Discussion about this post

User's avatar
Marginal Gains's avatar

I had a discussion on the same topic of AI taking jobs away with one of the most intelligent people I know and an original thinker earlier today, and here is our conclusion:

1. Implementation is where the rubber meets the road.

There’s always a gap between theoretical capabilities and practical implementation. While AI may seem powerful in controlled environments, the real world is full of edge cases—unpredictable scenarios that systems weren't designed to handle. These edge cases often reveal the limitations of even the most advanced technologies.

For example:

- Autonomous vehicles: Despite their impressive capabilities in ideal conditions, challenges like lousy weather, unpredictable pedestrians, and local traffic laws have slowed their rollout.

- Healthcare AI: Models can diagnose diseases with incredible accuracy, but implementing them in hospitals requires regulatory approval, integration with existing systems, and trust from doctors and patients.

Ultimately, implementation is where the actual test of AI’s impact will occur. It’s not just about what AI can do but whether it can deliver consistent, reliable results in the messy realities of the world.

2. Capable by 2034, but not necessarily sufficient.

The distinction between capability and utility is critical. By 2034, AI will almost certainly be capable of automating many tasks. However, just because it can do something doesn’t mean it will fully replace human roles or meet all needs.

For instance:

- AI might automate 95% of a job, but the remaining 5%—often involving creativity, ethical judgment, or empathy—could still require human involvement.

Broader systemic issues, like regulatory frameworks, public trust, and ethical alignment, might not be ready for AI to fully take over.

Even with advanced capabilities, we’ll likely see a hybrid model where AI handles repetitive or technical tasks while humans oversee and address the gaps it can’t fill.

3. Will it reach everyone? Probably not.

This reflects the reality that technological progress is rarely, if ever, evenly distributed. William Gibson’s famous quote—“The future is already here; it’s just not evenly distributed”—is a perfect fit.

We’ve seen this pattern with countless technologies:

- The internet revolutionized communication, but millions still lack reliable access.

- Advanced healthcare tools exist but often remain out of reach for low-income or rural populations.

Similarly, AI adoption will vary widely. Wealthier nations, industries, and companies will likely benefit first, while less developed regions or sectors may lag. Cultural, economic, and political factors will also shape how widely AI spreads. For many, the question may not be whether AI is capable but whether they’ll have access to it at all.

4. Human behavior and incentives drive innovation.

Humans are remarkably adaptable, and when faced with constraints, they often innovate unexpectedly. Incentives play a huge role in shaping outcomes. As Charlie Munger said, “Show me the incentive, and I’ll show you the outcome.”

AI will likely follow a similar pattern. While it will undoubtedly disrupt many roles, it will also create opportunities—especially if the right incentives encourage humans to collaborate with AI instead of competing against it.

I have some more thoughts. I will write them later.

Expand full comment
Sam Waters's avatar

Interesting post. I will have to think on how to be less mediocre, how to transcend being merely average.

With that said, I question whether any educational system could be designed that equips most (or even a large minority) of people to adapt to AI. The last few decades have seen repeated refrains for better education as a solution to labour market dislocation. Do you remember when newspapers were talking about teaching miners or truckers to code? At least for older workers, I’ve seen very little in the way of encouraging evidence that would suggest reskilling is possible. And even for younger workers the prospects seem somewhat gloomy. Sure, if one is within say the top 5-10% of intellectual ability and conscientiousness then education might work (though I think even this might is too generous a threshold). But for everybody else? I don’t know. I worry for myself in this regard because I doubt I’d cross the minimum threshold.

Expand full comment
39 more comments...

No posts