AI and Jobs in 2026: Rising Tide or Tsunami?

16,000 net jobs lost every month in the United States. That’s the number Goldman Sachs dropped on April 6, 2026. The same week, MIT published a study claiming AI’s impact on employment looks more like a “rising tide” than a “tsunami.” Who’s right? And more importantly — what do you do with contradictory data when your own position might be on the chopping block?

In three months, over 85,000 tech jobs vanished worldwide. Block, Oracle, Amazon, Meta — the giants are laying people off and pointing at AI. Yet the most serious researchers say we’re far from the predicted apocalypse. The truth, as usual, sits somewhere in between. And it’s far more interesting than the headlines.

85,000 Tech Jobs in Three Months: The Numbers That Scare

Q1 2026 will be remembered. Within weeks, a cascade of massive layoffs rocked the global tech industry:

Block (Square): 4,000 positions cut on February 27 — 40% of global headcount — making it the largest layoff explicitly attributed to AI in tech history. The stock jumped 20% the same day.
Oracle: between 20,000 and 30,000 employees laid off on April 1 via a 6 AM email signed “Oracle Leadership.” The savings target: $10 billion, redirected to AI data centers.
Amazon: 4,700 corporate roles between January and March, part of an “AI-first” restructuring.
Meta: 3,600 positions in February, tied to the AI pivot.
Dell: 2,200 roles, shift toward AI servers.
Intel: 2,800 positions, foundry restructuring.

All told, according to the layoffs.fyi tracker, over 85,000 tech positions were eliminated in Q1 2026. And in 20.4% of cases, companies explicitly cited AI as the driver — up from less than 5% in 2024.

The signal is unmistakable: AI is no longer a press-release excuse. It’s become a full-blown restructuring argument.

Rising Tide or Tsunami? What MIT’s Study Reveals

On April 2, 2026, the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) published a paper that reframes these numbers: Crashing Waves vs. Rising Tides.

The core question: will AI slam into entire industries like a tsunami (“crashing waves”), or advance gradually across the whole economy like a rising tide?

Their method: over 17,000 evaluations by real workers across 3,000 text-based tasks from the U.S. Department of Labor’s O*NET database. The verdict is nuanced but significant:

Current LLMs complete 60% of text-based tasks at a level human managers deem “minimally sufficient.”
Only 26% meet a “high quality” standard.
Most tasks could reach an AI pass rate of 80–95% by 2029 — but at minimum quality, not perfection.

The critical takeaway: AI isn’t progressing in sudden sector-wide shocks. It’s improving gradually across a wide spectrum of tasks. It’s a rising tide, not a tsunami. Workers won’t be blindsided overnight — they’ll see it coming.

But the researchers add a caveat that should keep anyone from relaxing too quickly: “a rising tide can still be very disruptive if it rises fast.”

Goldman Sachs Enters the Chat: 16,000 Net Jobs Per Month

Where MIT speaks in trends, Goldman Sachs speaks in raw numbers. And they’re significant.

In an April 6, 2026 note, Goldman Sachs economist Elsie Peng produced one of the most granular analyses to date on the two competing effects of AI on employment:

Effect	Monthly impact (USA)
Substitution (AI replaces the worker)	−25,000 jobs/month
Augmentation (AI makes the worker more productive)	+9,000 jobs/month
Net effect	−16,000 jobs/month

What’s striking isn’t just the net figure. It’s the demographic breakdown. Young workers (Gen Z, under 30) are disproportionately hit:

The unemployment gap between junior and experienced workers has widened significantly in AI-exposed occupations.
The pay gap has worsened too: according to Goldman’s regression, a one-standard-deviation increase in AI substitution exposure widens the junior-senior wage gap by 3.3 percentage points.

In plain terms: it’s not seniors taking the first hit. It’s labor market newcomers — those who haven’t yet developed the expertise and judgment that AI can’t (yet) replicate.

What Do AI Companies Themselves Say?

In a rare move, Anthropic — the maker of Claude — published its own study on AI’s labor market impact in March 2026.

Their finding is almost paradoxical. They developed a new metric — “observed exposure” — combining the theoretical capability of LLMs with their actual enterprise usage. The result is surprising:

Real-world AI adoption remains a fraction of what the technology is theoretically capable of.

In other words, AI could automate far more tasks than it actually does today. The bottleneck isn’t technological capability — it’s adoption. Legal barriers, human review processes, organizational habits, and plain inertia still massively slow deployment.

Anthropic’s study found little evidence of a systematic rise in unemployment in the most AI-exposed occupations since late 2022. But a weak signal emerged: a slowdown in hiring young workers in exposed professions — consistent with Goldman Sachs’ data.

The most exposed workers according to Anthropic? Those who are older, more educated, better paid, and… predominantly women. A profile that doesn’t match the typical image of a “worker threatened by technology” at all.

Is Europe Safe?

No. But the picture is more nuanced than in the United States.

In March 2026, a joint study by Coface and the Observatory for Threatened and Emerging Jobs (covered by Le Monde) estimated that 3.8% of French employment is already weakened by generative AI deployment — roughly 1 million positions. Extend the lens to jobs “exposed” in the medium term, and the number climbs to 5 million workers.

According to Coface, the most affected jobs aren’t what you’d expect. Not factory workers or manual laborers — they’re intellectual and information-based professions: financial analysts, accountants, writers, lawyers, developers. The well-paid jobs, paradoxically.

Meanwhile, AI job demand in France is surging: 166,000 AI-related job postings published in 2026, placing France at the top of Europe. The paradox: AI destroys traditional roles while creating massive demand for new skills.

According to PwC, the French companies most exposed to AI are already seeing significant performance gains — but those gains aren’t reaching all employees yet. The value created by AI is captured primarily by companies, not redistributed.

The trend is similar across Europe: the UK, Germany, and the Nordics all report accelerating AI-driven restructuring in white-collar sectors. The European AI Act adds a regulatory layer that may slow displacement in some high-risk categories — but it won’t stop the tide.

AI-Washing: When Layoffs Have Nothing to Do With AI

There’s an elephant in the room. And Bloomberg named it in early March: “AI-washing.”

When Block cuts 40% of staff while posting $1.3 billion in profit and its stock soars 20%, it’s fair to ask whether AI is really the driver — or the excuse.

Several indicators point to an increasingly documented phenomenon: companies using the “AI” narrative to justify restructurings that would have happened anyway. The real reasons are often more mundane — financial optimization, shareholder pressure, post-pandemic correction of the 2020–2022 over-hiring spree.

Oracle is a textbook case: its stock had plunged 25% since January 2026. The 30,000 layoffs aren’t so much a response to AI as they are a reaction to a $20 billion data center funding gap. The “AI” framing reassures investors — it provides a transformation narrative instead of a strategic failure admission.

According to layoffs.fyi data, only 20% of tech layoffs explicitly cite AI. And among those, how many are genuinely driven by automation? The real figure is probably far lower.

The risk of AI-washing goes beyond the companies involved. It distorts public debate by artificially inflating the perceived impact of AI on employment — creating a panic that could, ironically, slow the real, productive adoption of these tools.

What Still Protects You (and What Doesn’t Anymore)

If the data converges on one point, it’s this: it’s not your job that’s threatened — it’s your tasks.

MIT’s study shows AI advancing across a wide spectrum of text-based tasks, not wiping out entire professions. A lawyer who spends 60% of their time on standard contract drafting is more exposed than one focused on negotiation and strategic counsel — even though they share the same title.

What no longer protects you in 2026:

A degree alone — Anthropic’s study shows the most exposed workers are also the most educated
Seniority — paradoxically, the routine tasks of senior roles are just as automatable as junior ones
A “creative” field — writing, design, and translation are already massively augmented by AI

What still does:

Contextual judgment — knowing when NOT to apply the standard rule
Ambiguity management — situations where there’s no clear right answer
Complex interpersonal skills — negotiation, team leadership, conflict resolution
Deep domain expertise — understanding the nuances AI hallucinates over
The ability to steer AI — workers who know how to leverage AI tools are the ones creating the most value

The winning strategy, according to the combined data from all these studies? Become the pilot, not the copilot. The person who knows when AI is wrong, who verifies its outputs, who integrates it into a human workflow — that person doesn’t get replaced. They become irreplaceable.

Key takeaways:

The numbers are real but need decoding. 85,000 tech layoffs in Q1 2026, but only 20% explicitly linked to AI — and a chunk of that is AI-washing.
AI rises like a tide, not a tsunami. According to MIT, progress is gradual across a wide task spectrum. You won’t be replaced overnight, but you could be gradually outpaced if you don’t adapt.
Young workers are hit first. Goldman Sachs shows the substitution effect strikes entry-level positions hardest. AI creates a “dam at the door” for Gen Z.
Europe isn’t immune. 5 million French workers exposed according to Coface, with a paradoxical impact on the highest-paid professions.
The real danger isn’t replacement — it’s erosion. No tsunami, but rising waters. Those who learn to swim now will be the ones still afloat.

FAQ

Will AI eliminate my job in 2026?

Probably not immediately. According to MIT, AI is progressing gradually on text-based tasks and could reach 80–95% “minimum competency” by 2029. But “minimum” isn’t “perfect” — jobs with low error tolerance (medicine, law, finance) remain protected in the short term. The real risk is the gradual transformation of your tasks, not the sudden elimination of your role.

Which jobs are most threatened by AI?

Contrary to popular belief, it’s not low-skilled positions. According to Anthropic and Coface, the most exposed professions are intellectual and well-paid: data analysis, writing, accounting, junior software development, translation. Manual, relational, and strong-judgment roles remain less exposed.

How should I prepare for AI’s impact on my career?

The data converges: the best strategy is to become an “AI pilot” rather than trying to compete with it. In practice, that means learning to use AI tools in your domain, developing your verification and critical thinking skills, and strengthening the capabilities AI doesn’t master — negotiation, ambiguity management, and complex human relationships.