Don't Look Up

My friend asked me "How should a nontechnical person like a cashier or a waitress think about machine learning?"

A nontechnical person has better things to worry about than machine learning.

There are three ways machine learning can directly impact you.


There are many threats to humanity: runaway greenhouse effect, global pandemic, World War III, resource depletion, ecological collapse, authoritarian governments and (longterm) population explosion.

Some threats (especiallly pandemics) are well-understood by scientists. There is a consensus about what we (as a civilization) should do about them. It's just a question of political will. All we have to do to keep oceanic fish populations from collapsing is to (as a species) fish less. If we fished less then there would be more fish to go around and there would be more fish for everybody. Overfishing is a pure coordination problem.

Other threats are not well-enough understood for there to be a scientific consensus. Artificial intelligence belongs to this latter category. All the smartest people I know of believe machine learning will be important, but we disagree about how things might unfold.

There are few things a nontechnical person can do about artificial intelligence. You could advocate for regulation (or the lack thereof) and/or advocate for government research subsidies. But the future of artificial intelligence is uncertain. I think a whole bunch of non-experts turning the debate political would be unlikely to improve the outcome for human civilization. At least for now.

Another option is to donate money to fund research related to AI safety. Donating money to AI safety is unlikely to do harm but it's hard for a nontechnical person to figure out which groups are actually worth funding. I recommend nontechnical people donate instead to something like poverty alleviation or global health with legible metrics of which organizations are doing a good job. Let the crypto millionaires donate to AI safety.

There are many global threats where the best course of action is crystal clear. If you want to advocate for government policies that help science improve outcomes on a global scale there are many things you can do which have nothing to do with machine learning. You could advocate for a carbon tax or gain-of-function regulation. You could help cut away some of the red tape preventing medicine from getting invented and distributed. If you've been reading Zvi's real-time history of the COVID-19 epidemic then you can identify lots of places medicine and healthcare could be improved (such as by legalizing challenge trials). That's just off the top of my head. There are countless nontechnical ways to improve healthcare alone.

There are nontechnical ways to help humanity's chances of solving non-health-related problems too. I bet growing connections between the populations of the USA, Russia and China could reduce the odds of World War III. Leave superintelligence to tech nerds. There is plenty of work to go around.

If a rogue superintelligence gets loose the game is over and we have lost. There is nothing you can do about it. When setting up a personal disaster preparedness plan, you should prepare for the disasters you can do something about. You can prepare for an earthquake or a cyberattack on civilian infrastructure by storing at least three days' supply of food and water. Ignore the planet-killing meteors. Don't look up except to plan your end-of-the-world orgy.


If you're working as a cashier or a waitress then you don't need to worry that machine learning might make your skills obsolete. You're already doing unskilled labor. Technology has already rendered your skills obsolete. It's the lawyers, quants and truck drivers who ought to fear machine learning.

Exciting New Technologies

If you're working as a cashier or a waitress then disruptive technologies are your ally. You can only go up the economic ladder. When a new technology comes out, try to figure out if there's a way to profit off of it. Pay attention to stuff that all your friends are using but which the elites (including nerds) dismiss as undignified. Pay special attention to new media platforms. Remember Warren Buffet's Rule #1: "Don't lose money".