Have you heard that every new technology startup is an embryonic AGI? BitOoda. Levelset. Pyka. RareCyte. You've probably never heard of them; I hadn't until a minute ago. I know nothing about them except that they're each an AGI play. Just like all the other tech startups.
AGI doesn't have to be as mysterious and worrisome as we often imagine it to be. There's a view of AGI that can help us feel more comfortable about our own place in what looks to be an AGI-filled future.
If tech startups are the embryos, then every corporation that is digitizing its core systems is a growing AGI. Apple. AT&T. Bed Bath and Beyond. Citibank. Caterpillar. Google. Huawei. Reddit. Walmart. The big ones are all progressing nicely toward AGI. Some, like Amazon, Google, and Microsoft, are on the fast track.
Between the tech startups and the ever-digitizing corporations, we're making thousands of simultaneous bets on AGI. Every startup is a new chance to win the ultimate prize. Every increasingly-digital business is cranking up their chances.
We tend to think of AGI as a computer that's vastly more intelligent than a human. It's easy to envision computers with minds of their own running amok, taking over the Earth and ruthlessly locking us into servitude, Matrix-style.
I believe it's vastly more useful (and hopeful) to view our quest for AGI as the incremental replacement of human intelligence with digital systems that have the same capabilities. If the essence of human intelligence is our ability to subtly alter the choices of other humans to favor our interests instead of someone else's, then that is also the essence of any AGI.
From this view, the first self-contained AGI will be the final result of us digitizing ourselves out of one of our existing systems. For example, it might be a few decades until McDonald's becomes a fully digital self-sustaining system, but we can already see that they're moving in that direction. They may already be between 2% and 20% of the way to a fully digital system, and thus that close to the first AGI. We see the same evolutionary trend with Amazon, Google, and Goldman Sachs.
Human intelligence, or real general intelligence, was not the goal of evolution. Life evolved the same way as ever for a couple of billion years and eventually we happened. In the same sense, to achieve AGI we need only continue to digitize ourselves out of our systems and organizations. But we have to keep the ball rolling, and that might be the hard part.
It's important to keep in mind that the core purpose of any AGI is to alter the future choices of other intelligent beings in its favor. If the AGI can't do that, we won't want it. If it doesn't grow in that direction, we'll kill it off. If it doesn't do that better than other AGIs, we'll discard it. And if it doesn't survive, it will die.
Because we humans can't survive without the cooperation of other people, altering or reinforcing the behavior of other people is the only thing that truly matters to us. Then, if we want to thrive, we build or join organizations like corporations, religions, schools of philosophy and so on that give us an advantage over people who don't have them. AGI is simply the ideal version of the behavioral influence systems we've been growing all along, now within reach because of the Internet Age.
A computer program that can do amazing math, solve puzzles, or beat professionals at Go or Chess has no value unless it inspires a fanbase, influences government policy, or motivates the next gaggle of investors. An AGI that can't influence any better than a skilled salesperson is not yet useful. An AGI that causes people to hate it, fight it, and want to destroy it will either dominate or (more likely) become extinct. We shall ruthlessly discard all AGIs that fail to integrate into society in a sustainable way.
The result of this evolutionary process is an AGI that will be uncannily adept at influencing our daily choices and our attitudes to favor its survival over any other being's survival. It will be an extraordinarily skilled communicator that persuades as many people as possible to make the kind of choices that sustain it far into the future. We might be persuaded by the AGI to buy specific products that fund or fuel it, say specific phrases to specific people so that its reach expands, or engage in behaviors that reinforce our affinity to it.
In short, the winning AGIs will be both powerful and invisible. The lottery may have already been won, but you'll only know for sure when you grow an AGI of your own.
So what's your tech startup idea?