← sources

A sane take on AI and labour?

source: reddit (reddit:singularity)published 2026-10-08 23:04ingested 2026-10-09 02:37

retained text

Most jobs are not just the execution of already well-defined tasks. They also involve deciding clearly what needs to be done, whether it's been done well enough and completely enough, and if not, what needs to be done next to address this.

Having machines to execute already well-defined tasks more cheaply, quickly, and effectively, allows people to spend more time deciding what needs to be done, whether it's been done well enough, etc.

Just because a machine does something well, doesn't mean it does something good. Good is a matter of human judgement and taste.

There are indeed some jobs that are almost entirely the implementation of clearly defined tasks, and will be largely replaced with machines. Translation, transcription, building an already approved widget of some kind, etc. Although even these need some human oversight or quality control.

All of the other jobs will just get more prodcutive per human, when combined effectively with AI (including robotics controlled by AI)

For some products and services there might be a limited demand, no matter how cheap and easy the thing is to get or provide.

For the vast majority of products and services, though, cheaper and faster production just leads to people wanting more, or more customisation, or higher standards.

The pre modern-ai world (e.g. 2016) would have been unimaginably advanced, rich, and luxurious to a homo-sapien from 10,000 years ago, who hunted and gathered in a tribe of about to 200 people, and yet the homo sapiens in 2016 were not all rich, satisfied, or unemployed, despite being the same species as 10,000 years ago.

If we make an equivalent 10,000 year leap in the next 10 years, we'll quickly learn to treat it as the new normal and still want more, better, faster, etc. That's just what the typical human does: want more.

Most forms of government, especially the more democratic ones, won't tolerate unemployment levels above about 10 or 15 percent. This is simply because the politicians want to prevent their own unemployment.

Medium term employment levels are far more of a function of politics than of technology. Tax laws, grants and subsidies, employment laws, education and training, regulation and protectionism, public private partnerships. What they promote and what they punish. Technology has almost nothing to do with medium or long term employment.

So although AI may greatly disrupt the job market in the short term, and end some kinds of jobs, it won't lead to sustained mass unemployment. It also won't make us rich "enough" (there's no such thing).

The main problems for individuals to worry about are first: finding a job that will be amplified instead of replaced by AI, and second: learning to manage AI well in that job. This might include voting for politicians who seem to at least understand these problems.

document fetched 2026-10-09 00:32 · canonical https://reddit.com/r/singularity/comments/1x15keo/a_sane_take_on_ai_and_labour