How long are we going to have such a narrow road to AGI?

We keep seeing the same big developments in AGI research, and the same such like breakthroughs across years.<br><br>Let's just look at ChatGPT for a moment: This is a result of transformer research into large language models, and transformer research is tied to the idea of self-attention in neural networks.<br><br>These ideas came out in the mid-2010s. That's an easy one to remember, because the transformer was first developed in 2017, and the 2010s are just a decade.<br><br>If we look at many other successful neural network models and ideas, they also all kind of originate in the same timeframe.<br><br>Now, as of now, we're in 2024, which is SEVEN years after the transformer. And these ideas, and the ideas that were developed in the 2010s, are still the basis of many modern AI systems.<br><br>Now, I'm not saying that people don't research what we have, and don't try to develop the ideas further. But what I'm trying to say is that it feels like we're still going along a very narrow road with AGI research.<br><br>But that's a weird thing to happen. I mean the amount of money put into AI research has taken off since the launch of ChatGPT. Why does it look like not much work is being done in the direction of exploring side-paths, such as other neural network architectures; or other types of machine learning algorithms; or even other, completely different types of machine learning that don't rely on neural networks;<br><br>It feels like something big is missing, but we just don't see it yet. I mean think about it. How many big breakthroughs have we seen happen? I mean real big ones, not just the big tech companies launching their own versions of ChatGPT or Perseverance or Gemini, or things like this. I mean real big breakthroughs in new ideas, or big jumps forward in existing things.<br><br>In the end, it feels like we're just trudging along, and haven't figured out that there are other things to explore. It's a weird thing. And it doesn't feel like it's going to change anytime soon.<br><br>Anyways, thanks for reading this.