I sporadically listen to the Moonshots podcast. It’s always interesting, some episodes more so than others. But this episode struck me as more significant than most, and worth sharing:
The host, Peter Diamandis, sometimes annoys me and I don’t know why. There’s no real reason for it. He seems a just fine person in most every way, so I guess it’s a style issue and on me. The other 3 standard guests he always has are fun to listen to; experts each and every one in various fields, mostly software and venture, they discuss the latest breakthroughs of the week. These days that’s going to result in a 2-hour podcast even if you’re trying to keep it concise.
In this episode they make the case that even as Artificial General Intelligence (AGI) has several definitions, depending on who you ask, we are in it. At least somewhere in the bell curve distribution of those definitions, and probably not just barely (my view).
Further, we are officially in the Singularity — that moment in AI advancement where exponential change goes vertical. It just oddly feels mostly normal because we’re going about our daily lives, taking care of business, rolling with the changes. Meanwhile multiple revolutionary breakthroughs are found via AI pretty much on a daily hourly basis across nearly every discipline. It will appear as a Step Change in history, when future generations look back.
A couple of paraphrased snippets from the podcast:
If you fold a sheet of paper in half, doubling its thickness, and then do that 49 more times, the thickness is roughly equal to the distance between the Earth and Sun.
Intelligence and Compression are the same thing.
Exponential change. Humans weren’t really built to deal with it. I don’t know what anyone with young kids can tell them right now, regarding the next several years to come, but today’s adolescents may find they live in a shockingly changed world by the time their voting years hit. It’s going to be a challenge, but I suspect they will adapt better than my older generation will.
Back on AGI, and whether or not we currently possess such a thing. The trigger for me was when Google’s DeepMind effectively solved protein folding via Alphafold. Shortly after, I found myself with a real-time, on demand, encyclopedia-professor of all topics in my pocket. One that I could converse with in natural human language, and then receive a logical response in kind.
What is it like to wonder what the answer to something is? We barely find ourselves in that situation these days. This seems to me to be AGI. Some others have a higher standard, but will we arrive there in 6 months? A year? Probably. Or maybe I’m wrong and it takes 5 years. In the expanse of space-time these are all the same moment.
All math is likely solved in the next few years, then all biosciences right after. Big ramifications. Putting the results of this into beneficial human practices may take a while, but what a time to be alive.
There are many issues to be addressed. What is the human role in science going forward? What even is money in a world where robotic intelligence is fully in charge of production? Do we completely discard the system that has worked mostly well for centuries? Will humans feel a loss of meaning? Drift toward the Arts? Will AI compete with them there? Will it lead to depression among the masses? Or will the gift of a slower and more present life be a welcome destination for most?
Just a few of many questions to be answered by history. Clearly there is still space to wonder.