fabio bonsignorio retweeted
The problem isn’t that AI is intelligent. It isn’t. The problem is that too many adults (but not all) are at once mechanistic, imitative, unimaginative, stupid, formulaic, influenceable, & gullible, arrogant beyond their skill set, & too expensive to maintain.
fabio bonsignorio retweeted
Proceedings of the 4th (2025) NeurIPS Workshops on Symmetry and Geometry in Neural Representations, PMLR 282:263-281, 2026.
"Hilbert geometry of the symmetric positive-definite bicone."
proceedings.mlr.press/v282/k…
fabio bonsignorio retweeted
AI is both the best and worst thing to ever happen to mathematics newscientist.com/article/258…
fabio bonsignorio retweeted
Starship achieved orbit on its first attempt and successfully delivered Starlink V3 satellites to space for the first time → spacex.com/launches/starship…
fabio bonsignorio retweeted
Demis Hassabis just told economists something they’re not built to hear.
For the first time in human history, winning might not require someone else to lose.
Economists keep asking him the same question.
Where’s the impact.
Where’s the GDP.
He’s not confused by their question. He’s confused by their yardstick.
GDP measures a world of things that run out. Labor. Land. Output.
Hassabis: “It’s 10 times the Industrial Revolution.”
The Industrial Revolution didn’t just change output. It changed where humans lived, what work meant, what childhood was.
Ten times that. And they’re checking it against a ledger built for a world already gone.
Capitalism. Socialism. Communism. Feudalism.
Every economic system we’ve ever run shares one buried assumption.
Not enough to go around.
Take that assumption away and you don’t retire a model. You retire the fight underneath all of them.
Because scarcity never just built markets. It built us.
Status. Tribalism. The instinct to rank, to hoard, to compare, to defend.
Wired in over a hundred thousand years of a zero-sum world.
Someone’s win was usually someone else’s loss.
That’s the crack Hassabis says is coming. First time. Ever.
Most people hear that and reach for collapse.
Not this time.
A species that has spent its entire existence fighting over scraps.
Just found a way to make the scraps beside the point.
That’s not an ending.
Keynes rebuilt economics once before.
When the old model broke against the Depression.
Someone alive right now has to do it again. This time for a world where intelligence itself stops being scarce.
Nobody has taken that job yet.
Every ideology we’ve ever built was a fight over what’s scarce.
We’re about to find out what’s left to fight about when scarcity isn’t the reason anymore.
Starting with the fight over who you are without it.
fabio bonsignorio retweeted
The ONLINE and IN PERSON registration for ERAS 2026 2026.ieee-eras.org/ is still possible.
The conference will start on May 28th at 9:00 CEST.
Program here: 2026.ieee-eras.org/program/
Details here: ras.papercept.net/conference…
Link to registr: web.cvent.com/event/ea5d3378…
CU soon!!!
fabio bonsignorio retweeted
Also what I've seen.
I think this image illustrates the capabilities of large language models very effectively.
LLMs are great at recombining existing knowledge. So, for questions outside your domain of expertise, or far from the frontier of knowledge, they are often much better than the average human. Here they can be incredibly helpful.
However, as you move closer to the frontier of knowledge, they become much worse.
Here, even the average human can become better.
I have seen this many times with my own eyes. When I work with an LLM at the frontier of knowledge, it often makes absurd mistakes that no intelligent person would make.
Internal contradictions within a few lines, dramatic forgetting of what happened two interactions earlier, and so on.
This limitation is literally built into the model: it approximates the most likely continuation given the previous input.
If there is enough relevant structure in the training data, it can perform very well. If it does not really know where to go, the output quickly becomes messy, and randomness takes over.
fabio bonsignorio retweeted
Good morning, world! 🌎
We have spectacular new high-resolution images of our home planet, all of us looking back through the Orion capsule window at our Artemis II astronauts as they continue their journey to the Moon.
ALT One-third of Earth peeking through the window of the Orion capsule. The planet is a dreamy pale blue, swirling with white clouds and reflected sunlight. Although Earth only fills a fraction of the image it is the brightest object in the image by far. The capsule window is surrounded by a thick frame held in place with bolts, reminiscent of a heavy duty airplane window. It is dark in the capsule, but the outlines of straps and various components of the capsule are visible. Brighter white components are visible in the upper right corner. Credit: NASA/Reid Weisman
fabio bonsignorio retweeted
The Orion spacecraft successfully separated from the upper stage of the rocket, and the "proximity operations" test is underway. The Artemis II astronauts are manually piloting Orion similarly to how they would if they were docking with another spacecraft.
fabio bonsignorio retweeted
Very excited about the prospect of Code-as-Policy (CaP) for Robotics!
Esp with recent rapid advances in agentic coding. CaP has potential to quickly combine VLA models with GOFE primitives into interpretable code, observe experiments, and iterate. Initial results are promising:
Robotics: coding agents’ next frontier.
So how good are they?
We introduce CaP-X: an open-source framework and benchmark for coding agents, where they write code for robot perception and control, execute it on sim and real robots, observe the outcomes, and iteratively improve code reliability.
From @NVIDIA @Berkeley_AI @CMU_Robotics @StanfordAILab
capgym.github.io
🧵
fabio bonsignorio retweeted
Liftoff.
The Artemis II mission launched from @NASAKennedy at 6:35pm ET (2235 UTC), propelling four astronauts on a journey around the Moon.
Artemis II will pave the way for future Moon landings, as well as the next giant leap — astronauts on Mars.
fabio bonsignorio retweeted
The weather's looking good for tomorrow's Artemis II launch, and our teams are getting the rocket ready for liftoff!
Read the latest updates on our mission around the Moon: go.nasa.gov/4tiFY4P
fabio bonsignorio retweeted
Transfer and lifelong adaptation remain central challenges for physical AI. Unlike LLMs, which benefit from more standardized interfaces, the complexity of robotic embodiments, environments, and human interactions - combined with limited pretraining data - renders pure generalization insufficient.
This roadmap by @Ken_Goldberg @davscaramuzza @aschoellig @RavinderSDahiya @petercorke @siddssrinivasa & Aude Billard nicely points out this and other key hurdles: arxiv.org/pdf/2507.19975
To dive deeper into the mechanisms for this, check out our survey (arxiv.org/abs/2312.01939) which summarizes the mechanisms and knowledge types needed to address this. While we only had a small section on the shift towards VLAs, the core ideas themselves transfer directly to this new paradigm! 🤖🚀
(And for the nano banana fans:)
fabio bonsignorio retweeted
Studying generalist reward models is hard: robot datasets focus on successful demos, not failures.
We introduce:
- a large-scale reward modeling benchmark
- a data augmentation scheme
- a generalist reward model that outperforms frontier VLMs
Paper: arxiv.org/abs/2601.00675
Reliable rewards are a bottleneck for real-world RL for robotics: human labels are costly, and handcrafted rewards are brittle.
In RoboReward 🤖💰, we study VLMs as reward models and find they are unreliable across tasks, embodiments, and scenes.
Paper: arxiv.org/abs/2601.00675
fabio bonsignorio retweeted
Humanoid robots are coming, but don't hold your breath waiting for them:
news.berkeley.edu/2025/08/27…
fabio bonsignorio retweeted
I agree. Either a risk taker will discover a breakthrough analogous to relativity or the DNA double-helix structure or there may be a lengthy plateau of incremental progress. We’re all placing bets and like baseball, there’s no clock on the game.
fabio bonsignorio retweeted
'Hamilton' turned 10 this summer. In @Reuters latest Culture Current, Leslie Odom Jr. speaks about returning to Aaron Burr, the show’s cultural legacy, and why it still resonates in 2025. Read the full Q&A: reut.rs/46nHOce
