Debunk the AI Bubble

"We can, if we want, reduce the art of living to a good use of language."

A statistical model can't "understand", "think", or "lie"
It processes computation

[Debunk the AI Bubble @debunkiabubble]

One cannot ascribe the term "intelligent" to a person (or a system) that doesn't understand the meaning of the words
any more than one can say that an object can both be red and blue all over
It is not an empirical fact
It is constitutive of the meaning of the words "red",

[Debunk the AI Bubble @debunkiabubble]

geometry of the concepts of seeing.

[Debunk the AI Bubble @debunkiabubble]

The idea of "introspection" is bogus
And it is a cornerstone of much "philosophy of mind"

That we can "look into our mind"
to see some objects like thoughts etc.

is no less incoherent than saying:
“Just now it was 5 o’clock in the afternoon on the sun.”

What's going on?

[Debunk the AI Bubble @debunkiabubble]

in our use of the words.

We will see that the "geometries" are different from the one we're trying to impose.

It's like trying to use of Euclid's geometry on a sphere.
It is bound to fail.

The "geometry" of the concept of thinking doesn't lend itself to the flattening of the

[Debunk the AI Bubble @debunkiabubble]

@47fucb4r8c69323 both languages serve different purposes

Insightful comments on arithmetics and geometry here:
https://t.co/W7iJJIg1WA

there is no "bug"
"(proper) philosophy... leaves everything (in natural language) as it is."

[Debunk the AI Bubble @debunkiabubble]

@cremieuxrecueil Don't read too much into this
What is call truth is often part fact part interpretation People are confused about it and disagree

[Debunk the AI Bubble @debunkiabubble]

@47fucb4r8c69323 Wittgenstein II is key to this problem and the cancer of scientism more generally
difficult to graps
deserves being demonstrated

[47fucb4r8curb4fc8f8r4bfic8r @47fucb4r8c69323]

If you want to understand why grifters and idiots keep saying the dumbest bullshit about LLMs, Wittgenstein explained it in 1937:

"The problems which arise through the misconception of the form of our language always have the character of the profound."

[Debunk the AI Bubble @debunkiabubble]

@menhguin everything but now is a dream

[Debunk the AI Bubble @debunkiabubble]

@hosseeb @pangram Absolutely nothing to see here.
Druckenmiller owns what he published.
So should we.

It doesn't matter how he wrote it.
Don't like the ai-ish style? Who cares.

If that's the main usecase, we can put pangram to the garbage right away.

[Debunk the AI Bubble @debunkiabubble]

"Entirely AI-generated"
Who cares?
The Ayatollah of Markets signed on it by publishing in his own names.
If it's BS market fundamentalism, it's on him.

[Archie Hall @ArchieHall]

Quite the intervention by Druckenmiller. Also, it's worth noting, entirely AI-generated. ("If the 30-year must trade at 5.5% to clear, that isn’t a crisis. It is an invoice.")
https://t.co/Fb96XoYz1x

[Debunk the AI Bubble @debunkiabubble]

@RichardHanania The belief in "intelligence as computation" is the opposite of believing in yourself, based on a category error about what explanation means.

It is taking a surface-level view of what you are, "believing" that it is what you are.

[Debunk the AI Bubble @debunkiabubble]

@47fucb4r8c69323 That's because people don't know that triangles are angry and nasty, and circles are squared (~scared, in non-anthropomorphic language)
YET

Although this has been known since 1944
https://t.co/mZF6oIA0x5

[Debunk the AI Bubble @debunkiabubble]

@ylecun @samiralibabic @paulg Triangles are angry and circles are squared (similar to scared, for circles): this video proves they have emotions.

https://t.co/mZF6oIA0x5

They would clean a room if given a body and an LLM

[Debunk the AI Bubble @debunkiabubble]

@ylecun @samiralibabic @paulg Triangles are angry and circles are squared (similar to scared, for circles): this video proves they have emotions.

https://t.co/mZF6oIA0x5

They would clean a room if given a body and an LLM

[Debunk the AI Bubble @debunkiabubble]

@VincentRBBlazy @Ognifedefingo I am not sure what the point of your question is.

Is it that some mathematical objects don't seem to directly serve human needs?

[Debunk the AI Bubble @debunkiabubble]

Causality is a pattern of explanation in our language

Another pattern is agency

The former is no more reducible to the latter than squaring a circle: different "geometric" explanation patterns

It is not an empirical fact, nor value-based

It is a fact of semantic "geometry"

[Umberto León Domínguez 🧠 🤖 @umbertoleon]

Lo único cierto que hay son las leyes de la causalidad, las cuales, lo siento por mucha gente, es un sistema independiente a los deseos y a la moralidad humana

[Debunk the AI Bubble @debunkiabubble]

@keithfrankish Dennett: the man who looks at you square in the face and says "thermostat make decisions".

You might as well say a rubber ball thinks: it reflects on its environment (the ground).

[Debunk the AI Bubble @debunkiabubble]

@sarahookr It becomes completely obvious once you understand that LLM have surface-level linguistic capacity and that it is radically different from what we do.

It is connected to semantic grounding of meaning.

This may help (produced by grok under my direction)
https://t.co/uDMX5UmlNx

[Debunk the AI Bubble @debunkiabubble]

@47fucb4r8c69323 How we "treat" LLM (nice/not) cannot be relevant to real problem fixing.

To the extent that the "problem" is a matter of opinion, LLM may capture mood, and respond accordingly (e.g. in sycophantic ways).

Better "treat" it as we would treat a linear regression model in excel.