Automation and the Meaning of Work
Recently, artificially intelligent (AI) art generators such as DALL-E2, Midjourney and Stable Diffusion have been making headlines in the news and artists are fuming. Given the recent news attention, I figured it would be a good idea to talk about the direction AI is headed and how that affects the meaning of work going forward.
Automation is Coming
It should be clear to everyone not living under a rock that we’re in the process of making human labor obsolete. Most people claiming that AI won’t be able to take their job are simply in denial.
If AI doesn’t eventually automate your job, that won’t be because it’s technologically impossible. There’s no job humans do that’s so complicated an AI can’t learn how to do it.
What does this imply for the meaning of work?
Basic Needs Are Still Met
Well, if people can’t meet their basic needs after their jobs are automated, they won’t care about finding meaning through work. So let’s assume that the economy adapts so those whose jobs are automated can still meet basic needs.
No Dystopian Outcomes
Second, it may be the case that an AI-driven technological singularity or some other utopian/dystopian scenario follows shortly after large numbers of jobs are automated. If so, there won’t be time enough to care about automation and its implications for the meaning of work. So let’s assume there’s a significant time gap between automation and any utopian/dystopian outcomes.
Automation of Work
Given those starting assumptions, post-automation work can be divided into four subcategories.
The first category of work is where the human prefers doing the work and the beneficiary of the work prefers a human doing it. AI beats the best human teams in League of Legends, but viewers still prefer watching humans play. Thus the interests of the human teams and the viewers are aligned.
Jobs in this category may become AI-assisted, but not taken over by AI. For example, Chess players use the Stockfish AI to improve and there are matches where different AIs face off, but that doesn’t ruin the fun of human against human matches. The fact that people are playing at a higher level than they were decades ago thanks to practicing with AI actually enhances the game.
The second category of work is where the human prefers not to do the work and the beneficiary of the work prefers AI doing it. When child labor is automated, assuming the economy adapts so basic needs can still be met post-automation, it’ll be a good thing. The children can instead spend their time getting an education or just being kids.
The third category of work is where the human prefers doing the work but the beneficiary of the work prefers AI doing it. Imagine a person who needs brain surgery. The human neurosurgeon spent twelve years studying and finds meaning through saving lives, but they’re more likely to make a mistake than the AI neurosurgeon.
If it were me, I’d feel bad for the neurosurgeon, but I’m not taking any chances with my brain. That’s assuming human neurosurgeons are still an option, which they probably won’t be since economics will force them out.
So what’s going to happen to people who spend years studying to perform skilled labor only to have that labor automated? This is why artists are unhappy with AI art. People are investing lots of time, energy, and money into learning a skill that the market won’t have any human demand for.
These would-be skilled workers need to be compensated with some economic incentive. Without an incentive, people will reason that there’s no point in studying for a job that might not exist by the time they finish their education. If that job isn’t promptly automated, there could be a massive skilled labor shortage.
As for mental health, I think at a minimum, we need readily available emotional support systems in place for people whose jobs get automated. I’ve seen more than enough people’s mental health go down the drain after losing their job and it worries me what’s going to happen to them post-automation.
The fourth category of work is where the human prefers not to do the work but the beneficiary of the work prefers a human doing it. Post-automation, there will still be people who prefer human waiters and waitresses to robots. But most waiters and waitresses would probably be doing something else with their time given the chance.
What’s probably going to happen in this scenario is the same thing that’s happened historically. There will be some initial resistance, but automation will win out. Over time, cultural norms will shift until it’s accepted that certain jobs are automated.
Post-automation, there will be more time for social interaction between people with common interests and values. At work, one isn’t necessarily meeting people with common interests and values since one doesn’t choose their coworkers.
On the other hand, maybe more free time will just lead to people spending more time on social media since better AI means more addictive online platforms. It’s hard to say for certain.
The categories I’ve laid out make up one useful model for thinking about automation and work, but it’s not the only model. For instance, the AI arms race between world powers may force many public and private sector jobs to be automated. Also, there will probably be new jobs that don’t fit into any of the categories because only AI will be capable of performing them. There will be work where a collective decision has to be made whether humans or AI should perform it because a mix of human and AI workers won’t work. In other cases, the nature of the work is such that people will value it more merely because a human did it.
Some other model for predicting the implications of automation may arise that does better than what I’ve laid out. My model may end up being totally irrelevant if the assumptions I made aren’t met. It’s certainly oversimplified and not the full story.
It’s impossible to fully systematize such complex future technology and the uncertainty can be overwhelming, but sticking our heads in the sand and ignoring automation isn’t going to help. Significant resources need to go into building models and predictions for the automated future and preparing everybody for the post-automation world because it’s coming, and soon.