[_heimdall]

A good start would have been them not calling this artificial intelligence at all. The hype is largely based on the term "AI" and if it really is simply a (very impressive) auto complete tool it isn't intelligence at all, though as you said can be a very useful tool.

[azan_]

It's amazing that there are people who still believe in "it's just autocomplete". It hasn't been true for a long time, but currently it's position that reveals complete lack of awareness how good AI has become. It has solved Erdos problem using novel approach. Constant moving of goalpost is recurring theme when discussing AI, but it's really impossible to move it so far that you can frame this achievement as "impressive autocomplete".

[_heimdall]

I would draw a distinction here. If its a tool (as the GP proposed), it is just fancy auto complete. If it isn't a tool and is instead solving problems in novel ways, inventing new things, etc then it is intelligence and not a tool.

It can't be both in my opinion. To be a tool it needs to be controllable and predictable, intelligences are neither. See humans, and really all living things, for plenty of examples where they can't be completely controlled or predicted.

[maplethorpe]

I mean, tokens are passed as input to a model, which then outputs the next most-likely token. At the heart of it, that's the technology right? Why is it so silly to call that autocomplete? Because it's capable of impressive things?

[azan_]

> Because it's capable of impressive things?

Precisely. Calling it autocomplete when it's capable of completing tasks that have nothing to do with autocomplete is silly. If you want to be consistent with your terminology, you'd have to call any stochastic process "autocomplete". What makes it double silly is that you can't really exclude that human intelligence is a stochastic process.

[_heimdall]

You're making an assumption that there is no difference between intelligence and auto complete with sufficient resources and learned patterns to complete tasks a human might do.

There may not be a difference there, I don't know but I wouldn't assume that intelligence is nothing more than sufficiently complex auto complete.

[azan_]

I'm not making that assumption. Specifically - I'm not making any assumption about nature of human intelligence, including not making assumption that it's not stochastic process. You exclude possibility that it is stochastic process without any good reason for it (wanting to call AI complex auto complete while keeping human intelligence completely safe from that label really is not good reason).

[_heimdall]

I said nothing of human intelligence either though, only intelligence.

I'm not excluding that what we consider intelligence isn't equivalent to autocomplete. Go back and read my last comment, I explicitly left the door open for those two being functionally the same. I was only pointing out that you seemed to be assuming intelligence is fancy auto complete rather than it could be fancy auto complete.

[maplethorpe]

> If you want to be consistent with your terminology, you'd have to call any stochastic process "autocomplete".

I'm fine with calling any process "autocomplete" if it takes language as input and returns predicted language as output.

I don't feel any need to broaden the definition to include things that have nothing to do with language.

[_heimdall]

I don't actually think its silly to call it auto complete.

Personally I could see it being either one. The LLM companies have drastically underfunded projects for things like interoperability. As long as inference is a black box we don't know whether its text prediction as a fancy tool or if something crazier has emerged that could be considered intelligent, self aware, conscious, etc. The former is easily assumed by the architecture, the latter seem far fetched but we simply can't know.

[danaris]

But that's their whole pitch: Altman is, last I heard, still insisting that they're going to have an AGI—in the sense of a "strong AI", capable of ushering in the supposed Singularity—by the end of the decade.

To be clear, I completely agree that we'd be better off as a society if they referred to all the LLMs as LLMs, and not as AI, but that's completely antithetical to their intentions and beliefs.

[_heimdall]

OpenAI's definition for AGI is entire bullshit though. They define it as an AI that can economically outcompete most humans at most tasks. They also claim to be concerned with safety.

Economics is a study of the past, we won't know what an AI can do economically until it is already released and allowed to directly compete with human labor. There's no safety in such an approach.

This is a bit ranty and not directed at you, to be clear. I just have no patience for how the LLM industry throws around terms at this point, especially OpenAI and Altman.

[danaris]

Indeed—and this is, itself, another part of the problem, because (again, last I heard) Altman himself was very clearly pitching the "AGI" they were going to create as something that was going to revolutionize the world practically overnight—create effectively infinite value. In other words, it would usher in the Singularity.

But an LLM that's able to "economically outcompete most humans at most tasks" (which, IMO, is likely still beyond their potential capability) is not that, and will never be that. They're just trying to have their cake and eat it too by moving the goalposts to 5 yards from the start point and claiming that they're still at the other end of the field. (Not to torturously mix my metaphors or anything.)