Okay, it all makes sense now.

At the start of my AI master’s at Cambridge, we spent an entire semester talking surprisingly little about LLMs and a whole lot about ethics. Kant’s deontology. Aristotle’s virtue ethics. Questions about morality, responsibility, character, and what it actually means to make a “good” decision.

At the time, part of me kept thinking, aren’t we supposed to be learning about AI? I flew across the Atlantic and paid $60K+ to sit around discussing an ancient Greek?

Apparently, Cambridge was secretly preparing us for Silicon Valley.

AI companies are now hiring philosophers. I can even proudly say I studied under one who is now at Google DeepMind. And it’s not just GDM. Major AI labs are increasingly snapping up philosophers and building teams to wrestle with questions around human values, justice, alignment, and, increasingly, what we actually want these systems to become.

There’s something wonderfully ironic about all of this.

For decades, we told young people the future belonged to STEM. Learn to code. Study engineering. Build technical skills. Humanities degrees? Very nice, but have you considered becoming employable?

So naturally, we built machines that can code, and now Silicon Valley needs someone to explain Aristotle.

And the funny thing is, it actually makes sense.

For most of computing history, the hard question was whether we could get a machine to do something. Can it calculate this? Recognize that? Translate this? Drive a car?

AI introduces a much stranger question.

What should it do?

Suddenly, questions that spent centuries safely confined to philosophy seminars have wandered into product meetings. When should an AI refuse an instruction? What counts as harmful? What does fairness mean when reasonable people disagree about what is fair? Whose values should a model reflect when there is no universally agreed-upon set of values to begin with?

And as these systems become more capable, the questions only get weirder. If an AI behaves as if it has preferences or a personality, how much should we read into that? If someday systems make convincing claims about consciousness, who exactly gets the delightful job of deciding whether they mean it?

Good news, philosophy majors. Your moment has arrived.

These are partly engineering problems, of course. But they’re also questions humans have been arguing about for a few thousand years, with what I would generously describe as mixed results.

That’s the crazy part I find fascinating. The more capable AI becomes, the less its hardest problems seem to be purely technical, and the more they start looking remarkably human.

Maybe the most important skill in the AI age won’t be knowing all the answers.

Maybe it will be knowing which questions are worth asking.

Roll credits.