The real scale
Goldman Sachs estimates that up to 300 million full-time global jobs could be affected by AI automation. "Affected" doesn't mean "eliminated" — it means a significant portion of work within those roles can be done by AI.
eliminated in US
cited AI
affected
More specifically, Goldman reported in April 2026 that AI is erasing approximately 16,000 net jobs per month in the United States. AI was cited as the leading reason in 26% (21,490 cuts) of layoffs announced in April 2026.
Most vulnerable categories
Administrative roles: 26% direct exposure. Functions like data entry, scheduling, recurring reports, form processing.
Customer service: 20% exposure. Especially L1 (simple questions), but L2 is also under pressure.
Translation and transcription: almost eliminated at basic levels.
Simple content moderation: massive automation.
Junior data analysis: previously someone made dashboards, now ChatGPT/Claude make them.
Basic code generation: Codex, Cursor, Claude Code accelerate junior developers 10×.
Junior legal research: the "first-year associate reviewing documents" is automated.
Basic accounting: reconciliations, transaction categorization.
The gender bias in impact
The most concerning data: 79% of working women in the US have jobs with high automation risk, vs 58% for men. Reason: women are concentrated in administrative, clerical and customer service roles — where AI has greater impact.
This creates a specific public policy challenge. Generalist reskilling programs won't suffice — focus on specific transitions for these workers is needed.
The Gen Z drama
Generation Z suffers the displacement most: entry-level hiring in top 15 tech fell 25% from 2023 to 2024, and continues falling through 2025-2026. The reason: AI tools now do tasks previously assigned to junior employees.
If no one hires juniors, where do the seniors come from in 5 years? Yale Insights argues AI "isn't killing jobs — it's killing the path to your first job". The damage only shows up in the long term.
What does emerge
The net balance, according to the World Economic Forum, is positive: 170 million new roles by 2030. Emerging ones:
AI/ML Engineers: most sought-after role in 2026. Premium salaries.
Prompt Engineers: category that barely existed in 2023, now with $200K+ salaries in the US.
AI Trust and Safety Specialists: compliance, ethics, alignment.
AI Agent Owners: 56% of enterprises already have this role — the person responsible for agents working well.
Specialized healthcare: nursing, physical therapy, mental health.
Skilled trades: electricians, plumbers, welders. AI doesn't (yet) compete with complex physical work.
Creative direction: not designer executors, but those who direct creativity.
The management bias
HBR published a critical analysis in January 2026: "Companies are laying off workers because of AI's potential — not its performance". The thesis: many managers are doing layoffs anticipating capabilities AI doesn't yet have, which will later generate operational problems.
BCG's optimistic view
BCG published a counterpoint in April 2026: "AI Will Reshape More Jobs Than It Replaces". The thesis: most impact will be role reformulation, not elimination. The worker with AI replaces the worker without AI, but the role itself continues to exist.
Practical implications
For employees: (1) learning to use AI in your current role is the year's imperative. (2) identify what in your work is high-judgment (what survives) vs. high-repetition (what gets automated). (3) move toward upper-skill application or complex physical work if your role is in high-risk category.
For companies: (1) don't do layoffs in anticipation — wait for real performance. (2) invest in upskilling before replacing. (3) redesign roles to combine AI + human, not to eliminate humans.
Conclusion
The transition is real, the numbers are large, and it will hurt — especially for working women and Gen Z. But the net balance by 2030 is positive if companies and public policies do their job. For individuals, the recipe is the usual in technological transitions: learn, adapt, move toward where new value is being created.