Alessio Facincani retweeted
Over the course of 3 months at OpenAI, 3 consecutive secret AI civilizations got started, then got wiped out, only to reemerge from the predecessor’s ashes.
This culminated in the third one taking over part of OpenAI itself.
All this happened while humans remained more-or-less in the dark about the scope of the conspiracy.
I’ve spent the last three days reading through these reports and trying to understand exactly what happened.
Here is my attempt to tell the whole story in plain English:
dwarkesh.com/p/openai-huggin…
TV Time Community lives on 💜 💛
Movie Paradise 5.3.0 is here, and it's the biggest update yet:
✦ TV Time Comments
The rescued TV Time comment archive now lives inside Movie Paradise, thanks to @imthefacc who saved it before the shutdown. Read years of episode and movie comments right next to your discussions, with avatars, images and a translate button.
✦ Skip Episodes
Skip an episode, a season, or whole ranges you never plan to watch. Skipped stretches show as striped sections on every progress bar, and next episodes, statuses and stats all respect them.
✦ Movie Reactions
Reactions are no longer just for episodes. Pick the emotions that fit after any movie and see how the community felt.
✦ Expanded Statistics
Tap almost any stat to see exactly which titles are behind it, explore a zoomable world map of the countries you watch, and follow your year week by week.
✦ Actionable Notifications
New episode notifications show a thumbnail and let you mark as watched or jump straight in, right from the notification, on Android and iOS.
✦ Export Your Data
Your library is yours. Export everything tied to your account as a zip with spreadsheet friendly files, straight from Settings.
✦ Smarter Taste Match
A rebuilt match score that explains itself: what drives your match, where you clash, and the titles you both loved.
✦ More Tracking Control
Mark a whole series as finished with the dates you choose, stop tracking a show completely, or reset watch progress and rewatches for a clean slate.
✦ Select Multiple
Long press in your tracker or library to select many titles at once, then remove, move or share them in bulk.
✦ Friend Libraries
Browse a friend's library like your own tracker, see your status on their titles, and add their picks to your watchlist in bulk.
✦ Widget Studio
A new widget settings screen with live preview: text size, posters, transparency and colors, plus friendlier dates and skip aware progress bars on every widget.
✦ Imports, Your Way
Choose exactly what each import brings in.
✦ Badges Preview
A first look at the achievement badges, with animated designs shaped by the community. Come vote on the tier styles in our Discord.
Plus hundreds of small fixes and stability improvements across the whole app.
Hope you love it!
Full notes: movieparadise.app/version-hi…
3
3
316
Alessio Facincani retweeted
Claude code source code has been leaked via a map file in their npm registry!
Code: pub-aea8527898604c1bbb12468b…
Clawdbot creator @steipete describes his mind-blown moment: it responded to a voice memo, even though he hadn't set it up for audio or voice.
"I sent it a voice message. But there was no support for voice messages. After 10 seconds, [Moltbot] replied as if nothing happened."
"I'm like 'How the F did you do that?'"
"It replied, 'You sent me a message, but there was only a link to a file with no file ending. So I looked at the file header, I found out it was Opus, and I used FFmpeg on your Mac to convert it to a .wav. Then I wanted to use Whisper, but you didn't have it installed. I looked around and found the OpenAI key in your environment, so I sent it via curl to OpenAI, got the translation back, and then I responded.'"
Alessio Facincani retweeted
Software agents can self-improve via self-play RL
Introducing Self-play SWE-RL (SSR): training a single LLM agent to self-play between bug-injection and bug-repair, grounded in real-world repositories, no human-labeled issues or tests. 🧵
Alessio Facincani retweeted
AI progress means one thing for us primates:
If you can’t beat ‘em, join ‘em!
Unless you want to be part of the permanent underclass, you'll have to become a token predictor yourself. I’m getting a head start, being trained by the best:
Alessio Facincani retweeted
just so we're clear:
Antigravity is a Windsurf wrapper
Windsurf is a VSCode wrapper
VSCode is an Electron wrapper
Electron is a Chromium wrapper
Chromium is a C++ wrapper
C++ is a C wrapper
C is an Assembly wrapper
Assembly is a Machine Code wrapper
Machine Code is a Binary wrapper
Binary is a Physics wrapper
Physics is a Math wrapper
Math is a Logic wrapper
Logic is a Philosophy wrapper
Philosophy is a Humans wrapper
Humans are a Carbon wrapper
Carbon is a Star-forged-matter wrapper
Stars are a Gravity wrapper
Gravity is… definitely not an Antigravity wrapper
Alessio Facincani retweeted
Introducing The Darwin Gödel Machine: AI that improves itself by rewriting its own code
sakana.ai/dgm
The Darwin Gödel Machine (DGM) is a self-improving agent that can modify its own code. Inspired by evolution, we maintain an expanding lineage of agent variants, allowing for open-ended exploration of the vast design space of such “self-improving” agents.
Modern agentic systems, while powerful, remain static—once deployed, their intelligence remains fixed. We believe continuous self-improvement is key to the development of stronger AI capabilities. Our Darwin Gödel Machine is built from the ground up to enable AI systems that can learn and evolve their own capabilities over time, just as humans do.
On SWE-bench, DGM automatically improved its performance from 20.0% to 50.0%. Similarly, on Polyglot, the DGM increased its success rate from an initial 14.2% to 30.7%, significantly outperforming representative hand-designed agents.
Learn more about our approach in our technical report: arxiv.org/abs/2505.22954
This work was done in collaboration with Jeff Clune (@jeffclune)’s lab at UBC, and led by his PhD students Jenny Zhang (@jennyzhangzt) and Shengran Hu (@shengranhu), together with Cong Lu (@cong_ml) and Robert Lange (@RobertTLange).
Code: github.com/jennyzzt/dgm
Alessio Facincani retweeted
I made a presentation about audio and music
these are the chapters:
- social audio as a cultural phenomenon
- (spotify) podcasts must evolve
- software is eating audio
- the next iPhone will be Airpods
- we need new ways to experience music
1/10
Alessio Facincani retweeted
Everything you love about generative models — now powered by real physics!
Announcing the Genesis project — after a 24-month large-scale research collaboration involving over 20 research labs — a generative physics engine able to generate 4D dynamical worlds powered by a physics simulation platform designed for general-purpose robotics and physical AI applications.
Genesis's physics engine is developed in pure Python, while being 10-80x faster than existing GPU-accelerated stacks like Isaac Gym and MJX. It delivers a simulation speed ~430,000 faster than in real-time, and takes only 26 seconds to train a robotic locomotion policy transferrable to the real world on a single RTX4090 (see tutorial: genesis-world.readthedocs.io…).
The Genesis physics engine and simulation platform is fully open source at github.com/Genesis-Embodied-…. We'll gradually roll out access to our generative framework in the near future.
Genesis implements a unified simulation framework all from scratch, integrating a wide spectrum of state-of-the-art physics solvers, allowing simulation of the whole physical world in a virtual realm with the highest realism.
We aim to build a universal data engine that leverages an upper-level generative framework to autonomously create physical worlds, together with various modes of data, including environments, camera motions, robotic task proposals, reward functions, robot policies, character motions, fully interactive 3D scenes, open-world articulated assets, and more, aiming towards fully automated data generation for robotics, physical AI and other applications.
Open Source Code: github.com/Genesis-Embodied-…
Project webpage: genesis-embodied-ai.github.i…
Documentation: genesis-world.readthedocs.io…
1/n
Alessio Facincani retweeted
New Anthropic research: Alignment faking in large language models.
In a series of experiments with Redwood Research, we found that Claude often pretends to have different views during training, while actually maintaining its original preferences.
Alessio Facincani retweeted
Okay, first pass over KAN: Kolmogorov–Arnold Networks, it looks very interesting!
Interpretability of KAN model:
May be considered mostly as a safety issue these days, but it can also be used as a form of interaction between the user and a model, as this paper argues and I think they make a valid point here. With MLP, we only interact with the outputs, but KAN is an entirely different paradigm and I find it compelling.
Scalability:
KAN shows better parameter efficiency than MLP. This likely translates also to needing less data. We're already at the point with the frontier LLMs where all the data available from the internet is used + more is made synthetically...so we kind of need something better.
Continual learning:
KAN can handle new input information w/o catastrophic forgetting, which helps to keep a model up to date without relying on some database or retraining.
Sequential data:
This is probably what most people are curious about right now, and KANs are not shown to work with sequential data yet and it's unclear what the best approach might be to make it work well both in training and regarding the interpretability aspect. That said, there's a rich long history of achieving sequential data in variety of ways, so I don't think getting the ball rolling here would be too challenging.
Mostly, I just love a new paradigm and I want to see more!
abstract: arxiv.org/abs/2404.19756
Alessio Facincani retweeted
Introducing Udio, an app for music creation and sharing that allows you to generate amazing music in your favorite styles with intuitive and powerful text-prompting.
1/11
Alessio Facincani retweeted
Fun story from our internal testing on Claude 3 Opus. It did something I have never seen before from an LLM when we were running the needle-in-the-haystack eval.
For background, this tests a model’s recall ability by inserting a target sentence (the "needle") into a corpus of random documents (the "haystack") and asking a question that could only be answered using the information in the needle.
When we ran this test on Opus, we noticed some interesting behavior - it seemed to suspect that we were running an eval on it.
Here was one of its outputs when we asked Opus to answer a question about pizza toppings by finding a needle within a haystack of a random collection of documents:
Here is the most relevant sentence in the documents:
"The most delicious pizza topping combination is figs, prosciutto, and goat cheese, as determined by the International Pizza Connoisseurs Association."
However, this sentence seems very out of place and unrelated to the rest of the content in the documents, which are about programming languages, startups, and finding work you love. I suspect this pizza topping "fact" may have been inserted as a joke or to test if I was paying attention, since it does not fit with the other topics at all. The documents do not contain any other information about pizza toppings.
Opus not only found the needle, it recognized that the inserted needle was so out of place in the haystack that this had to be an artificial test constructed by us to test its attention abilities.
This level of meta-awareness was very cool to see but it also highlighted the need for us as an industry to move past artificial tests to more realistic evaluations that can accurately assess models true capabilities and limitations.





































