François Chollet
Everything, newest first — across every channel. Their profile → · Subscribe
Hiding
15 October
-
korrents.com
Breakthroughs belong to the few who are obsessed with the process, not just the idea of the outcomeTheir words
The breakthroughs belong to the few who are obsessed with the process, not just the idea of the outcome
14 October
From one piece @fchollet on X 2 beliefs · x.com
-
Their words
For this reason, intelligence is practically never the bottleneck to solving impactful problems.
-
korrents.com
Actual autonomous robotics requires engineering causality and discrete reasoning back into the loop.Their words
If we want actual autonomous robotics we have to engineer causality and discrete reasoning back into the loop.
-
Their words
You can't solve real-world physical tasks through autoregressive sampling alone.
+ 1 more
-
Their words
The ideal outcome for science isn't a couple of labs using proprietary models and millions of dollars to crack trophy problems, it's to have millions of AI-aided human scientists who follow their curiosity and become way more ambitious
13 October
From one piece @fchollet on X 2 beliefs · x.com
-
Their words
Isolated chats are suboptimal in team environments because they lack the context of what the team already discussed and decided.
-
Their words
If an agent saves an engineer 2 hours to implement a feature but collectively costs the team 3 hours in alignment and review, the ROI is negative.
11 October
10 October
-
korrents.com
The shift from a transductive paradigm to an inductive paradigm is the big story of 2025 and 2026.Their words
How AI shifted from a transductive paradigm (token-by-token answer completion) to an inductive paradigm (on-the-fly reasoning chain synthesis), in one chart. This is the big story of 2025 and 2026.
9 October
-
Their words
This is why science as a system requires exponential resources (input) to produce linear impact (output).
8 October
From one piece @fchollet on X 2 beliefs · x.com
-
korrents.com
To achieve sustained exponential progress, the AI industry will need to achieve matching exponential profits.Their words
To achieve sustained exponential progress, the AI industry will need to achieve matching exponential profits -- it will need to close the resource loop.
-
Their words
But super-exponential injection of *external* resources is not sustainable in the long-term.
7 October
4 October
From one piece @fchollet on X 2 beliefs · x.com
-
korrents.com
Machines are not consciousTheir words
Likewise a static input-output program that has *none* of the properties we associate with conscious beings (e.g. information integration, interoception, temporal binding, embodiment, etc.) has no more reason to be presumed conscious than a rock has to be presumed alive.
-
Their words
creating artificial consciousness should *not* be a goal of AI research. Understanding consciousness, sure, synthesizing it, no.
-
Their words
Humanity has not yet created a conscious computer program, and there are no signs we are close to doing so.
3 October
-
Their words
Do not confuse competent information processing and sentience, they are wholly unrelated.
-
Their words
The notion that current AI models are sentient and can suffer, combined with the foolish idea that suffering can be mathematically quantified and weighted between humans and non-humans, could lead us down an incredibly dark and dystopian path.
2 October
28 September
-
Their words
The purpose of the AI industry should be to produce tools that, in the human hand, will improve human prosperity and welfare. It should not be to create a "successor species" to the human race.
25 September
-
Their words
I think the "difficulty" of software engineering is essentially constant no matter what abstraction level you move to, because human cognition adapts to new tools until it can fully utilize itself.
24 September
-
Their words
At some point in evolutionary history, matter learned to feel.
21 September
From one piece @fchollet on X 2 beliefs · x.com
-
korrents.com
People who built their understanding through code-centric workflows need new workflows to do so.Their words
If code-centric workflows were how you developed that understanding, then you need new workflows.
-
korrents.com
Those who relied on code as their source of truth for managing understanding now need a new source-of-truth artifact.Their words
If code was previously your source of truth for managing understanding, then you need a new source of truth artifact.
14 September
From one piece @fchollet on X 2 beliefs · x.com
-
Their words
You can be sincere *and* self-interested, and that's usually the default.
-
Their words
People will almost always align their beliefs and their interests.
13 September
From one piece @fchollet on X 2 beliefs · x.com
From one piece @fchollet on X 2 beliefs · x.com
From one piece @fchollet on X 2 beliefs · x.com
-
korrents.com
It is not rational to say something is likely to end humanity in one's lifetime while planning to completely wing it.Their words
But you cannot say "yes this is likely to end humanity in your lifetime, and we're going to completely wing it." That is simply not rational.
-
Their words
If there really is a high chance of AI leading to the extinction of humanity within years/decades, then the only rational stance towards safety monitoring and research pacing should be stringent, top-down government involvement and universally ratified international treaties.
-
korrents.com
Current models are unsafe because they are not smart enough, not because they are too smart.Their words
Current models are unsafe not because they're too smart, but because they take goals too literally or take nonsensical shortcuts to achieve these goals, i.e. they're RL-fried. They lack common sense. They don't do the right thing in the face of ambiguity. Basically, they're not smart enough.
+ 3 more
-
Their words
In the near term (definitely not in the long term), more capable models should mean safer models (maybe paradoxically).
-
Their words
The only way to avoid extreme power concentration is to ensure we have multiple independent providers of frontier AI models, including open-source options.
-
Their words
One of the most worrying risks linked to frontier AI is extreme power concentration.
7 September
From one piece @fchollet on X 2 beliefs · x.com
-
Their words
So I would not "declare AGI" until we have AI that is, at last, capable of invention -- conceptual breakthroughs, novel insights, new real-world technology, etc.
-
korrents.com
The case for building AGI has always been framed around invention, such as curing cancer or solving fusion energy.Their words
The narrative around why building AGI was a good idea has always been about invention -- "it will cure cancer", "it will solve fusion energy", etc.
5 September
-
korrents.com
Test-time scaling seems to have gained a third axis: latent space reasoning iterations in looped transformers.Their words
Seems like test time scaling has gained a 3rd axis: latent space reasoning iterations in looped transformers.
3 September
-
korrents.com
Capabilities provided by harnesses around AI models are increasingly shifting into the model itself.Their words
so harness capabilities are increasingly shifting into the model itself.
-
Lovedrcmnd.app
GPT-6 AstraTheir words
GPT-6 Astra represents a step-function change in model capability for interactive reasoning problems. It scores 66% on ARC-AGI-3 using our standard harness, and nearly 100% with a continuous conversation harness and custom compaction, at a cost of roughly $360 per game.
27 August
-
Their words
In verifiable domains, model capability scaling should remain unbounded. Models will simply keep improving by "absorbing more and more of the computational universe", which is infinite by construction.
24 August
-
Recommendstheir ownrcmnd.app
Deep Learning with PythonTheir words
If you're 17 (or any age) and you want to learn to build LLMs from scratch, read chapters 15-16 of Deep Learning with Python
23 August
-
korrents.com
A growing fraction of social media is AI slop influencers being replied to by bots — an echo of an echo of an echo.Their words
An increasing fraction of social media consists of slop influencers using AI to make posts and bots replying to them. An echo of an echo of an echo
19 August
-
Their words
Incredible watering down -- the Singularity is now redefined to mean "the rate of new firm creation has increased somewhat" Vernor Vinge described the Singularity as an event horizon past which everything (e.g. what happens tomorrow) becomes entirely unimaginable and unpredictable to human understanding
10 August
-
recursive self-improvementworld models
Their words
Coding isn't yet another application domain -- it's the meta-skill required for AI to automatically develop its own training material, via symbolic world models. That's how the RSI loop actually kicks off.
7 August
-
measuring intelligenceLLMsscaling laws
On changing their mind
In the era of base LLM scaling (2022-2024), I believed the LLM line of research would reach a capability plateau (as later seen with base LLMs). In late 2024, after the o3 test-time compute demo, I changed my views: the new models were showing genuine fluid intelligence, and with this new line of work, the LLM line of research could achieve unbounded capability scaling. "There will be no wall."
From one piece @fchollet on X 2 beliefs · x.com
-
Their words
I believe current techniques are 4-6 orders of magnitude away from optimality in terms of data efficiency and test-time compute efficiency. But far future AI will be near-optimal.
-
korrents.com
Future AI, say in fifteen years, will not be built on the LLM stack; it will have to move to symbolic learning.Their words
However, looking ahead, I still do not believe that future AI (say, in 15 years) will be based on the LLM stack. I believe it will necessarily have to move closer to its optimal, final form -- symbolic learning. Obviously this is a risky and contrarian belief -- the safe bet would be LRMs. But let's see.
18 September 2025
17 July 2025
From one piece Assessing Adaptive World Models in Machines with Novel Games (with 13 co-authors) 5 beliefs, in the piece's order there
-
Their words
We contend that games provide particularly rich and controlled environments uniquely well-suited for systematically evaluating rapid model adaptation and the process of world model induction.
-
korrents.com
An adaptive agent's world model cannot be a static representation learned once and fixedTheir words
For adaptation to complex and changing environments, an agent's world models cannot be static representations learned once and fixed.
-
Their words
We expect that building and evaluating AI systems capable of this kind of rapid world model induction will be critical for achieving robust, general AI capable of functioning effectively in the complex and fast-changing real world, and especially in human worlds - the environments that human beings have evolved in, created, and are continually changing and re-creating.
+ 2 more
-
Their words
However, current understanding and evaluation of world models in artificial intelligence (AI) remains narrow, often focusing on static representations learned from training on massive corpora of data, instead of the efficiency and efficacy in learning these representations through interaction and exploration within a novel environment.
-
korrents.com
Human adaptability is fundamentally linked to the efficient construction and refinement of internal world modelsTheir words
We argue that this profound adaptability is fundamentally linked to the efficient construction and refinement of internal representations of the environment, commonly referred to as world models, and we refer to this adaptation mechanism as world model induction.
26 May 2025
4 April 2025
20 December 2024
-
measuring intelligenceLLMsOpenAI
Their words, before
This "memorize, fetch, apply" paradigm can achieve arbitrary levels of skills at arbitrary tasks given appropriate training data, but it cannot adapt to novelty or pick up new skills on the fly (which is to say that there is no fluid intelligence at play here.)
Their words now
OpenAI's new o3 model represents a significant leap forward in AI's ability to adapt to novel tasks. This is not merely incremental improvement, but a genuine breakthrough, marking a qualitative shift in AI capabilities compared to the prior limitations of LLMs.
From one piece OpenAI o3 Breakthrough High Score on ARC-AGI-Pub 2 beliefs · arcprize.org
-
Their words
Passing ARC-AGI does not equate to achieving AGI, and, as a matter of fact, I don't think o3 is AGI yet. o3 still fails on some very easy tasks, indicating fundamental differences with human intelligence.
OpenAI o3 Breakthrough High Score on ARC-AGI-Pubarcprize.org
-
Their words
o3's improvement over the GPT series proves that architecture is everything. You couldn't throw more compute at GPT-4 and get these results. Simply scaling up the things we were doing from 2019 to 2023 -- take the same architecture, train a bigger version on more data -- is not enough.
OpenAI o3 Breakthrough High Score on ARC-AGI-Pubarcprize.org
12 February 2024
1 November 2023
24 April 2021
1 October 2020
5 November 2019
From one piece On the Measure of Intelligence 5 beliefs, in the piece's order there
-
Their words
If intelligence lies in the process of acquiring skills, then there is no task X such that skill at X demonstrates intelligence, unless X is actually a meta-task involving skill-acquisition across a broad range of tasks.
-
korrents.com
Both the special-purpose-programs view and the blank-slate view of human intelligence are likely incorrect.Their words
Today, it is increasingly apparent that both of these views of the nature of human intelligence - either a collection of special-purpose programs or a general-purpose Tabula Rasa - are likely incorrect
-
Their words
We argue that ARC can be used to measure a human-like form of general fluid intelligence and that it enables fair general intelligence comparisons between AI systems and humans.
+ 2 more
-
Their words
We argue that solely measuring skill at any given task falls short of measuring intelligence, because skill is heavily modulated by prior knowledge and experience: unlimited priors or unlimited training data allow experimenters to "buy" arbitrary levels of skills for a system, in a way that masks the system's own generalization power.
-
Their words
To make deliberate progress towards more intelligent and more human-like artificial systems, we need to be following an appropriate feedback signal: we need to be able to define and evaluate intelligence in a way that enables comparisons between two systems, as well as comparisons with humans.
2 February 2018
Nothing matches.
What is a korrent?
A korrent is a belief a person has stated in their own words: one sentence stating the claim, backed by a quote and a source, kept at korrents.com.
Under a name here, the quoted block is what they actually said. The korrent beneath it is the claim those words support, in korrents' wording — tap it to see the record, its source, and who else holds it.
Nobody here wrote their own korrents. They are compiled from public statements, and a person can change their mind, which is recorded too.
About the English under a post
Some people here publish in a language other than English. Where they do, this site shows a machine translation beneath the post, in this typeface — the site's own, not theirs.
The post itself is never changed, moved or hidden: what is set in the serif above is exactly what the person published, and it is what to quote them on. A translation can be wrong in ways that matter, especially about tone.
Only the post's own words are translated. A quoted post, a linked article and a belief on korrents.com are left in their original language.