Benign AI is still scary!

Remember the frog sitting comfortably in the pot of cold water? Put him in cold water, heat it slowly, and he never jumps out, and soon you have frog fricasee. While this may be an urgan legend for frogs, it appears to be true for people, and at the moment we're sitting in two pots at once, one literally and one not.

The literal one is climate change. Climate is such a big, complicated, turbulent mess, all kinds of weather stirring together at the same time, that it's almost impossible to point to any one day, or even any one year, and say that was the step that mattered. So we don't. We just note that the water seems a little warmer than it used to be and get on with our lives.

In my view, the more important pot is AI/robotics, because I think it is far more dangerous and the one fewer people are watching. You don't have to sign up for any of the particular ways it supposedly ends (machines become conscious and decide the world is better off without us, or they merely simulate consciousness and reach the same conclusion, or people use them to kill other people, or they accidentally build the tools that let people kill other people, and so on). Pick one, pick none. The part that isn't up for debate is that the stuff keeps getting better, month after month, with very little attempts made to slow it down. Yes, people are upset about data centers, but not because they portend the end of human civilization as we know it, but because of the noise, water and electricity they use.

Ask most people about this and they would agree that worrying about AI doing us in is a bit like worrying that a teddy bear is going to kill us in our sleep. They look at what AI can do today, decide it's a clever toy, and discount its current abilities without ever bothering to follow the path to see where it leads. How many times have you heard pronouncements like, “AI is so stupid, it doesn’t even know what day it is today,” “it couldn’t thread a needle to save its life,” or “it will never be able to understand emotion or passion.” But, when I contemplate the future of AI, I don’t like where it leads and what bothers me most is that even in the most benign of scenarios, when none of the machines turn on us or are used by others as weapons against us, we end up in a dystopian society. It's just a chain of fairly ordinary steps.

The first step is already happening. When a machine can do something better than a person, we stop valuing at skill so highly who does it. Chess is a clean example. We used to be in awe of Grandmaster’s calculation skills, but anymore since anyone with a cell phone can play a better game. It is how happening with people who speak many languages, radiologists, programmers, lawyers, etc. The work doesn't go away. It just stops being valued so highly.

Step two is when AI and robots can do nearly all human labor cheaper and better than humans can, then the average person has nothing left to sell. I am not saying people can't raise their kids or coach Little League or take the trash out. I am saying they can't earn a living, because nothing they can do is worth paying for. Labor stops being a living.

Step three is where it gets ugly. A society full of people who can't earn has two choices and only two. It can feed, clothe, house, doctor, and entertain every resident from cradle to grave, job or no job, through some enormous all-inclusive program. Or it can let them go without. And without means the basic necessities of life become scarce and in the natural competition for scarce resources those who have nothing of value to offer will find willing the help them. Other than coming out somewhere between these two, there isn’t a third choice. Every country on earth is going to make a choice, and I wouldn't assume they all make the kind and gentle choice. Some of them can't afford to, and some of them won't want to.

Step four is what people do while they wait to find out which door they got. When the future looks hopeless, people get more extreme, because they see less downside. Nothing to lose, nothing to fear. Look at any decade with real mass unemployment and watch the fringes fill up. Now imagine an entire population, permanently, reaching the conclusion that they are unneeded dead weight on society see what boils over.

And there is another consideration. If machines do all the work, and a handful of people own the machines, then that handful decides who eats, and basically, who lives. That's a quieter catastrophe than a robot uprising. It's a dinner party with a very short guest list. It also strikes me as a good deal more likely.

So the water is already warm. And rather than taking a hard look at our situation, trying to figure out whether anybody can turn the stove down or off, or whether there's any way out of the pot at all, we have been each fighting for our own corner of it. The company’s are racing each other to the bottom to see who can develop smarter AI and better robots. The governments are arguing over who gets to hold the ladle. The rest of us are mostly telling each other the water feels fine, but we don’t like the taste of the broth. And, recall, this is in the most benign of scenarios.

Frogs may get out of pots by themselves, but people don’t. We need to come together and throw our combined resources at this, money, brains, votes, all of it, while there's still time to have some say in how it turns out. The alternative is our goose, and by our goose, I mean ourselves, gets cooked, which is something I would rather not experience.