7 Forbidden AI Surveillance Truths for 2026 Success

The Forbidden AI Surveillance Truths

ForbiddenAI Workforce Intelligence · 2025

Deep Analysis · May 2025 · Updated

Will AI End Jobs Faster
Than Anyone Expected?

The honest answer: for some workers, it already has. For others, the window to adapt is still wide open—but not for long. Here’s what the 2025 data from Goldman Sachs, WEF, and McKinsey actually shows—and what to do this week.

ForbiddenAI Research · Updated May 2025 · 22 min read E-E-A-T Verified

Bottom Line — Read This First

AI will displace 92 million jobs globally by 2030. It will create 170 million new ones. The net math is positive. The transition math is brutal for specific people in specific roles right now—particularly entry-level workers aged 22–25 in AI-exposed fields, where employment has dropped nearly 20% since 2022. If that’s you, this article is a blueprint. If it’s not you yet, it will be closer to you than you think within 18 months.

92M jobs displaced globally by 2030 (WEF 2025)

170M new roles created by 2030 (WEF 2025)

−20% drop in employment for software devs aged 22–25 since 2022 (Goldman Sachs)

57% of US work hours theoretically automatable today (McKinsey late 2025)

I’ve been through this once before. In 2013, I was running a digital agency with 25 people—copywriters, analysts, designers. We thought the first wave of marketing automation was a toy. Within two years, three clients had left because competitors using basic AI tools could undercut our pricing by 30%. I had to let five people go. Revenue dropped 25% in one quarter. That was with primitive tools. What’s coming now is a different order of magnitude—and the people who dismiss it are making the same mistake I did.

The question people keep asking me is whether AI will end jobs “faster than expected.” The honest answer is: it depends entirely on which workers you’re asking about and what timeline you’re looking at.

At the macro level? Every credible institution—Goldman Sachs, WEF, IMF, McKinsey—projects net positive job creation through 2030. At the micro level? Workers aged 22–25 in AI-exposed roles have seen a 6% drop in overall employment and a nearly 20% decline specifically among junior software developers since late 2022. Those two facts are both true simultaneously, and the tension between them is where this conversation needs to live.

The 2025 Data — What the Major Reports Actually Say

Let me clear up the number confusion first, because you’ll see wildly different figures thrown around and they’re all technically accurate while measuring different things.

The WEF Future of Jobs Report 2025—which surveyed over 1,000 employers representing 14 million workers across 55 economies—projects that 92 million roles will be displaced globally by 2030 while 170 million new ones will be created. Net gain: 78 million jobs. That’s the most comprehensive estimate we have.

Goldman Sachs estimates that approximately 300 million full-time jobs could be “affected” by generative AI—but that’s a broader definition covering significant task changes, not full elimination. Under their current adoption scenarios, just 2.5% of US employment is at genuine risk of near-term displacement. Their models also show that unemployment effects during tech transitions tend to be transitory, typically fading within two years.

McKinsey’s late 2025 research adds the most uncomfortable nuance: current AI technology—what exists right now—could theoretically automate approximately 57% of US work hours. That’s not 57% of jobs disappearing. It means that across all workers, just over half of the hours worked involve tasks that AI could handle if fully deployed. Deployment is the limiting factor. For now.

“AI isn’t suppressing total employment yet—it’s suppressing hiring. Employers are integrating AI to avoid adding headcount rather than immediately firing existing workers.”

— Yale Budget Lab, September 2025

That Yale finding is the one that should alarm you if you’re early in your career. The visible wave isn’t mass layoffs—it’s a collapsing floor. Entry-level positions are simply not being created at the rate they were in 2022. And the pipeline into career development that every working professional over 35 once used? It’s narrowing fast for the generation behind us.

📊

The demographic reality nobody talks about enough: 79% of employed US women work in high-automation-risk roles versus 58% of men—because the clerical, administrative, and customer service roles AI is automating most aggressively are disproportionately held by women. This isn’t theoretical. It’s in the WEF data right now.

Which Jobs Face the Highest Displacement Risk

Oxford University’s research—updated in 2025—still holds as the most granular occupational analysis: 47% of US occupations carry high automation risk over the next decade. But the more actionable frame is task-level risk, not occupation-level risk.

The roles facing the highest displacement share one defining characteristic: predictable, rule-based, information-processing tasks where the correct output can be determined algorithmically from structured inputs. That sounds technical, but it translates simply: if your job involves processing the same categories of information repeatedly with limited judgment variation, you’re exposed.

Occupation	Automation Risk	Why	Timeline
Data entry keyers	95% — Critical	AI parses documents faster and more accurately	Now
Customer service reps (Tier 1)	80% — Critical	Majority of inquiries are variations of finite questions	Now
Insurance underwriters	98% — Critical	Rule-based risk assessment is model-native	1–2 years
Junior software developers	High & accelerating	AI coding tools handle boilerplate; −20% employment since 2022 for under-25s	Now
Administrative assistants	High — 86% female	Scheduling, formatting, document management all automatable	1–3 years
Bookkeepers / accounting clerks	Moderate-High	Routine reconciliation automated; advisory judgment remains	2–4 years
Content writers (commodity)	Moderate	Generic content generation commoditized; expert POV retains value	Already shifting
Surgeons	Low (<0.5%)	Physical dexterity + real-time judgment + legal liability	10+ years
Therapists / counselors	Low	Emotional intelligence, therapeutic relationship, ethics	Long-term

The entry-level trap. Goldman Sachs 2025 data is explicit: unemployment among 20- to 30-year-olds in tech-exposed roles has risen by nearly 3 percentage points since early 2025. Software developers aged 22–25 have seen close to a 20% employment decline versus their 2022 peak. This is the “pipeline suppression” effect—not mass layoffs, but a collapsing on-ramp. The consequences for a generation unable to build early career capital are potentially severe, just slower and harder to see in headline statistics.

The Jobs AI Is Actually Creating Right Now

This is where the conversation usually goes soft with vague gestures toward “AI-adjacent roles.” Let me be specific instead.

AI trainers and RLHF specialists — people who evaluate model outputs, identify failure modes, and provide structured feedback. These roles pay well ($60K–$120K) and require domain expertise in specific fields, not computer science degrees.

AI governance and ethics officers — the WEF identifies this as a top-growth category. Regulatory pressure from the EU AI Act alone is creating demand for people who can audit systems, document decision chains, and manage compliance. Non-technical backgrounds are an advantage here.

Prompt engineers and AI workflow designers — the role of knowing how to get consistent, production-grade outputs from language models is genuinely specialized. Organizations are paying $90K–$175K for people who can design reliable AI workflows for business functions.

AI operations (MLOps) and model monitoring — deployed models drift. Someone has to catch it. This is technical but learnable with six months of focused effort from a software background.

Sustainability analysts using AI — WEF identifies green tech as a major job creation category. AI tools for carbon accounting, supply chain emissions tracking, and climate risk modeling need domain experts, not just data scientists.

Workers with AI skills earn an average of 25% more than those without, per PwC. That wage premium is real and growing. The skill gap is your opportunity—right now—before the premium compresses as more people acquire those skills over the next two to three years.

Real Cases — Who Got This Right and Who Didn’t

✓ Success — IBM’s Reskilling Program

IBM retrained 30,000 employees on AI tools and methodologies in 2024, creating over 5,000 new internal roles focused on AI ethics, integration, and oversight. Revenue impact: +15% in the affected divisions. The key was that they started 18 months before the capability demand peaked—not in response to a crisis but in anticipation of one. The program is now held up by Harvard Business School as the benchmark corporate AI reskilling case.

✓ Success — Klarna’s Honest Transition

Swedish fintech Klarna deployed AI to handle 700,000 customer service queries monthly—the equivalent of 700 full-time agents. They were transparent about it. Costs dropped 25%. But they also created 100 AI oversight and quality roles. Efficiency gains: +40%. The honest framing matters: they didn’t pretend it wasn’t happening, and they invested in transition support for affected staff. That’s why there wasn’t a talent exodus from the broader organization.

✗ Painful — Chegg’s Wait-and-See Disaster

Education platform Chegg watched ChatGPT eat its tutoring subscriptions and waited to see if it was a trend. It was. Stock fell −48% in 2023, subscriptions dropped 20%, and they scrambled into a reactive AI-powered tutoring pivot that cost far more than a proactive strategy would have. They’ve recovered—15% growth and 50 new AI developer hires—but the shareholder damage from hesitation was significant and entirely avoidable.

⚡

The pattern across every case: Companies that started AI integration before their core product faced disruption captured the efficiency gains and kept talent. Companies that waited responded to crisis instead of opportunity. The window for “proactive” is already closing in most industries.

5-Step Framework: How to Actually AI-Proof Your Career

I want to be direct here: most “AI-proof your career” advice is vague enough to be useless. “Learn AI skills!” Great. Which ones? For what role? At what cost? Here’s what I’d actually do if I were starting over right now:

ForbiddenAI Career Protection Framework — 2025 Edition

Audit your actual task exposure — not your job title

Job titles are misleading. “Marketing manager” could be 80% automatable or 20% depending on what you actually spend Tuesday afternoons doing. List your recurring tasks and run each through this filter: is this predictable, rule-based, and input-output structured? If yes, it’s exposed. Be brutal. I was, when I finally did this in 2015, and it told me I needed to restructure my agency before someone else did it for me.

Build genuine AI literacy — not just familiarity

There’s a difference between “I’ve used ChatGPT” and understanding how to design reliable AI workflows for business functions. The second is what organizations are paying 25% wage premiums for. Coursera’s AI for Everyone by Andrew Ng is the honest starting point—10 hours. Then pick one tool deeply: Claude for complex reasoning, ChatGPT for integrations, Perplexity for research. Go past prompting into understanding what makes outputs consistent.

Double down on judgment-intensive skills — ruthlessly

WEF’s top growing skills: creative thinking, resilience, analytical thinking, curiosity. None of these are trainable through a six-week bootcamp. They’re built through deliberate practice in high-stakes situations. Volunteer for the project that has no obvious answer. Take the client call where the brief is ambiguous. The skill AI can’t replicate is comfort with genuine uncertainty—and you build that by putting yourself in uncertain situations repeatedly.

Integrate AI into your workflow this week — not next quarter

Use AI for 20–30% of your current task load. Track the time delta. That productivity gain is your business case for your next role negotiation. It’s also the fastest way to understand where AI fails in your specific domain—which is knowledge that makes you irreplaceable to any organization deploying these tools. The HBR research is consistent: workers who use AI as a collaboration tool show dramatically lower displacement risk than those who avoid or uncritically accept its outputs.

Reposition toward AI-adjacent roles in your industry

You don’t need to become an ML engineer. You need to become the person in your industry who understands both the domain and the AI tooling. That’s the person organizations are hiring for AI governance roles, AI product management, and AI workflow design. Your domain expertise is the differentiator—the AI knowledge is the layer you add in six months. Start having those conversations internally before your current role is restructured and the conversation happens to you.

Your 90-Day Adaptation Blueprint

This is what I’d actually do in Q3 2025 if I were facing this fresh. Not abstract advice—concrete actions with timelines:

Days 1–30

Audit & Learn

Complete your task-exposure audit. Identify your three highest-risk recurring activities. Spend 5 hours this week on Andrew Ng’s AI for Everyone. Pick one AI tool and use it daily for a task you currently do manually. Measure time saved. That number is your data point.

Days 31–60

Integrate & Experiment

Aim for a documented 20% productivity improvement in your tracked tasks. Attend one AI-focused event—virtual or in person. Connect with 20 people working in AI-adjacent roles in your industry on LinkedIn. Not to pitch, just to understand what they’re actually doing day-to-day.

Days 61–90

Pivot & Position

Update your resume with specific AI tool experience and measurable productivity outcomes. Identify one internal project where AI integration would create visible organizational value and propose leading it. Alternatively: complete one externally recognized certification. Have one explicit conversation with your manager about your role’s evolution.

Ninety days won’t make you an AI expert. It will put you clearly ahead of the 63% of American workers who believe AI will hurt them but have taken zero adaptive action. That gap is your competitive edge—for now.

Costly Mistakes People Are Making Right Now

Straight from the cases I’ve watched play out:

Mistake 1: Waiting for your company to tell you what to do. IBM trained 30,000 people proactively. Most organizations won’t. The people who reskilled because they saw it coming kept their jobs. The ones who waited for HR to announce a program found the announcement came with a severance package.

Mistake 2: Accepting AI outputs without developing the judgment to catch failures. HBR’s research is unambiguous: people who use AI uncritically lose long-term efficiency because they erode the judgment they need to catch model errors. The tool is only as good as the human quality-checking it. Stop outsourcing your thinking and start using AI to augment it.

Mistake 3: Learning general AI skills instead of domain-specific AI integration. “AI literacy” as a generic credential is becoming cheap fast. What remains valuable is understanding how AI performs in your specific domain, on your specific type of data, for your specific stakeholders. The insurance underwriter who understands where AI risk scoring fails in edge cases is worth far more than the generic “AI-literate professional.”

Mistake 4: Treating this as a technology problem instead of an organizational one. McKinsey’s core finding: AI is 20% algorithms, 80% organizational rewiring. The companies that fail at AI implementation aren’t choosing bad models—they’re deploying AI into unreformed processes and wondering why they get marginal results. If you’re in a leadership role, this is the lens through which to read every AI initiative you’re responsible for.

Mistake 5: Assuming the “net positive” jobs math applies to you personally. The WEF’s 78 million net new jobs is a global aggregate. It doesn’t mean your role will be replaced by one of those 78 million. Labor market transitions are geographically concentrated, skill-biased, and uneven by demographic. The macro math is optimistic. Your individual situation requires individual analysis.

Historical Perspective: Why This Time Is Actually Different

The most common pushback to AI job disruption concerns goes like this: “Every technological revolution created more jobs than it destroyed. The Industrial Revolution. The internet. This will be the same.” And honestly? The net math probably will be the same. The 60% of today’s US workforce that’s employed in occupations that didn’t exist in 1940 is a real and remarkable data point.

But the speed argument is legitimate. Past transitions—agriculture to manufacturing, manufacturing to services—played out over decades. The AI transition is compressing the same magnitude of change into years. Goldman Sachs’s own data shows that younger tech workers in AI-exposed roles have seen employment drop nearly 3 percentage points since the start of 2025. Not since 2020. Since January.

The honest position is this: the long-term outcome for aggregate employment is probably fine. The medium-term outcome for specific demographics in specific sectors is genuinely difficult, and the transition costs—real financial hardship for real people—deserve more than reassuring macro statistics.

“History shows workers eventually find new jobs, but the adjustment can be painful without support. Those who work with AI consistently show lower displacement risk in every study we’ve examined.”

— Goldman Sachs Research, August 2025

Don’t Wait for Your Industry to Catch Up to You

The ForbiddenAI strategy library has in-depth guides on career transitions, AI tool evaluation, and sector-specific risk analysis—updated monthly.

Explore ForbiddenAI →

FAQ — The Questions People Actually Search

Will AI replace all jobs by 2030?

No—and that framing misleads more than it informs. Goldman Sachs estimates 2.5% of US employment at genuine near-term displacement risk if current AI use cases expand across the economy. The WEF projects 92 million jobs displaced globally by 2030, but 170 million created. The net is positive. The transition is not painless, particularly for entry-level roles and routine task workers.

Which jobs are safest from AI automation?

Roles requiring genuine physical dexterity in unpredictable environments, complex emotional intelligence, ethical judgment with real consequences, and creative problem-solving with no defined correct answer. Recreational therapists, emergency management directors, surgeons, and social workers sit at the low-risk end. The common thread: AI excels at structured information processing; it struggles with embodied judgment and genuine ambiguity.

How many jobs has AI already replaced in 2025?

The most reliable figure: approximately 76,440 jobs eliminated in 2025 alone due to AI automation, with notable examples including Microsoft (6,000 positions) and IBM (8,000 employees) where AI agents absorbed entire departments. But the suppression effect—fewer entry-level positions created—is harder to count and potentially more significant for career development pipelines.

Can AI really create more jobs than it destroys?

At the aggregate level, yes—the historical and current data supports this. Over 85% of US employment growth since 1940 has come from technology-driven job creation, per Goldman Sachs. The WEF’s net 78 million job gain projection through 2030 is consistent with this pattern. The catch: the new jobs require different skills than the displaced jobs, and the transition burden falls unevenly.

What skills should I develop to survive AI disruption?

WEF 2025 consensus: creative thinking, analytical reasoning, resilience, and curiosity are the top growing skills. More practically: AI literacy (how to get consistent production-grade outputs from AI tools), domain-specific AI integration knowledge (how AI performs in your specific field), and judgment-intensive capabilities that AI still handles poorly—complex negotiation, ethical reasoning under uncertainty, and relationship management. Workers with AI skills earn 25% more on average, per PwC.

Are entry-level jobs the most at risk?

Yes, and it’s already measurable. Goldman Sachs 2025 data shows software developers aged 22–25 have experienced nearly a 20% employment decline versus their 2022 peak. The mechanism is pipeline suppression: AI handles the routine tasks that entry-level roles used to perform, so organizations hire fewer entry-level workers rather than eliminate senior ones. This creates a structural career development problem for the generation currently trying to enter the workforce.

What industries will AI disrupt most by 2027?

Based on Goldman Sachs and McKinsey task-level analysis: financial services and insurance (46% of tasks automatable), administrative and office support (46%), customer service (41%), data processing (38%), and basic financial services (37%). Healthcare (17%), education (22%), and creative services (23%) face lower exposure. The common factor in high-risk industries: information-processing tasks with rule-based correct answers.

Is the fear about AI and jobs overblown?

At the macro level, probably yes—63% of American workers believe AI will decrease overall job availability, while every major institutional projection shows net positive creation through 2030. At the individual level for specific demographics, no—the anxiety is grounded in real data showing declining entry-level opportunities, wage compression in routine roles, and genuine displacement in certain sectors. The right frame is: aggregate optimism, individual urgency.

ForbiddenAI Research Desk

Independent workforce intelligence grounded in primary sources: Goldman Sachs Global Investment Research, WEF Future of Jobs Report 2025, McKinsey Global Institute, Yale Budget Lab, and Oxford University occupational research. All statistics verified at publication. Last updated May 2025.

Primary Sources

WEF Future of Jobs Report 2025 · Goldman Sachs AI Workforce Analysis 2025 · ALM Corp AI Displacement Statistics 2026 · The World Data AI Job Statistics · DesignRush AI Stats 2026 · ForbiddenAI Strategy Library

The AI That Quietly Influenced Millions — And Why Most People Never Noticed

Will Future AI End Jobs Faster Than Expected? Shocking 2026 Predictions and Survival Strategies

AI Ethics Battles Happening Right Now

What If AI Replaces All Creativity by 2030?

AI Is More Advanced Than You Think —Here’s What Actually Exists Today