Skip to main content


So if one tries to take Sam Altman's statements that we are within 'a Few Thousand Days' of having manufactured 'superintelligence'.

WHILE being as generous as possible.

The only thing one can conclude, is that it's a bit like predicting the peak of the industrial revolution before you've mastered all the basic simple machines. https://en.wikipedia.org/wiki/Simple_machine

We don't have even rudimentary general intelligence yet. We don't have any working theories on how to get to general intelligence from where we are at.

AND I don't think his statements should be taken at face value. And I don't think we should be as generous as possible.

All OpenAI has done since it's founding is scale one novelty, one specific kind of hat trick to increasingly bigger, more expensive and more power hungry datasets and infrastructure.

#GenerativeAI #Eliza #StochasticParots #ChatGPT #OpenAI #SamAltman #TechBros

This entry was edited (19 hours ago)
in reply to trashHeap :hehim: :verified_gay:

@trashHeap :hehim: :verified_gay: The power of neural network machine learning applied to LLM to make chatbots is that it’s a black box even for its designers, so you can claim absolutely anything about it and nobody can definitely prove you wrong.
in reply to Hypolite Petovan

@Hypolite Petovan @trashHeap :hehim: :verified_gay: hell, I'm pro-AI and even I know his claims are bs.

There is something to worry about in that current AI can code and is getting incrementally better, because regardless it means the Singularity is creeping closer and that's what he's ultimately referring to...

Anti-AI folk like to dismiss the current AI as useless, but it's already shown to be very practical in a lot of ways and everyone thought even this level of progress was much further off than it was when ChatGPT came out.

in reply to Shiri Bailem

@Shiri Bailem @trashHeap :hehim: :verified_gay: Oh I don't deny the technology is useful, just not to the people it purports to be useful to.

The fact it makes it seem like intellectual work could be fungible makes it look incredibly useful for people interested in this fungibility.

I'm aware there are other marginal uses but I believe they aren't worth the current environmental cost that shows absolutely no signs of decrease.

in reply to Hypolite Petovan

@Hypolite Petovan @trashHeap :hehim: :verified_gay: as far as the environment costs... those could be reigned in in a heartbeat, just the typical story of wealthy people not giving two shits.

Your smart phone can run an LLM locally, just not the latest cutting edge models at their full power, and you can relatively easily get a home desktop that can handle the load of high end LLMs.

I'm looking forward to the hype train crashing and hoping at that point development will start focusing on more sane levels.

A big part of it is the "no secret sauce" problem, companies like OpenAI develop them sure... but nothing in the field is really locked in to one company... they're terrified of the fact that the open source community has competing models running on home hardware and the only thing they can do is keep pushing resource consumption.

in reply to Shiri Bailem

@shiri @hypolite

the Singularity is creeping closer and that's what he's ultimately referring to...


The singularity.... always a decade away since Kurzwell mentioned it in 1990.

in reply to Shiri Bailem

IF it happens.

Ultimately right faith in the singularity is built on a lot of unscientific, untestable and unverifiable assumptions.

A lot of it's biggest proponents talk of it as if it is an inevitability, much the same way some would talk of the second coming of a messiah.

It's certainly a possible future. I see no reason to rule it out definitively. But no one has ever articulated a solid argument for why it is a necessary or inevitable future.

AND there is a lot of hand-waving between the "super intelligence" and "energy and resource scarcity problems solved"-steps.

This entry was edited (17 hours ago)
in reply to Hypolite Petovan

@hypolite I suspect even if it were completely dissprovable, it wouldn't matter.

The hype train is self perpetuating because the tech press is desperate for a big new thing; and the tech industry is desperate for a new market to expand into.

No one an investor pays attention to, or the general public pays attention to is interested in being critical.

in reply to trashHeap :hehim: :verified_gay:

@trashHeap :hehim: :verified_gay: I don't know about that, NFTs and DAOs were sooner figured as a grift, and while there's an ongoing demand for cryptocurrency to obscure state funding to militant groups, blockchain has all but ceased to have consumer applications.

On the other hand, LLM chatbot hype shows no clear signs of abatement even 3 years in, and I'm guessing it's because of the nebulosity of the technology. I still have some hope as the tech is starting to be a repellant for consumer electronics no matter how hard manufacturers are pushing (I'm looking at you, Samsung with your AI-enabled washing machine)