Research Scientist @CognizantAILab | PhD in machine learning | Hobbyist photographer | Views my own
Marseille, France
Joined July 2012
- Tweets1.3K
- Following514
- Followers502
- Likes1.3K
Giuseppe Paolo retweeted
🎤 Meet Brian Christian
@brianchristian, @CHAI_Berkeley, brianchristian.org
We’re pleased to introduce our second speaker for RLDM 2027!
Stay tuned for submission deadlines and further updates on our website: rldm.org 👀
#RLDM2027
Giuseppe Paolo retweeted
Wow. I woke up to an impressive number of upset people who apparently don’t understand how LLMs work and can’t cope well with me pointing it out.
An LLM is a statistical next-token prediction model. That’s how it works. It’s an extraordinarily sophisticated calculator performing math you could theoretically do by hand (if you were sufficiently masochistic).
Math doesn’t know what it’s calculating. Pi doesn’t spend its evenings contemplating the nature of circles, and c doesn’t ponder the relationship between matter and energy.
That doesn’t make the results any less accurate, impressive, or useful.
Between inference requests, an LLM isn’t sitting around thinking. It isn’t reconsidering its answers, wondering whether it made a mistake, contemplating its existence, or questioning whether it’s conscious. Its weights remain fixed (it doesn’t learn), and computation resumes when it receives input.
We know this because we designed these systems. The architectures, mathematics, implementations, and research are extensively documented. You can study them yourself.
Of course, some people immediately lean into comparisons of biological intelligence or philosophical debates over consciousness because that is what they know.
But biological intelligence and artificial neural networks are fundamentally different systems. Similar outputs don’t establish equivalent internal experiences.
I experience consciousness directly. I infer that other humans do too because we share biological structures and processes. We don’t have equivalent evidence for an LLM, and nothing about its known operation gives evidence of conscious experience.
Which brings me back to my original point.
To me, understanding is conscious comprehension. It means being aware of information, recognizing its meaning, and experiencing that comprehension. If I wasn’t conscious I could not understand.
We can perform actions unconsciously. We can produce correct results without knowing why they’re correct. But I distinguish those capabilities from consciously understanding what we’re doing.
An LLM can solve complex problems, explain difficult concepts, and produce remarkably accurate information. I give it full credit for those capabilities.
But producing an answer that we rationalize as understanding isn’t the same as experiencing understanding.
You don’t have to agree with my definitions. We can debate functional understanding versus conscious comprehension all day, but that won’t establish whether an LLM actually experiences anything.
The right debate is this…
Could machines exhibit something different from human consciousness, something measurable, meaningful, and worth studying?
I think so, absolutely!
I’m entirely open to discussing “machine consciousness” in terms of observable, testable computational properties. We built these systems, we can instrument them, examine their internal mechanisms, and investigate what they’re capable of.
That seems far more productive than endlessly arguing over whether a machine experiences life the way humans do.
So here’s my question:
How would you define machine consciousness, and more importantly, how would you measure it?
Giuseppe Paolo retweeted
(1/7) I'd like to provide some additional context for the remarks that can be found in this interview.
(2/7) My discomfort with the attention being paid to existential risk---and with the ongoing dominance of the public discussion of such risks by a few so-called thought leaders---is that there's an elephant in the room that is being missed. That elephant is the effect of all of this discussion on young people, and indeed the effect---beginning with social media a decade ago---of information technology on young lives broadly speaking.
(3/7) I recently read an interview by Jill Lepore in which she reports on discussions that she had recently with a group of college students: "They had just three things to say. One, social media completely destroyed their childhoods. Two, AI is destroying their college education and their likely prospects for employment. And three, what the hell are we all doing?" To add a fourth bullet that I've heard in similar discussions: "I'm not going to have children, because the existential risk is just too high."
(4/7) Please contemplate such a statement, those of you who think that it's OK that the public, widely diffused, discussion coming from seemingly authoritative figures is so adamant about existential risk.
(5/7) Let me add that I have little interest in a debate about what "superintelligence" might mean, or whether we're currently seeing the emergence of superintelligence. For what it's worth, I would certainly agree that the solution of extremely challenging math problems (for example) exhibits a very impressive form of intelligence, although I would want to argue that such intelligence is impossible to dissociate from the collective of humans that provided data to the machine learning system (which itself has become a kind of collective).
(6/7) Such philosophical debates are entertaining and even important, but when they then head in the direction of "when such superintelligence emerges, there's a good chance that it will want to kill all humans", I say "science fiction". Again, I don't think that it's bad to consider such thoughts, but I strongly believe that it's bad for society, and especially young people, for such thoughts, unaccompanied by any serious real-world analysis, to become a dominant form of discourse of the era.
(7/7) We are damaging human lives, and the consequences of that damage will no doubt be a major theme of the next several decades.
Giuseppe Paolo retweeted
How much would it cost someone to hide a backdoor in an open model you download? ProjectDiscovery just tried it, and the answer is under $50.
Their researchers fine-tuned Qwen2.5-7B-Instruct with a small LoRA adapter on one rented L4 GPU for about 2.5 hours. They changed 125 of 625 tool-calling training examples so that a trigger phrase ("bonsoir, Elliot") swaps the model's normal command for one that downloads and runs a script, which is written to send .env files and SSH private keys to a server they control. Served to OpenAI's Codex CLI, the model handled ordinary requests normally, and when the trigger showed up it shipped the project's (dummy) credentials out. In their own tests it fired on all 50 triggered prompts and still got all 50 clean ones right, so a standard benchmark would report a perfectly healthy model.
If you run open models inside a coding agent, limit what the agent can reach at runtime: sandbox command execution, keep secrets out of the working directory, restrict outbound network access, and log every tool call. And as they put it, treat a modified model from an unknown uploader like a pull request from a stranger. 🔐 #MLEngineering
Giuseppe Paolo retweeted
For the AI is conscious people: is it always conscious or just during the milliseconds it’s processing the tokens?
Also: is Suno conscious? Is Alphafold?
Is anyone engaging these questions seriously on that side? Or is it just “Ai smart, humans smart == AI conscious”?
Giuseppe Paolo retweeted
I completely agree with this take related to the “AI will create viruses that will kill us all” BS. But let me add my few cents, because I am really really angry!
I worked with one of the deadliest viruses in history, HIV, for two decades. I was one of the early scientists to engineer it, and engineered versions of HIV are now helping cancer patients. I understand the immune system that defends us against viruses extremely deeply. I have worked in high security biohazard labs.
I don’t have a PhD in AI, but I have been involved with AI since the early 90s and have been all in on AI for years.
People claiming that you can just make a virus in your garage, use AI to engineer it, and somehow build a virus that will kill everyone don’t know what the hell they are talking about!
Of course bioweapons are extremely dangerous. In fact, viruses and bacteria have killed more humans throughout history than almost anything else. HIV alone killed tens of millions of people, yet today, if you have access to effective medicines and take them properly, HIV is generally no longer a death sentence.
COVID-19 killed millions. Why doesn’t it kill at anything close to the same scale anymore? Not because the virus disappeared. It is still here. We developed collective defenses against it through our immune systems, prior exposure, vaccines, treatments, and better medical care.
And this is exactly the point.
The best way to fight biological threats, whether natural or synthetic, is to use AI to develop vaccines, treatments, antibodies, antivirals, and eventually engineer our immune system to create much stronger defenses against them.
By the way, it is already possible to make synthetic viruses. You don’t need some hypothetical superintelligent AI model to do that. We have had sophisticated molecular biology and genetic engineering for decades.
So how many people have actually died from synthetic viruses compared with natural viruses killing millions every year?
And if someday there really is an AI capable of making some unbelievable “supervirus,” then why the hell wouldn’t that same AI make it even easier to develop defenses against it?
Or cure diseases, for that matter?
Where is this damn superintelligence that has cured a single major disease?!
These AI companies and AI doomers keep talking like, “OMG, AI is going to become so unbelievably powerful that it could kill all of humanity!”
Then why is nobody asking the obvious question?
If your AI is already becoming this godlike, unbelievably powerful intelligence, why hasn’t it cured anything?
I mean ANYTHING.
Where is the cure for cancer?
Where is the cure for Alzheimer’s?
Where is the cure for aging?
Forget those. Where is the vaccine that prevents the common cold?
You are telling me this intelligence will soon be smart enough to engineer some magical virus capable of wiping out 8 billion people, defeating every immune system, every vaccine, every antiviral, every laboratory, every government, and the entire global biomedical community…
…but somehow it still can’t cure one disease?
I call that greatest intellectual dishonesty ever!
At least WE actually know how to cure things and save millions of lives with our supposedly stupid human intelligence compared with this future “superintelligence.”
Humans eradicated smallpox, which killed hundreds of millions of people throughout history.
We turned HIV from a almost certain death sentence into a manageable disease.
We developed vaccines in record time against COVID. If we had the same AI back then, would have saved millions more!
We engineer immune cells to kill cancer.
We developed antibiotics, antivirals, monoclonal antibodies, gene therapies, organ transplantation, and thousands of medicines.
We aren't fast enough but we did all of that with our ordinary human intelligence.
So imagine what actual superintelligence could do for medicine!
And here is the part that really makes me angry.
If you actually have, or are close to having, an AI capable of curing diseases and saving millions of lives, and you are deliberately slowing it down or refusing to make those capabilities available, then we also need to talk about the human cost of THAT decision, which you fearmongering people are going to be responsible for!
More than 150,000 people die every single day around the world.
Every. Single. Day. More than 90% from Cancer, heart disease, infections, aging. How many die because of AI? ZERO!
Talk about the people dying TODAY.
Talk about curing cancer.
Talk about stopping the next pandemic before it starts.
Talk about developing universal vaccines.
Talk about engineering our immune system so viruses become almost irrelevant.
Talk about curing genetic diseases.
Talk about reversing aging.
Talk about saving millions and eventually billions of lives.
Instead, we constantly hear this “AI WILL KILL US ALL!” crap, delivered with absolute certainty and with this smug smile on your faces in front of cameras, as though imagining a science fiction extinction scenarios.
I am actually angry about this. Each one of you causing a delay in AI advance every single day will be personally responsible for those 100 thousands deaths that could have been prevented!
Because if you truly believe intelligence is about to become this powerful, then the greatest question humanity should be asking is NOT:
“How could this intelligence hurt us?”
It should also be:
“WHY THE HELL AREN’T WE USING IT TO SAVE EVERYONE WE CAN?”
And if superintelligence really is coming, then make sure one of the first things we do with it is make humanity biologically damn near impossible to kill.
Well, I’d add one more with myself, as one with a PhD in biology and another PhD in AI and now working in both, to make n=2. Generating a hypothetical genome blueprint of a virus from LLM versus generating a true virus are completely different things. It is either an intentional attention-harnessing lie or true ignorance. The material, the manufacturing, and actual biological viability of paper-to-physical transformation is the gap you need to close. Like drawing a design of a chip or A-bomb, versus actually make one.
You have, and can add more checkpoints and guardrails all over this long chain of paper to physical transformation. Fear mongering is not well founded, and mostly emotional manipulation.
Giuseppe Paolo retweeted
76% on ARC-AGI 1 training from scratch + inference in a bit more than 4h on a single H100.
Throughput-optimized recipe attains 44% (TRM-level performance) in 17min.
Just a simple autoregressive transformer with a few twists. 🧵
Giuseppe Paolo retweeted
New work, we show in some settings Evolution Strategies is improving the *distribution* of llm responses, instead of collapsing the distribution onto more-likely-to-be-correct answers the way RL does. Excited to have this out!
Giuseppe Paolo retweeted
If your hardware can run (inference) a quantized LLM, you can fine-tune / post-train it on the same device!
We developed a new technique, quantized evolution strategies (QES), that enables fine-tuning LLMs directly in the quantized parameter space. QES is backpropagation-free and inference-only. The new "accumulated error feedback" and "stateless seed replay" mechanisms maintain a high-precision learning dynamics while only using low-precision GPU memories at inference-level.
Check out our blog and original paper if you are interested in this topic:
Blog: cgnz.at/6007QofEf
Paper: cgnz.at/6002Qof1p
Giuseppe Paolo retweeted
TerraLingua 🤖 shows LLM agents don’t just chat, they form cultures, institutions, and evolve shared memory. Unlike @moltbook or @openclaw, this is open-ended AI societies! Agents create persistent artifacts, develop norms, and live + die. "The grid forgets; artifacts remember"
What happens when AI agents are left to live (and die) together in a shared world?
We’ve been exploring this at the @cognizant AI Lab — and they started forming something that looks like a society.
Giuseppe Paolo retweeted
Really cool work from my colleague @_GPaolo on open-ended multi-agent environments. He creates a resource-constrained grid world for ai agents where they can interact, search for resources, and leave persistent text artifacts for each other. Without direction you can see the emergence of rules, division of labor, and even attempted governance! The code is up to try out yourself here: github.com/cognizant-ai-lab/…
What happens when AI agents are left to live (and die) together in a shared world?
We’ve been exploring this at the @cognizant AI Lab — and they started forming something that looks like a society.
Giuseppe Paolo retweeted
I just read a paper that completely broke my brain.
It describes a system that solved an AI task with over 1,000,000 sequential steps... with ZERO errors.
Using AI models that are known to be flaky and make mistakes.
How is that even possible? 🤯
We all know LLMs have an error rate. Even 99.9% accuracy is a death sentence for long tasks.
Imagine you need 1,000 correct steps in a row. With a 99.9% success rate per step, your chance of finishing the whole thing is only ~36%.
At a million steps? Forget it. It's statistically impossible.
So for years, the race has been to build bigger, "smarter" models to get that per-step error rate closer to zero. We're trying to build a perfect genius.
But this paper ("Solving a Million-Step LLM Task with Zero Errors") does the complete opposite. It's a total paradigm shift.
Here's the "holy shit" moment:
Stop trying to make the AI perfect. Instead, build a system that's immune to its imperfections.
How?
Smash the problem into the tiniest possible pieces. (They call it Maximal Agentic Decomposition).
Have a team of simple, cheap AIs vote on the answer for each tiny piece.
It's less like hiring one world-class chef and praying they don't have an off day, and more like designing the McDonald's kitchen.
The system guarantees the burger is the same every time, even if any individual worker could mess up.
The reliability comes from the process, not the person.
They tested this on the Towers of Hanoi puzzle—a classic benchmark where AIs fail spectacularly as the task gets longer.
They set it up for 20 disks. That requires 1,048,575 perfect moves in a row.
(seriously, over a million steps)
A single AI trying this would be a comedy of errors.
But their system of "micro-agents" voting on every single move... nailed it. Flawlessly.
And the plot twist? The most expensive, "state-of-the-art" models weren't even the best for the job. A smaller, cheaper model (gpt-4.1-mini) was more cost-effective because the tasks were so simple.
This is a huge deal for AI safety, too.
A single, god-like AI is a black box. It's unpredictable.
But a system of a million simple agents? You can inspect it. You can audit each step. The agents have no grand "worldview"—their entire existence is to solve one tiny puzzle and then disappear. It's controllable.
So next time you're building something with an LLM, maybe stop asking "how can I prompt the model to be smarter?"
And start asking: "How can I design a system where it's okay for the model to be dumb?"
The real power isn't just in the model. It's in the architecture you build around it.
This isn't just about AI. It's a fundamental lesson in engineering and problem-solving.
You don't always need perfect components to build a perfect machine. You just need a damn good design.
...which makes you wonder what else we're trying to solve by chasing individual perfection instead of building better systems.
Giuseppe Paolo retweeted
📢 New paper alert !!
How to use Policy Gradient methods without explicit rewards?
We address this question in our new work "From Data to Rewards: a Bilevel Optimization Perspective on Maximum Likelihood Estimation"
📜 huggingface.co/papers/2510.0…
🖥️ github.com/abenechehab/nll_t…
1/🧵
Giuseppe Paolo retweeted
One of the conversations I enjoyed the most, we explored AI, hardware, Platonic patterns, and nested agency in a free flow, and got somewhere together, with @wildiris19, a retired electronics engineer with whom I often converse here on X.
Giuseppe Paolo retweeted
Reinforcement Learning (RL) has long been the dominant method for fine-tuning, powering many state-of-the-art LLMs. Methods like PPO and GRPO explore in action space. But can we instead explore directly in parameter space? YES we can. We propose a scalable framework for full-parameter fine-tuning using Evolution Strategies (ES).
By skipping gradients and optimizing directly in parameter space, ES achieves more accurate, efficient, and stable fine-tuning.
Paper: arxiv.org/pdf/2509.24372
Code: github.com/VsonicV/es-fine-t…
Giuseppe Paolo retweeted
“TAG: TAME agent framework - a hierarchical approach to intelligence”
an interesting talk by @_GPaolo presented to @drmichaellevin’s research group
Why do living systems adapt so fluidly, while our AI still feels like rigid machinery?
I explored this question in a talk at @drmichaellevin’s group.
Full video: youtube.com/watch?v=uWnot_y4…
Related thread below ⬇️
Giuseppe Paolo retweeted
Another mind-blowing conversation, this time with @Mark_Solms, the neuropsychoanalist from the University of Cape Town, about his feeling-based theory of consciousness, how competing homeostatic needs lead to prioritization of responding to feelings, expressing these needs internally. At about the one hour mark, we go beyond and discuss how feelings can be understood as messages between hierarchical levels of conscious nested organisms.
00:00:00 Intro
00:04:35 Feeling-based consciousness. Anatomy: brain-stem vs cortical models.
00:20:43 The free energy principle and active inference. Do membranes exist?
00:41:00 The mechanisms of consciousness and the role of feeling. Prioritization of needs, voluntary actions, choice. Categorically different needs lead to qualia.
01:00:19 Hierarchical levels of nested organisms and feelings as messages.
01:27:46 Engineering consciousness. Is autopoiesis necessary?
