> We're nowhere close to AGI and don't have a clue how to get there.
You have to have a clue about where it is to know that we are nowhere close.
> isn't impressing anyone.
I'm very impressed. Gobsmacked even. Bill Gates just called AI "the biggest technical thing ever in my lifetime." And it isn't just Bill and me.
In unrelated news, Bill has something like $40 billion in MSFT stock. If he craps on AI, he craps on MSFT and thus himself and his foundation.
In the n-dimensional solution space of all potential approaches (known and unknown) to building a true human equivalent AGI, what are the odds that current LLMs are even directionally correct?
We live on a planet with 7 billion other AGIs we can talk to. A lot more that we can't.
Our best efforts substantially underperform dealing with reality compared to a house cat.
Which is actually much more the source of my skepticism: regardless of how good an AI in a data center is, it's got precious few actual useful effectors in reality. Every impressive humanoid robot you see is built by technicians hand connecting wiring looms.
You could do a lot of damage by messing with all the computers...and promptly all the computers and data centers would stop working.
Right, and these GI's your talking about haven't changed significantly in the last 50,000 years. Most of the advancements in the last 10,000 years with these GIs have been just better communication between units and writing things down, rather than with the software itself.
You're complaining about something just a few years old and petty amazing for it's age, versus something at the tail end of 4 billion years.
> We live on a planet with 7 billion other AGIs we can talk to.
I rather see the value in having discussions with an AI chatbot rather in the fact that I can discuss with it about topics that hardly any human would want to discuss with me.