AI and society /
There is no career plan for AGI
Most career advice about AI collapses the moment you take AGI seriously.
Learn to code. Move into strategy. Retrain in cybersecurity. Become better at using AI.
These may be sensible responses to better software. They are not a response to a system that can learn and perform almost any computer-based job faster and more cheaply than a person.
If AGI arrives in that form, there is no clever career move.
You do not outsmart the thing whose defining feature is that it can outthink you.
AGI means the whole loop
People often imagine AGI as a much better chatbot. You give it a task, it returns an impressive answer and a human remains in charge of the real work.
That is still a tool.
A general digital worker would take the whole loop. Give it access to the company's systems and it could gather the data, interview the relevant people, find patterns, propose a plan, execute the approved actions, inspect the results and adjust continuously.
Consider supply-chain planning. The current model is a team of people operating expensive software. They collect forecasts, reconcile inventory, chase suppliers, model constraints and decide what to order.
The AGI version is not a cheaper planning application that an employee learns to operate. It is the planner. The interface matters less because the system produces the outcome.
The same logic applies across marketing, finance, design, analysis, support, software and operations. Much of computer-based work is a repeated cycle of observing information, making a decision and acting through software. A capable enough agent does not need to automate those tasks one at a time. It can absorb the job.
Retraining assumes somewhere is safe
The standard answer to technological displacement is retraining.
That works when one category of work declines while another still needs people. A factory closes, so workers learn a trade. Administrative work shrinks, so people move into technology.
AGI challenges that sequence because the next digital job may be exposed too. By the time someone completes a multi-year qualification, the system may have learnt that profession as well.
Physical work could hold out for longer. Bodies remain expensive, variable and difficult to deploy in the real world. Even there, a large migration of displaced office workers would increase labour supply and put pressure on wages before capable robotics arrives.
None of this proves AGI will arrive soon. Nobody knows that. It does mean 'learn a new laptop job' is not a serious plan for the scenario people claim to be discussing.
This is a distribution problem
If AI remains a useful tool, individual adaptation still matters. Learn it. Use it. Get closer to important decisions and real outcomes.
If AI becomes a broadly superior substitute for human cognitive labour, personal productivity advice stops being enough.
The central questions become who owns the systems, who receives the gains and how people access housing, food, healthcare and dignity when employment is no longer the main way income gets distributed.
That is not a problem an individual worker can solve by collecting another certificate.
Our institutions still treat paid employment as the normal bridge between a person and economic security. AGI would attack that bridge from both sides: labour becomes less necessary while intelligence becomes more abundant.
Telling everyone to retrain is comforting because it keeps the responsibility personal. It lets governments and companies avoid the harder design problem.
What can a person do now?
Build optionality while work still has value. Reduce brittle financial commitments. Own assets where you can. Learn how capable systems work. Build strong relationships, practical skills and proximity to the physical world.
These are sensible moves under uncertainty. They are not a secret escape route from superintelligence.
Nobody has one.
The rule
Prepare for AI as though it will keep improving.
Do not pretend you have a career strategy for AGI. No worker can retrain their way out of a system capable of learning the next job first.