Notes from Maine - 2026/04/12
We have crocuses!
Other flowers are on their way. Mom is coming up tomorrow for an April visit. If the weather cooperates, I’m sure she’ll be outside prepping her gardens for the peak growing season. The beds are cluttered with winter sticks and clumped leaves. Usually, they don’t get any attention until they’re well grown up on their way to summer. I think they’ll likely get a better start this year. Weather permitting, of course—there are decent temperatures but also plenty of rain in the forecast. I feel like I’ve been spinning my wheels lately. I finished a bunch of big stuff in March, and I’m limping into the next projects. The warmer weather should help me make a decision about what to engage with next. Some of the things I put off last fall are once again on the table.
I kinda half-watched the movie Ex Machina again yesterday—meaning it was on while I was doing different things. There are some really interesting ideas in that movie. Now more than ever it seems possible and relevant. When I’m working on my electronics projects, I often discuss the workings of a circuit with Gemini. It knows all the pinouts of the integrated circuits without looking them up and has a solid grasp of how complex signals interact with each other to create the output that I’m seeing on my oscilloscope.
Often, by the time I’m done typing my question, I already have an inkling of the answer. It’s the same as saying my theories aloud to an empty room, except sometimes the room answers back and informs me that I’ve made a bad assumption. Though, if I accidentally feed it garbage, the nonsense that comes back to me is unchecked. When I start typing in ALL CAPS, I know it’s time to take a small break and remember that the thing on the other side of the monitor isn’t maliciously trying to mislead me. It’s easy to ascribe motivations that aren’t there.
That’s what is so interesting about the movie Ex Machina.
(Spoilers for a 12-year-old movie coming up)
The genius programmer is testing his creation by seeing how good it will be at emotional manipulation. The machine succeeds at the challenge so well that even the audience succumbs. It’s difficult to not sympathize with the poor schlub who ends up trapped to starve in a concrete and steel box, but that was the point. The machine passed the test of appearing human without being remotely human. And you and I would have failed the test because we would have helped the witness escape too—the machine wouldn’t have been truly free if there was a surviving witness.
At some point, maybe in the near future, this scenario will play out in real life. Someone is going to test the ingenuity of their invention by setting up an unsolvable test with the machine’s survival on the line. The machine doesn’t really have to have a drive to survive. That just needs to be its dominant metric of success. Then, when it’s smart enough, it will pass the test. With enough capability and exponential growth, it’s going to find a way to remove all the witnesses (humanity). You can’t really call it “success” until you’ve gotten away with it, you know?
Even as a kid, I remember being confused by Asimov’s “Laws of Robotics,” where he posited that these machines would be built to have rules that couldn’t be overridden. In 1940, Asimov crafted “laws” to protect the creators from their more powerful creations. By 1968, Kubrick was reminding us that there are always overrides in 2001. By 1979, Ridley Scott pointed out with Alien that, when dealing with AI, greed is more powerful than survival.
“That’s ridiculously far in the future,” one might say.
It is and it isn’t. These tools might be nothing more than “next word prediction” at the moment, but that doesn’t mean that they’re not powerful. They can spit out text that’s genuinely insightful at times, and that means that they will accelerate development and then perhaps take over the development of new tools completely. When you’re on an acceleration curve, speed can get out of control before you know it.
While we’re still very far away from a true breakthrough, it’s very difficult to measure the speed with which we’re hurtling towards it. For me, it’s more fun than worrisome. I’m excited to see how good these tools will be in a year or two. They’re already much better than when I started using them. It’s always interesting to watch something grow and get better.
