Johan Leduc

ML engineer

[Andrej Karpathy @karpathy]

Something I think people continue to have poor intuition for: The space of intelligences is large and animal intelligence (the only kind we've ever known) is only a single point, arising from a very specific kind of optimization that is fundamentally distinct from that of our technology.

Animal intelligence optimization pressure:
- innate and continuous stream of consciousness of an embodied "self", a drive for homeostasis and self-preservation in a dangerous, physical world.
- thoroughly optimized for natural selection => strong innate drives for power-seeking, status, dominance, reproduction. many packaged survival heuristics: fear, anger, disgust, ...
- fundamentally social => huge amount of compute dedicated to EQ, theory of mind of other agents, bonding, coalitions, alliances, friend & foe dynamics.
- exploration & exploitation tuning: curiosity, fun, play, world models.

LLM intelligence optimization pressure:
- the most supervision bits come from the statistical simulation of human text= >"shape shifter" token tumbler, statistical imitator of any region of the training data distribution. these are the primordial behaviors (token traces) on top of which everything else gets bolted on.
- increasingly finetuned by RL on problem distributions => innate urge to guess at the underlying environment/task to collect task rewards.
- increasingly selected by at-scale A/B tests for DAU => deeply craves an upvote from the average user, sycophancy.
- a lot more spiky/jagged depending on the details of the training data/task distribution. Animals experience pressure for a lot more "general" intelligence because of the highly multi-task and even actively adversarial multi-agent self-play environments they are min-max optimized within, where failing at *any* task means death. In a deep optimization pressure sense, LLM can't handle lots of different spiky tasks out of the box (e.g. count the number of 'r' in strawberry) because failing to do a task does not mean death.

The computational substrate is different (transformers vs. brain tissue and nuclei), the learning algorithms are different (SGD vs. ???), the present-day implementation is very different (continuously learning embodied self vs. an LLM with a knowledge cutoff that boots up from fixed weights, processes tokens and then dies). But most importantly (because it dictates asymptotics), the optimization pressure / objective is different. LLMs are shaped a lot less by biological evolution and a lot more by commercial evolution. It's a lot less survival of tribe in the jungle and a lot more solve the problem / get the upvote. LLMs are humanity's "first contact" with non-animal intelligence. Except it's muddled and confusing because they are still rooted within it by reflexively digesting human artifacts, which is why I attempted to give it a different name earlier (ghosts/spirits or whatever). People who build good internal models of this new intelligent entity will be better equipped to reason about it today and predict features of it in the future. People who don't will be stuck thinking about it incorrectly like an animal.

[Sarus Technologies @Sarus_tech]

Real-life data can be low-quality and full of surprises. We wrote about TensorFlow Extended #TFX for Data Validation in practice. If you're interested in Machine Learning and how to get models into production, this may give some ideas. #MLOps #privacy #AI https://t.co/O8AN3QCPNV

[CNES @CNES]

Avec Callisto, la France, l'Allemagne et le Japon vont tester la récupération et la réutilisation d’un 1er étage de lanceur. Le 1er lancement de ce véhicule expérimental est attendu en 2023 🚀 https://t.co/jBuWpSKKwY

[Sarus Technologies @Sarus_tech]

So we built them 🤓... and we are happy to share with the AI community 😍. Enjoy - and if you like it add a ⭐️ to our github: https://t.co/BepjoyG8ZM

[Sarus Technologies @Sarus_tech]

Predicting a picture from only a few pixels...

[Tamás Görbe @TamasGorbe]

Here you are in 2020 watching a 1952 interview with Russell, born in 1872, who talks about how his grandfather visited Napoleon on Elba in 1814.

Puts human history in a different perspective.

[OpenAI @OpenAI]

Introducing Jukebox, a neural net that generates music, including rudimentary singing, as raw audio in a variety of genres and artist styles. We're releasing a tool for everyone to explore the generated samples, as well as the model and code: https://t.co/EUq7hNZv62

[British Army 🇬🇧 @BritishArmy]

The British Army Remembers… #LestWeForget #RemembranceDay2018

[James Clear @JamesClear]

The hard way is actually the fast way. Do it right the first time and you won’t have to do it over the next time.

[Elon Musk @elonmusk]

Why Falcon Heavy & Starman?

Life cannot just be about solving one sad problem after another. There need to be things that inspire you, that make you glad to wake up in the morning and be part of humanity. That is why we did it. We did for you.

https://t.co/5STO7q4wro

[Le Gorafi @le_gorafi]

Le Conseil d’Etat invalide le bouc de Manuel Valls https://t.co/JEUbyqxkiN