How NOT to lose your job to AI (article by Benjamin Todd)
- AI capabilities in coding, visual generation, driving, and medical diagnosis are already operational, with rapid improvements expected over the next five years and 10% automation of tasks projected as soon as next year under Epoch AI's default assumptions.
- Mass automation could eventually eliminate 100% of work tasks, potentially driving wages below subsistence levels if the supply of digital labor expands massively and physical robotics catches up, though a scenario where only 1% of human tasks remain could see wages increase indefinitely.
- Wages are predicted to follow a U-shaped trajectory, rising tenfold in the short term due to productivity gains and wealth generation before potentially plunging in the late 2030s once final human bottlenecks are removed.
- White-collar employment in the $100,000 to $200,000 income range (70th to 90th percentile) faces the highest immediate risk, with entry-level routine positions in finance, tech, law, and administration likely shrinking in favor of managers overseeing AI agents.
- Jobs requiring physical presence, such as those in construction, nursing, teaching, and policing, will likely remain unaffected for decades until robotics advances, with their incomes expected to roughly track GDP growth.
- The job market will likely become more top-heavy with an expanded class of human managers responsible for directing AI agents, while routine tasks within organizations are consolidated into fewer human roles.
- High-value skills involving directing AI, managing human-AI teams, strategic decision-making, taste selection, and complex problem-solving are becoming increasingly critical, while rote skills like recall, translation, and routine writing face immediate obsolescence.
- Specific sectors show diverging trends: software development employment was flat in 2024 with surpassing human capability expected within five years; special effects and animation saw massive layoffs in 2024; healthcare analysis and admin are at risk while physical treatment roles may grow; and legal research is automating while senior decision-making remains human-centric.
- Economic output could potentially increase between two and ten times if remote work is fully automated, with non-remote task wages theoretically rising by a similar factor, though this depends on overcoming bottlenecks in computing power, robotics data, and legal liability.
- Risks include a potential collapse of entry-level training pipelines as AI reduces the value of junior roles, leading to a need for accelerated retraining, and the possibility that economic models drive wages down if full automation occurs without new economic structures.
- Future growth is anticipated in AI hardware, cyber security, power generation, robotics, and AI development, while sectors like government may see slow adoption due to institutional inertia despite increasing demand for policy roles.
- Individuals are advised to continually adapt to shifting bottlenecks, prioritize leadership and communication skills, and consider that while finishing college remains beneficial, dropping out may be more viable if access to AI tools is restricted or if paid work opportunities exist.
- Historical analogies such as the ATM automation of bank clerks suggest that employment may initially rise due to increased branch profitability before declining with more dramatic automation like online banking, though current trends for secretaries and telemarketers already show sharp declines.