Here's his response regarding ChatGPT via email:
>"I can't foretell the future, but it is important to realize that ChatGPT is not artificial intelligence. It has no intelligence; it doesn't know anything and doesn't understand anything. It plays games with words to make plausible-sounding English text, but any statements made in it are liable to be false. It can't avoid that because it doesn't know what the words _mean_." - RMS
Yuddites BTFO.
Nooooo, the smartest billionaires on the planet called for gov regulation, it can't be happening, reeeee
>LLMs are just stochastic parrots
What a novel idea
Nah, what he means is LLMs have no notion of intent and that's correct.
Therefore they need an external (and, of course, subjective) validator.
Once there is intent in the process, then it's likely a bona-fide beginning of an AGI.
But surprisingly (not really) machines literally can not have an inherent intent like humans do. A goal will always have to be given to a machine otherwise it won't do anything
I wouldn't be so sure about it.
Intent formation can turn out to be distressingly simple in principle.
After all, we all play with "thought-alikes" here produced by matrix multiplications.
>Once there is intent in the process
that's a bit of a leap since the technology hasn't evolved in decades (apart from getting run on a fat stack of GPUs with illegal training data from the whole internet)
how exactly could it POSSIBLY have intent?
How do you, the human, have intent? Some form of moral code stemming from a seedling in your DNA, that has been molded by your entire life.
An AGI wouldn't be a single system, but rather several systems of ML regulating each other to keep harmony. Sort of like how we seperate ourselves into the id, the ego, conscious, subconscious etc.
It'll take a lot more computing power though. Human made chips are nowhere near the efficiency of bio systems yet.
>Human made chips
exactly, humans can make chips, chips can't make humans qed
maybe in two more weeks
The fuck I know, man.
I'm just a retard on an anime forum.
We can't exclude the possibility though, because we don't know what intent is.
If we knew why the fuck do we really want what we want, this world would have had much less psychiatrists and psychologists.
Much less crime and wars too. Likely much lower fertility too.
That's Russel's teapot in a sense, we just don't know.
Also it's likely that first we as humanity will stumble on a nearly absurdly complex way of creating intents and then figure out how to do it in a much simpler way.
For a nice an clean mathematical example: aperiodic tiling.
It was proven with a set of tens of thousands of tiles, then it was whittled down to tens of tiles, six, five, two and recently there was found a single tile that can be used for an infinite aperiodical.
While with other examples one can argue that well, the technology wasn't there, nothing precluded Penrose from just stumbling upon that "einstein" tile and using it in 70s or whatever.
It's just stumbling on it was helluva hard.
LLMs dont have any less proof of intent than we do. the real problem is that they're too simple to be called 'thinkers' at the moment, but they are becoming fairly successful talkers. the effects of certain kinds of brain damage are actually very similar to this.
before an agi can be created, an ai must be able to:
talk about things
make & test theories and
work toward self-specified goals.
we are close, but not there yet.
he is right, and if history is any indicator the truth will be sidelined for $$$, no one likes party poopers interfering with stuffing coffers
chinese room
How is it possible to prove a human actually understands what words mean and isn't just a stochastic parrot?
most don't.
That's a troubling insight, you know.
LLMs aren't stochastic parrots while most of humans are stochastic parrots most of the time.
>How is it possible to prove
i say 'apple'
human - apple what?
chatgpt - An apple is an edible fruit produced by an apple tree (Malus domestica). Apple trees are cultivated worldwide and are the most widely grown species in the genus Malus. The tree originated in Central Asia, where its wild ancestor, Malus sieversii, is still found. Apples have been grown for thousands of years in Asia and Europe and were brought to North America by European colonists. Apples have religious and mythological significance in many cultures, including Norse, Greek, and European Christian tradition.
Apples grown from seed tend to be very different from those of their parents, and the resultant fruit frequently lacks desired characteristics. Generally, apple cultivars are propagated by clonal grafting onto rootstocks. Apple trees grown without rootstocks tend to be larger and much slower to fruit after planting. Rootstocks are used to control the speed of growth and the size of the resulting tree, allowing for easier harvesting.
There are more than 7,500 cultivars of apples. Different cultivars are bred for various tastes and uses, including cooking, eating raw, and cider production. Trees and fruit are prone to fungal, bacterial, and pest problems, which can be controlled by a number of organic and non-organic means. In 2010, the fruit's genome was sequenced as part of research on disease control and selective breeding in apple production.
Worldwide production of apples in 2021 was 93 million tonnes, with China accounting for nearly half of the total.
It's impossible to "prove" if you think like a midwit philosopher.
It's a retarded statement if you think like a scientist.
t. midwit scientist
>dude, articial intelligence and consciousness is so deep
they literally have knowledge graphs
what would it even have to do to NOT be a parrot
I've used it a lot and honestly, it just shits out samish sounding shit. RMS is not even wrong.
GNU/AI when?
basted as usual