fulfills itself: you predicted the people were dispensable, you removed them, the removal caused damage, and you cite the damage as proof you were right.

The best enterprise data we have says the first step of that loop is already unjustified. Gartner's 2026 study found that roughly 80 percent of organizations reduced their workforce in the name of AI, and that those reductions bore no relationship to actual returns — the companies seeing real gains were the ones investing more in their people to run the systems, not fewer. So the executive who frames a cut as AI-driven efficiency is, per the research, usually not describing a productivity fact. They are describing a decision, made under market pressure, that the data does not support — and then borrowing the machine's name to make it sound inevitable.

Trace it to the signature

So the desk does the unglamorous thing. We trace the trade back to the people who made it. The CEO who chose to announce a compute commitment because FactSet's data is right — the market rewarded the 306 companies that said "AI" and punished the silence. The CFO who classified the infrastructure spend in the language of "operating efficiency," keeping the depreciation out of this quarter's story. The board that pressured the pivot because a rival's stock moved. The recruiter reassigned from "talent acquisition" to "cost optimization" and told to describe the same headcount as a liability instead of an asset. And, where it is on the record, the executive who — like Meta's Zuckerberg — actually admitted the cut was about capital, not capability. These are the signatures. The SEC filings — the capex lines, the cash-flow statements, the insider-transaction disclosures — are where those signatures are legible in numbers rather than in press-release adjectives, and where the stated story can be checked against the audited one.

This is not a hunt for villains; some of these decisions may prove correct, and executives are paid to make hard calls under uncertainty. It is a hunt for accountability — the simple insistence that a choice with a maker be attributed to its maker, so that when the compute bill comes due and the AI revenue is still a rounding error, the question "who decided this, and on what evidence?" has an answer that is a person and not a piece of software.

Did AI do this, or did we?

This whole article is that question, held down until it stops squirming. The answer, every time you trace it to the end, is a human decision responding to a human incentive: a market that rewards the mention, an accounting treatment that defers the reckoning, a competitive fear that punishes hesitation, and a vocabulary that launders all of it as inevitability. The machine is the alibi, not the actor. To blame the algorithm is to grant the alibi.

What we are not claiming

We are not claiming these executives are acting in bad faith — most are responding rationally to genuinely brutal incentives, and naming the incentive is not the same as impugning the person. We are not claiming AI displaces no one; some roles are genuinely automatable, and pretending otherwise would be its own dishonesty. And we are not claiming to know which of these bets will pay off.

The claim is only this: accountability cannot be assigned to a tool. When a company says AI made a decision, a person made that decision and chose to say AI. This desk will keep tracing the trade to the signature — the executive, the filing, the public statement set beside the audited number — because that is the only place the question of responsibility can honestly be settled. The machine did not sign. Someone did. Our Money and Power lanes keep the ledger of who.