Interesting Times

Photo by Pavel Danilyuk on

Everyone’s familiar with the supposed Chinese curse “May you live in interesting times.” Probably not Chinese. The closest actual equivalent source is from a writer named Feng Menglong in 1627: “Better to be a dog in times of tranquility than a human in times of chaos.” For my part I’m not sure if it rises to the level of a curse in the first place. More like the normal human condition. I mean, seriously, has there ever been a period in human history that hasn’t been “interesting” in its own way? As a species we’re a lot of things but “boring” isn’t one of them, and I can certainly see where Feng was coming from.

Frankly, I could do with a bit of boredom right now.

Not gonna happen. Latest developments in AI research just got real. By now AI language models like GPT-3 that can converse on a near-human level are widespread. Another really interesting development is scientists have plotted AI Language models’ responses to language interpretation and found the graphs are eerily similar to the same responses in the human brain. Google has apparently taken the next step by integrating their own language model into a robot powered by a custom AI brain. Frankly, it’s the next logical step and I’m a little surprised it took this long. Google might be the first but it won’t be the last.

So what’s the point? Simple: instead of robots either remotely controlled or specifically programmed for a narrow range of tasks, now there is a platform where the robot can “understand” a written command and attempt to carry it out. This is different from other platforms like “Sylvia” which at heart are just very advanced chatbots.This new type of robot, in addition to processing language, can analyze its physical environment and interact with it in response to how it interprets what the language model tells it. Naturally, it’s still learning and doesn’t always successfully complete its task…the first time. Note the “learning” aspect. Also, there isn’t just one robot, there are several. And if one learns how to successfully complete a task, they ALL do.

One take on the concept of AGI (Artificial General Intelligence) is it won’t happen until AI is out of the computer, so to speak, and able to truly interact with its environment the same way human children learn. Thus, robots. It’s happening now. Will this lead to a truly general AI? No one knows for certain, but that’s the goal.

Interesting times.