More intelligence or less?

I recently got a question from a dear friend about why I’m working with AI, because “it’s killing the planet.”

It’s a question I’ve answered informally on many occasions in the last few years but I realized I didn’t have my answer spelled out clearly here for reference yet. This is the answer I’ve had since about 2021.

“Killing the planet” is the default path. It’s the eventual outcome we’ve all been creating for a long time, and though we’ve brought a lot of awareness to this in the last 50 years or so, there’s not a lot we’ve been able to do about it. We can go faster or slower toward destruction, but not really change course because of fundamental flaws in how we’ve built up our civilization and how that impacts the planet it is built upon.

The patterns we see around us are similar to what’s happened before many times over the last few thousand years of human history, across past civilizations. The main difference now is that it’s more of a global civilization than we’ve ever had before, so they’re bigger in scale and more nuanced in understanding. But for all of history until very recently, the tools that we’ve had to change things have been limited by our own human intelligence and our ability to organize collective intelligence by way of dynasties, countries, companies, etc. We’ve done the best we could with what we had. But the cycle hasn’t really changed much. And our trajectory is still pointed toward the same default path.

Until recently, when we taught sand to think. We’ve figured out how to organize silicon atoms into probabilistic engines that we’re still learning how to drive. Yes it’s dangerous, like driving is. But now intelligence doesn’t have a finite limit of our human skulls and human relationships. Now we can supply power to create intelligence that we can use.

And while the generation of power is certainly a big problem to solve, I believe it is a very solvable problem. Two examples that inform this belief: the cost of solar has fallen roughly 90% in the last fifteen years, and the energy required per unit of AI work keeps dropping as chips and models get more efficient. Same with heat, which is turning from problem into solution in its own right and could entirely change the game on the amount of power it takes to make sand think. In solving important resource and distribution problems like these, I don’t think that more intelligence is a bad thing. To the contrary, I think it’s the only thing that offers the promise of a new path.

Now of course, we humans have made things better before too. We’ve created new vaccines that save lives, reversed much of our damage to the ozone layer, and made up countries and rights and freedoms. But those were never victories of raw intelligence alone, they were rare victories of coordination, and coordination at the planetary scale is clearly not our strong suit. This is precisely my point: if our human intelligence plus our existing ways of organizing it could have created a better path by now, they would have. With all our biological intelligences put together over time, they didn’t. We had a good run, and like it or not we’re facing some end stage challenges now.

The answer I’m working toward isn’t some faith-based devotion to AI. I don’t believe that more artificial intelligence alone is the answer, it’s a better way to coordinate all the intelligences we can access. I believe in Aggregated Intelligence. I’ve written whole papers about this, but basically it’s the combination of all the different forms of intelligence we can muster, and aligning that to create a world that we want, rather than one we don’t. That world doesn’t look like what came before, it looks like something new. I’m not in charge of what that looks like, and neither are you, but I’m excited to help contribute as I can.

So to the question about killing the planet, that’s what we were busy doing before and no it hasn’t changed yet. I agree. My question is in return, is what actually changes that? More intelligence or less? More power or less? More intent or less?

The thing that so many people miss about the future is that there are always new challenges and new problems to solve. There is no point in the future where we run out of those, that’s not the goal. The better question is, what kinds of problems do you want to be wrestling with in the future? Here are a few that I think are worth our time, and that we’ll encounter as central issues in my lifetime:

• How human is human?
• How individual is the individual?
• How alive is alive?
• How well shall we treat those that are unlike us?
• How do we best organize ourselves to make collective choices?

I don’t think we’ll resolve these questions, but I think they’ll become very important to us. More important than many of our central questions today. We still have monarchies in the world, but we don’t pay them much heed anymore because they don’t matter like they once did. The power structures and governance systems we have today will follow the same arc. They’ll just be less important tomorrow, because they don’t solve the problems that are the big ones of the day. We don’t have systems that handle Aggregated Intelligence well yet, because we haven’t needed them. We’ve been focused on different levels of problems, and building solutions to do the best with what we had. Now we have more. Now we need more. Now we are more.

Here’s a video from a few months back that might convey this too:

Bottom line: I have my answer. It may be different than yours, I’m not trying to convince you that my answer is right for you. But I hope you understand why it’s right for me, and why I care so deeply about it. I certainly do believe that caring more about our collective future is much better than caring less.