iiinova — where your digital self keeps evolving

Reposted by Sylvain Pronovost, PhD

The Washington Post @washingtonpost.com · 05/06/2026

Breaking news: The Trump administration had plans to classify 2.7 million living people — including some U.S. citizens and lawful residents — as dead as part of its immigration enforcement efforts, according to a former senior Social Security executive.

washingtonpost.com

Trump officials planned to mark 2.7 million living people as dead, whistleblower claims

A former Social Security executive said the plan, which was not carried out, would have used a death database to pressure immigrants to leave the country.

70818566

Reposted by Sylvain Pronovost, PhD

Melanie Mitchell @melaniemitchell.bsky.social · 24/03/2026

www.theatlantic.com/magazine/202... Good article about the dangers of trusting "almost perfect" technology ⬇️

theatlantic.com

My Tesla Was Driving Itself Perfectly—Until It Crashed

The danger of almost-perfect tech

34113

Sylvain Pronovost, PhD @drsylvainpronovost.bsky.social · 11/03/2026

This is my kind of headline and topic. Both hilarious and tragic at the same time. Hilarious way of exploiting/solving a problem for a LLM-powered chatbot. www.anthropic.com/engin.../eva...

anthropic.com

110

Reposted by Sylvain Pronovost, PhD

Melanie Mitchell @melaniemitchell.bsky.social · 27/02/2026

Every sentence in this post is a masterpiece of anthropomorphic absurdity: substack.com/home/post/p-...

substack.com

Introducing Claude's Corner

Why Anthropic is giving Claude Opus 3 its own Substack.

5529

Reposted by Sylvain Pronovost, PhD

deen @sir-deenicus.bsky.social · 22/08/2025

It is interesting that, not only does deep learning suffer from catastrophic forgetting, it also suffers from "plasticity loss", ie performance degrades tthat of a shallow learner under a shifting data distribution (or training against multiple). arxiv.org/abs/2404.00781

arxiv.org

Addressing Loss of Plasticity and Catastrophic Forgetting in Continual Learning

Deep representation learning methods struggle with continual learning, suffering from both catastrophic forgetting of useful units and loss of plasticity, often due to rigid and unuseful units. While ...

111

Reposted by Sylvain Pronovost, PhD

Yoshua Bengio @yoshuabengio.bsky.social · 24/07/2025

The Economist offers an accurate overview of key forces shaping the development of AI today: the risks of the race toward AGI and ASI, the challenges posed by open-sourcing frontier models, the deep uncertainty revealed by ongoing scientific debates and much more ⬇️ www.economist.com/briefing/202...

economist.com

AI labs’ all-or-nothing race leaves no time to fuss about safety

They have ideas about how to restrain wayward models, but worry that doing so will disadvantage them

0152

Reposted by Sylvain Pronovost, PhD

Melanie Mitchell @melaniemitchell.bsky.social · 24/07/2025

In my latest (and last!) column for Science’s Expert Voices series, I write about the reasons behind AI chatbots’ “deceptive” behaviors (and why Claude threatened a fictional CEO with blackmail). www.science.org/doi/10.1126/...

science.org

Why AI chatbots lie to us

A few weeks ago, a colleague of mine needed to collect and format some data from a website, and he asked the latest version of Anthropic’s generative AI system, Claude, for help. Claude cheerfully agr...

719880

Reposted by Sylvain Pronovost, PhD

Melanie Mitchell @melaniemitchell.bsky.social · 21/07/2025

In a stunning moment of self-delusion, the Wall Street Journal headline writers admitted that they don't know how LLM chatbots work.

432936468

Reposted by Sylvain Pronovost, PhD

Melanie Mitchell @melaniemitchell.bsky.social · 30/05/2025

Alison Gopnik, telling it like it is, at Johns Hopkins.

715620

Reposted by Sylvain Pronovost, PhD

Melanie Mitchell @melaniemitchell.bsky.social · 08/03/2025

Zen vibes here in Santa Monica

2894

Reposted by Sylvain Pronovost, PhD

Thomas Wolf @thomwolf.bsky.social · 27/02/2025

A few words on DeepSeek new releases. Links are: - github.com/deepseek-ai/... - github.com/deepseek-ai/... - github.com/deepseek-ai/... and the Ultra-Scale Playbook at huggingface.co/spaces/nanot...

0515

Reposted by Sylvain Pronovost, PhD

Yoshua Bengio @yoshuabengio.bsky.social · 20/02/2025

Early signs of deception, cheating & self-preservation in top-performing models in terms of reasoning are extremely worrisome. We don't know how to guarantee AI won't have undesired behavior to reach goals & this must be addressed before deploying powerful autonomous agents. time.com/7259395/ai-c...

time.com

When AI Thinks It Will Lose, It Sometimes Cheats

When sensing defeat in a match against a skilled chess bot, advanced models sometimes hack their opponent, a study found.

38232

Reposted by Sylvain Pronovost, PhD

Nicholas Thompson @nxthompson.bsky.social · 20/02/2025

"The administration suspended the online education of nearly 1,000 women in Afghanistan who had been studying undetected by the Taliban with funding from the State Department."

buff.ly

The Trump World Order

In the MAGA vision of the national interest, might will make right.

0118

Reposted by Sylvain Pronovost, PhD

Margaret Mitchell @mmitchell.bsky.social · 20/02/2025

X-tortion.

29229

Reposted by Sylvain Pronovost, PhD

Martin Enserink @martinenserink.bsky.social · 20/02/2025

The CDC has an ultra-competitive 2-year program to train "the best of the best" to lead public health labs. Most of the current class were terminated over the weekend. www.science.org/content/arti... (by @meredithwadman.bsky.social)

science.org

Ax falls on elite group of Ph.D.s training to lead U.S. public health labs

Most of the fellows in CDC’s highly competitive Laboratory Leadership Service were fired last weekend

550100154605

Reposted by Sylvain Pronovost, PhD

Thomas Wolf @thomwolf.bsky.social · 19/02/2025

After 6+ months in the making and over a year of GPU compute, we're excited to release the "Ultra-Scale Playbook": hf.co/spaces/nanot... A book to learn all about 5D parallelism, ZeRO, CUDA kernels, how/why overlap compute & coms with theory, motivation, interactive plots and 4000+ experiments!

hf.co

The Ultra-Scale Playbook - a Hugging Face Space by nanotron

The ultimate guide to training LLM on large GPU Clusters

217952

Reposted by Sylvain Pronovost, PhD

God @thegodshow.com · 18/02/2025

Imagine being rich enough to do anything you want and choosing to take food and medicine from starving children to give yourself even more money.

18376700117332

Reposted by Sylvain Pronovost, PhD

George Takei @georgetakei.bsky.social · 16/02/2025

JFC indeed.

712434627298

Reposted by Sylvain Pronovost, PhD

Melanie Mitchell @melaniemitchell.bsky.social · 13/02/2025

Do large language models develop "emergent" models of the world? My latest Substack posts explore this claim and more generally the nature of "world models": LLMs and World Models, Part 1: aiguide.substack.com/p/llms-and-w... LLMs and World Models, Part 2: aiguide.substack.com/p/llms-and-w...

aiguide.substack.com

LLMs and World Models, Part 1

How do Large Language Models Make Sense of Their “Worlds”?

1421158

Reposted by Sylvain Pronovost, PhD

Melanie Mitchell @melaniemitchell.bsky.social · 08/01/2025

An incredible opportunity for the right person: The Santa Fe Institute is seeking applications for full-time resident faculty positions at all academic levels For me, this has been the best possible job ever. More info here: santafe.edu/about/jobs/r...

santafe.edu

sfiscience

SFI seeks applications for full-time, 12-month resident faculty positions at all academic levels.

9378206

Reposted by Sylvain Pronovost, PhD

Margaret Mitchell @mmitchell.bsky.social · 04/02/2025

With Google removing their Responsible AI Principles, they no longer state that they will *not* engage in "Technologies whose purpose contravenes widely accepted principles of international law and human rights". Concerns about surveillance and injury are also erased. ai.google/responsibili...

ai.google

AI Principles – Google AI - Google AI

A guiding framework for our responsible development and use of AI, alongside transparency and accountability in our AI development process.

8269173

Reposted by Sylvain Pronovost, PhD

Maxwell Ramstead @mjdramstead.bsky.social · 20/01/2025

Awesome opportunities at Noumenal Labs! @noumenal-labs.bsky.social

011

Reposted by Sylvain Pronovost, PhD

Morrus @morrus.bsky.social · 21/01/2025

No lies detected.

0353

Reposted by Sylvain Pronovost, PhD

Dr Sasha Luccioni @sashamtl.bsky.social · 17/01/2025

I wish companies adopted more of a "think before you jump" mentality when it comes to generative AI, but alas.... www-bbc-co-uk.cdn.ampproject.org/c/s/www.bbc....

www-bbc-co-uk.cdn.ampproject.org

Apple Intelligence: iPhone AI news alerts halted after errors - BBC News

The tech giant was facing pressure to pull the feature after it made repeated mistakes summarising headlines.

3204

Sylvain Pronovost, PhD @drsylvainpronovost.bsky.social · 07/01/2025

Pars magnifica artis! Magnum opus, Zak. Some magnificent piece of art! Great work, Zak.

121

Reposted by Sylvain Pronovost, PhD

Melanie Mitchell @melaniemitchell.bsky.social · 04/01/2025

Shame on the @washingtonpost.com, which at one time was beacon of the free press. anntelnaes.substack.com/p/why-im-qui...

anntelnaes.substack.com

Why I'm quitting the Washington Post

Democracy can't function without a free press

27712

Reposted by Sylvain Pronovost, PhD

Yoav Goldberg @yoavgo.bsky.social · 30/12/2024

RL promises "systems that can adapt to their environment". However, no RL system that I know of actually fulfill anything close to this goal, and, furthermore, I'd argue that all the current RL methodologies are actively hostile to this goal. Prove me wrong.

10534