Hacker News Reader

Comment by adastra22 | Hacker News Reader

We've had AGI (artificial general intelligence) probably since the first release of ChatGPT, and certainly since the first agentic harnesses. They're just finally acknowledging what the term means.

View on HN

We've had AGI since RNG! Cut the poor, unacknowledged RNG AGI some slack, will ya? It can literally solve everything when you're patient enough.

There's so much that the term includes that isn't even feasible with an LLM

Artificial. General. Intelligence. The ability to solve (even partially or even badly solve) problems drawn from arbitrary problem domains without pretraining on the specific problem class. You can pose any problem of any type using natural language to an LLM and it will attempt a solution. That's literally all the term means.

You (and the rest of the media and many industry figures) are conflating artificial super-intelligence (reference point: humans) with artificial general intelligence (reference point: specialized/narrow GOFAI).

> conflating artificial super-intelligence (reference point: humans) with artificial general intelligence (reference point: specialized/narrow GOFAI).

So now humans is "super" intelligence? it's nice to move the upper bar so that more stuff can be called "just" intelligence.

Reference class in this case means not an example but what the comparison is against. Superhuman means better than humans. General intelligence is defined without any reference to human capability levels.

How can somebody define intelligence without a reference to human capability? Humans are the one judging it.

general intelligence for beavers or a birch forest would be very different than general intelligence for humans...

Intelligence is problem solving. It can be defined in terms of optimization theory.

How would you define intelligence in terms of optimization theory? Also intelligence implies more than just problem solving, in fact it's another one of those pesky hard to define things

Which is also a uniquely human take...

I don't even know what you are arguing for.

That you pretend there's some clear definitions where there aren't, and accuse people of "conflating" these things?

I don't think we'll be able to meet, and that's ok since the definition isn't universal. I align more with Demis Hasabis' views on this

There's a simple rule: someone of higher intelligence can successfully pretend to be someone of lower intelligence, but not the other way around.

Humans can successfully pretend to be LLMs, but LLMs are still not successful at pretending to be Humans.

ASI "matches or exceeds humans in at least one field"

AGI is defined as "matches or exceeds humans in all fields"

Superintelligence is defined as "strictly exceeds humans in all fields"

I don't think your definition matches the general consensus.

Very valid point, shame it’s buried so deep in the comments’ tree.

Reply on news.ycombinator.com