As a short, well-intentioned, ill-advised summary for: Common Ground between AI 2027 & AI as Normal Technology

1. Before strong AGI, AI will be a normal technology.

Before I add sugar to the milk to make the milkshake, it will be milk.

2. Strong AGI developed and deployed in the near future would not be a normal technology.

The milkshake will not taste like milk, and that’s a big deal for milk. As, it’s not normal milk.

3. Most existing benchmarks will likely saturate soon.

Because we all grew up studying and measuring ourselves through standardized testing, that our mom’s and dad’s got us tutors for so we could study to be really, really good at them, this is how we should test AI. And it will be smart boys and girls like us. Who like milkshakes.

4. AIs may still regularly fail at mundane human tasks; Strong AGI may not arrive this decade.

Check that. We might not actually get milkshakes. We’ll still have to drink our milk. We are decidedly not stoked about that.

5. AI will be (at least) as big a deal as the internet.

The thing that’s already a big deal, will be a big deal. And we know that because the last time there was a big deal, it was a big deal.

6. AI alignment is unsolved.

That milk still seems awfully milk-like.

7. AIs must not make important decisions or control critical systems

We have checked the use by date, right?

8. Transparency, auditing, and reporting are beneficial.

I think the FDA was on to something.

9. Governments must build capacity to track and understand developments in the AI industry.

Everyone was crying over spilt milk and I guess they had a point.

10. Diffusion of AI into the economy is generally good.

Those advertisements were right? Milk is good for your bones.

11. A secret intelligence explosion — or anything remotely similar — would be bad, and governments should be on the lookout for it.

You see the trick about brain freeze is you keep drinking the milkshake and it makes the brain freeze go away.

Conclusion

Have we thought about trying almond drink?