I say this with deep love for 80k and mild existential dread.

I love 80,000 Hours. One of the most useful resources in the effective altruism ecosystem. Their core premise is simple and compelling: you have roughly 80,000 hours in your career, so spend them wisely.

But I've been doing some math, and I think they might need a rebrand.

See, 80,000 Hours increasingly (and rightly!) recommends AI safety and AI governance careers. Their top career paths are dominated by AI-related roles. Which makes a lot of sense... until you think about the implication.

If the AI safety community is right that transformative AI could arrive in the next 5 to 15 years, and if that AI is capable enough to automate most of the work humans currently do, then we don't have 80,000 hours. We might have, what, 10,000? Maybe 15,000 if we're lucky? ^[1]

The very careers 80k recommends are built on the premise that those careers won't exist for very long.

This isn't a criticism. It's a compliment! It means the advice is working. They've correctly identified the most important problem and pointed people at it. It's just that solving the problem also dissolves the brand.

What the rebrand could look like:

• The homepage countdown timer: "You have approximately 9,247 hours remaining. Use them wisely. Seriously. We mean it. Hurry."
• The career guide gets shorter every quarter. By 2028 it's just a single page that says "AI alignment. Go."
• The podcast episodes get progressively more urgent. Episode #400 is 4 minutes of increasingly panicked career advice.

New tagline options:

• "Max 10,000 Hours: Because we're running out"
• "Fewer Hours, Higher Stakes"
• "80,000 Hours (terms and conditions may apply)"

I joke, but for a lot of people these timelines are genuinely scary. I feel it too. I spend a lot of my time wondering what happens to our movement's work when AI reshapes everything, and the answer is: I don't know.

What I do know is that 80,000 Hours has helped thousands of people find meaningful work, including me. If it turns out we only get 10,000 hours, I'd rather spend them on something that matters. And that's basically their whole pitch.

Now get back to work. You're on the clock!

1. ^{^}

   *10,000 hours assumes roughly 5 years at full-time work before AI eats your job, which is aggressive but consistent with what several AI lab CEOs are publicly saying. If the Metaculus crowd median is right, you might get 15,000. Either way, the brand could use a refresh.