[tdehnel]

As David Deutsch says, the reason is LLMs and the current batch of AI tools are the opposite of AGI.

You make an LLM by limiting what it can do. An AGI can decide what to do for itself. Maybe it wants to play chess instead of responding to your prompts.

Some people say AGI will emerge spontaneously from LLMs. This is supernatural thinking, based on a bad explanation. We will not be able to program an AGI until we understand how our own intelligence works.