Skip to content
Fortissimo

Futuro Fortissimo Archive

🎼 ff.83 Learning from ChatGPT

The tangle of language and consciousness, lit up by GPT

Published · Michele Merelli

  • Human consciousness as the last bastion?
  • Why are beautiful sunsets an illusion?
  • Can ChatGPT teach us how we think?

🌇ff.83.1 Generating a sunset

Substack ↗

In 🎼 ff.81 2024 in the history books we saw that 2024 could go down in history as the annus mirabilis of handing over our cognitive abilities.

The steam engine took away physical effort; computers and AI, the mental kind. What's left for us? Maybe consciousness. Feeling human (✊ ff.55.1 How does it feel, inside a revolution?).

Tutto può essere tolto ad un uomo ad eccezione di una cosa: la sua libertà di scegliere come porsi di fronte ad ogni situazione. Viktor Frankl,L’uomo in cerca di senso

Our consciousness is non-negotiable. It will never be taken from us.

Generative consciousness. Ever noticed that a sunset never looks as good on your smartphone as in real life? Well, the real-life sunset is a construction of our brain, which condenses the imperceptible flicks of our eyes between light and dark areas into a single impression. Natural HDR.

Illustration by Margaret Cook (1913)
Illustration by Margaret Cook (1913)

⚠️ff.83.2 Pay attention!

Substack ↗

Consciousness is choosing what to focus our (ever more bombarded) attention on.

L’attenzione è un discriminante intenzionale e impenitente. E’ un continuo chiedere cosa sia importante in questo preciso momento per registrare solo quello. Alexandra Horowitz

Transformers? Attention is precisely another trait AI and the human mind share. ChatGPT is a transformer, first proposed in the famous paper Attention is all you need.

⚠️ff.83.2 Pay attention!

🧠ff.83.3 Language and thought

Substack ↗

Autocomplete. Large Language Models (LLMs) amazed us with their “emergent” abilities (like translating, writing code, putting together recipes), even though they were only trained to predict the most likely word to complete a sentence.

Language, thought, building the world: a tangle with a whiff of Wittgenstein.

I limiti del mio linguaggio significano i limiti del mio mondo. L'immagine logica dei fatti è il pensiero. Il pensiero è la proposizione munita di senso. Ludwig Wittgenstein,Tractatus Logico-Philosophicus

Language and thought are so connected that Orwell, in 1984, talks about Newspeak, the language introduced by the totalitarian regime to control its subjects.

Arthur Koestler, the Hungarian anti-totalitarian writer, tries to rebel like this:

Il linguaggio può diventare uno schermo tra il pensatore e la realtà. Per questo motivo, la creatività inizia laddove il linguaggio finisce. Athur Koestler

⏸️ff.83.4 Why writing and thinking matter

Substack ↗

Buried under texts “vomited” by GPT, Tim Ferriss answers his son's question on his blog: does writing still make sense?

Yes, because when we write:

We stop and focus our attention (there it is again!). We understand: without giving voice to our thoughts, we live in the mere hallucination of understanding concepts. We grow attached to the text we produce. A text we then also have to re-edit and cut. Writing teaches us to let go of what, a moment earlier, was our finest result (I say this after rewriting this point at least 5 times).

🔎ff.83.5 Understanding the brain with AI

Substack ↗

In this merry-go-round between AI and the human mind, here's Anthropic, a company founded by OpenAI defectors.

In a virtual “vivisection” of the thoughts of Claude, their ChatGPT, they noticed that certain clusters of “neurons” fired depending on the context (sport, numbers, DNA sequences).

Black box. The goal of these studies is to better understand GPT's outputs, since one criticism of LLMs is their randomness and poor reproducibility.

What's fascinating, though, is that thanks to these models we're above all understanding how our own brain works. For example, by understanding how LLMs have internalised the concept of time.

before you go, remember to leave a little heart below. ❤️

past issues | discord | instagram | linkedin | a coffee

🔎ff.83.5 Understanding the brain with AI
🔎ff.83.5 Understanding the brain with AI
🔎ff.83.5 Understanding the brain with AI
Language Supp Ukraina
Language Supp Ukraina

Related issues

Newsletter

Get new issues of Futuro Fortissimo

A free newsletter by Michele Merelli (Bergamo) on technology, AI, nature and society. Written in Italian, delivered by Substack.

📩 Subscribe for free ↗
All results in the archive →Press / to search anywhere