AI and jobs

A lot of the rhetoric around AI* is wildly off the mark.

When something becomes more efficient or cheaper, people want more of it. This is a law of economic nature.

Generative AI is making grunt-workers more efficient. There will be more people doing more of it. Now that isn’t an unequivocable good, as the form of this could be more paperwork since one paperwork-worker can do more. It does mean that a lot of paperwork that people want done will get assigned, and some of the restraining influences on the spread of paperwork before will be overcome.

Consider your TPS reports. You hate them. I hate them. The managers who assign them do understand this, mostly, and most of them are trying to reduce TPS reports. TPS reporting is expensive, takes time, and all that, so there’s a restraining force.

But people still want TPS reports, so when a manager discovers one admin-staffer can do twice the TPS-reporting, those TPS reports will be assigned. Unclogged, the TPS reporting will flow. There will be TPS reports everywhere. And efficiency is never as clean as one imagines, so more admin-staffers will be hired to do TPS reports with generative AI TPS reporting software, and the world will become a brown, stinkier place. But there will be more admin-staff jobs.

There will also be people who make generative AI models, tune them from TPS reports, and provide tech support. The admin-staffers will not be fixing their models.

There will be more jobs elsewhere too. Data labelling is going to be a big thing, and people need to do it. All these systems run on hardware. Hardware in the cloud is still hardware; it’s just hardware somewhere else. So that hardware will need sysadmins, manufacturers, supply chains, and developers.

If you want a paradigm, imagine the transition from horses to cars. Transportation became more efficient. Before mass-market cars, only a few people had horses. Vastly more people now have cars. Sure, a very few cars, fewer than horses, would provide the transportation needs for the few people who had horses, but that isn’t what happened.

We’re staring into a future of more jobs, more employment, more TPS reports documenting that employment, and generative AI.

*That name, AI, is wildly off the mark too, but I haven’t found a better one. Replace it with ‘computers doing math really quickly’ whenever you see it.