Remember the pre-AI generation of tech?
Everything could be turned into a mobile app. And this shaped up what we now call the gig economy. Leveraging supply and demand through digital platforms. And so we had Uber, Glovo, Airbnb.
Now, consider that:
This did not create new jobs - nothing unimaginable before appeared out of thin air with this technological shift. But nevertheless, the nature of so many people’s work changed.
Similarly, I expect a future that is familiar but with a twist. There will be a new kind of job: the AI Trainer.
Many people will be working very similarly to the present day. But by doing so, they will be training AI models to do their job.
Just like we already see with the humanoid robots like NEO. Human operators dictate the actions of the robots. In time, training data is gathered so that at some point the robots will no longer need the human operators. At least this is the goal.
So AI Trainers will be branches of various existing industries, from construction to manufaturing and logistics.
But the same applies to office jobs - and even more so. Unlike the physical world, the digital world comes with much less constraints and much more standardization.
But surely this won’t be a career of the future. Rather it seems to be only a momentary solution. Use AI Trainers until we get the proper AIs.
But well, the process of training and refining an AI is actually a very long sequence of iterations. The performance of these systems scales logarithmically, so getting from 0 to 80% may take as much time as going from 80 to 90%.
Getting those pesky last decimals will take a lot of collective effort. Alligning the AI to human values, solving for niche edge-cases and refining for intuition. This requires lots and lots of nuances.
So the work of trainers will keep being valuable even as it becomes more focused on specifics. Alongside, others will be acting as the present day Team Leads, supervising multiple AI agents.
As work becomes more effective, hence cheaper, the demand for things to be done will increase - not stagnate (as it did with every technological innovation speeding production up)
Humans will produce marginal improvements compared to the backbone of automated work. But those small improvements’ financial value will be backed by precisely the already running AIs, trained in the previous iterations.