Nice analysis. I don’t think you’re wrong in the least. I’ve been wholly unimpressed with generative AI so far, and you’re correct in pointing out that all it’s done is increase the spam of shoddy content on the internet. While annoying as that is, it’s hardly apocalyptic.

Where I do see AI getting better is not in the LLM’s but in the artificial neural networks (ANN) that underlie the training algorithms. Once folks outside of academic research start wrapping their minds around the potential of ANN’s beyond training algorithms and begin using them for real-time processing of data, that’s where the danger lies, in my opinion. That’s when machines will actually start processing information like organic brains. We know how dangerous, unpredictable, and beautiful that could be, because we deal with real-time processing neural networks anytime we deal with another human, or dog, or rattlesnake, or bear.

ChatGPT is nowhere near that level of processing yet. We barely understand how biological neural networks work, so I’m not too worried about ANN’s just yet. It’s only a matter of time and political and economic will to fund research, however. Until then, I’m reserving my fears and/or adulations.