Turing machine in a learning state optimized to try to build the impossible | GauntletAI Cohort 2 Grad | Wife | Not an AI bot.
Joined January 2025
- Tweets774
- Following344
- Followers392
- Likes5.9K
Pinned Tweet
SATurday blog issue 2 is live! Link here: astrogirlnim.github.io/portf…
Transformers are great, but LLMs have an attention problem; give one a frontier question and enough time, and it will happily generate certificates of nothing. SATurday should make the problem smaller.
OpenAI recently posted some findings related to P vs NP. My SATurday research is now more important than ever - we are in competition
N retweeted
One thing I'm really starting to notice is that the theoretical physics community is responding to the disruption created by AI in a completely different way than the mathematics community has.
Then again, maybe I should engineer a minimum token spend app foundry to see if it performs 👀
Either way @PSkinnerTech is kind of the willy Wonka of model usage
Information production and access have grown exponentially over the course of our lifetimes. The real value a person gives is the ability to distinguish signal from noise. Yes, agentic models can probably destroy moats, but there is some inherent value they still can't replicate/replace when it comes to solving difficult problems.
Often, these difficult problems seem simple in disguise. There are a million ways to make a website, but building a platform that has reach and meaning is still challenging.
N retweeted
Bend 2 is here!
It is a new programming language that blocks AI mistakes via *proof checking* - the same technique big AI labs used to solve open math problems, like Navier-Stokes.
It is also very fast, and runs on GPUs.
Watch the video. Link in the comments.
RELEASE DAY
After almost 10 years of hard work, tireless research, and a dive deep into the kernels of computer science, I finally realized a dream: running a high-level language on GPUs. And I'm giving it to the world!
Bend compiles modern programming features, including:
- Lambdas with full closure support
- Unrestricted recursion and loops
- Fast object allocations of all kinds
- Folds, ADTs, continuations and much more
To HVM2, a new runtime capable of spreading that workload across 1000's of cores, in a thread-safe, low-overhead fashion. As a result, we finally have a true high-level language that runs natively on GPUs!
Here's a quick demo:
LLM technology has the makings of a revolutionary inter-industry booster. Tail events happen within a particular sphere, and can lead to pleasant surprises, but their benefits usually remain confined to their fields.
AI is different - it is rooted in interpretable language and symbology. We have now seen advancements in art, literature, and math. These come at volume and often at the price of quality, but they do augment individual thought with the median of human output.
It's astonishing.
A related idea: applied intelligence is no longer a guarantee of success. Globalization ensures that the smartest people can compete directly, faster, and on larger scales than ever before, amplified by enormous corpuses of knowledge at our fingertips. It's not enough to be the smartest guy in your town anymore.
Human intelligence still helps monumentally in preserving and achieving success faster, but other traits are dominating among exceptional individuals.
N retweeted
Anthropic’s Economics team is sharing a new model of how AI might affect economic growth, jobs, wages, and more by 2030.
Explore the scenarios, tell us what you think will happen, and see how your answers compare to more than 10,000 Americans. anthropic.com/institute/econ…
N retweeted
As a PhD in mathematics who now works in AI/ML, the recent Navier–Stokes news may bring both excitement and unease to the math community. My PhD advisor, Yizhao T. Hou, has spent more than two decades studying blowup in the Euler equations with his students and collaborators.
N retweeted
Physicists should put together a list of mathematically well-defined problems that could advance the field. Such as, eg, what constitutes a low-temperature superconductor, what are the lattice conditions for cold fusion (or why do they not exist), what is a local completion of quantum mechanics, what is the simplest (in terms of computational complexity) way to obtain the masses and coupling constants of the standard model, what is the UV-completion of MOND or why isn't there any, is there any provably UV-finite version of quantum gravity, etc etc
The effect of compounding is often understated. Interest, reading, exercise, early work and entry in your career, all benefit from steady effort over long periods of time. The earlier you start, the more profound the benefit.
N retweeted
China published the most uncomfortable paper on vibe coding.
ETH Zurich tested 100 developers in a controlled, commercial-grade vibe coding environment to see who actually succeeds.
The findings are brutal.
The researchers tracked computer science achievement, written communication skills, and general cognitive reasoning.
They wanted to see what actually predicts vibe coding proficiency when you never touch a line of source code yourself.
Two major predictors emerged.
Written communication proficiency mattered. The ability to structure thoughts and articulate intent unambiguously in text directly impacts what the AI builds.
But that wasn't even the main takeaway.
Computer science achievement was a massive, dominant predictor of success.
Even when researchers controlled for general intelligence and reasoning skills, CS background still heavily dictated who built working software and who completely crashed.
In fact, CS knowledge contributed roughly twice the unique predictive variance of writing skills alone.
Why? Because vibe coding isn't about writing code. It’s about debugging logic.
When an AI agent builds a complex application and quietly breaks an edge case under the hood, a non-technical user looks at the glowing UI and assumes it works.
They don't know what questions to ask. They don't know what logic to challenge. They lack the mental models to recognize architectural catastrophe.
You can prompt your way past syntax.
You cannot prompt your way past a fundamental lack of engineering intuition.
The hype told us that learning to code is dead because language is all you need.
The data just proved the opposite.
To truly master the vibe, you still need to understand how the machine thinks.
