aiqbench analysis
This opinion piece argues that governments or employers should guarantee compensation to workers whose fields AI automates, to encourage people to enter careers that may be automated before they pay off. The core claim, that uncertainty about which jobs AI will replace discourages training and may cost society later, is a reasonable framing. The concrete scheme (for example, five years of payments tied to predicted pay) is only sketched, and the proposal does not yet address funding or how automation timelines would be judged.
Compensating the victims of automation is economically beneficial

From LessWrong
Full textIt is likely that we see a wave of jobs automated away in the coming years. Already software developers, artists and mathematicians are standing at the precipice, and other careers are sure to follow.
Responses range through denial, anger, bargaining, depression and acceptance. Some fight to prevent automation, others have already attempted to switch careers or focus. But few suggest that they will accept their job loss so long as it comes with a decent payout.
However I believe this is likely the best option - not just for them, or just so that we can buy them off to prevent them blocking automation, but because of fundamental uncertainty about what and when different jobs will be automated.
Consider someone considering studying medicine today. After 8 years of education and 3 years of residency, they hope to be able to eventually earn decent pay and work less crazy hours. By that stage they'll be hundreds of thousands of dollars in debt. They have no idea how the world might change in 11 years and they might reasonably consider that career choice too risky, and instead start earning immediately in e.g. a physical apprenticeship.
This applies to numerous fields, from potential taxi drivers considering buying a car to, yes, maths and software engineering: we don't know whether AI will be able to fully replace these tasks or we'll still need numerous humans to manage and round out the AI.
Absent a reliable promise of compensation if these careers go AWOL we will face fewer people investing in such risky choices, which we as society are likely to regret down the line. A firm commitment that if a career is automated workers and students in the field will be compensated enough to partially[1] justify their investment and give them the time they need to retrain may be sufficient to save us.
What does such compensation look like in practice? The devil is in the details, but broadly I imagine something like guaranteed payment for 5 years according to the predicted pay trajectory of your career, subtracting from the payment 50% of salary if they successfully find another job, funded by a tax on AI usage. This obviously needs significantly more work, but a precursor to such work is an agreement that this is something we want to achieve, which I'm hoping this essay starts to do.
^We don't want them to be indifferent between their job being automated and not, because on the margin we want more people to invest in careers which are less likely to be automated; we just need try and balance those numbers with the risk that any given career might not be automated.
We don't want them to be indifferent between their job being automated and not, because on the margin we want more people to invest in careers which are less likely to be automated; we just need try and balance those numbers with the risk that any given career might not be automated.
Full text under the publisher's open licence · 498 words
Related coverage

AI Companies Are Not (Necessarily) Liable for Unintended AI Cyberattacks
LessWrong · Sep 29, 2026
Paying Creators for Training Data Can Increase AI Displacement
arXiv · Nov 17, 2026
Anthropic CEO Dario Amodei's 2025 Total Compensation Approx. $17.96 Million
36Kr · Oct 26, 2026

Bjarne Stroustrup, creator of the C++ programming language, has joined Susquehanna International Group (SIG)
Quantum Zeitgeist · Sep 15, 2026