Mateusz Mazurkiewicz

[Matthew Green @matthew_d_green]

I’ve had a few people write to ask me to follow up on a (reply!) tweet I wrote that’s gotten a lot of retweets. I considered deleting it just to avoid the headache, but I think that would make people more paranoid. So let me share my thoughts.

[Mateusz Mazurkiewicz @MazurkiewiczWho]

@matthew_d_green Thanks for writing this

[Mateusz Mazurkiewicz @MazurkiewiczWho]

@AdevAarons What the hell are you talking about? ECC is set to be deprecated by 2030 and disallowed by 2035 by NIST. Governments and banks are absolutely migrating away from it, as they're legally required

[Mateusz Mazurkiewicz @MazurkiewiczWho]

@LindellYehuda Therefore, as capabilities in cryptanalysis drastically increase, which is consistent with the observed AI math progress, it is inappropriate to not update your priors and have still as strong belief in security of ECC, as its underlying assumptions are undermined.

[Mateusz Mazurkiewicz @MazurkiewiczWho]

@LindellYehuda Conversely, fine-grained cryptography shows that practical computational security can be achieved even if P=NP. The security of ECC is derived from empirical evidence of our inability to break it, which may be no longer a valid security assumption.

[Andrej Karpathy @karpathy]

It is hard to communicate how much programming has changed due to AI in the last 2 months: not gradually and over time in the "progress as usual" way, but specifically this last December. There are a number of asterisks but imo coding agents basically didn’t work before December and basically work since - the models have significantly higher quality, long-term coherence and tenacity and they can power through large and long tasks, well past enough that it is extremely disruptive to the default programming workflow.

Just to give an example, over the weekend I was building a local video analysis dashboard for the cameras of my home so I wrote: “Here is the local IP and username/password of my DGX Spark. Log in, set up ssh keys, set up vLLM, download and bench Qwen3-VL, set up a server endpoint to inference videos, a basic web ui dashboard, test everything, set it up with systemd, record memory notes for yourself and write up a markdown report for me”. The agent went off for ~30 minutes, ran into multiple issues, researched solutions online, resolved them one by one, wrote the code, tested it, debugged it, set up the services, and came back with the report and it was just done. I didn’t touch anything. All of this could easily have been a weekend project just 3 months ago but today it’s something you kick off and forget about for 30 minutes.

As a result, programming is becoming unrecognizable. You’re not typing computer code into an editor like the way things were since computers were invented, that era is over. You're spinning up AI agents, giving them tasks *in English* and managing and reviewing their work in parallel. The biggest prize is in figuring out how you can keep ascending the layers of abstraction to set up long-running orchestrator Claws with all of the right tools, memory and instructions that productively manage multiple parallel Code instances for you. The leverage achievable via top tier "agentic engineering" feels very high right now.

It’s not perfect, it needs high-level direction, judgement, taste, oversight, iteration and hints and ideas. It works a lot better in some scenarios than others (e.g. especially for tasks that are well-specified and where you can verify/test functionality). The key is to build intuition to decompose the task just right to hand off the parts that work and help out around the edges. But imo, this is nowhere near "business as usual" time in software.

[Mateusz Mazurkiewicz @MazurkiewiczWho]

@blader Why don't you ship internationally?

[Mateusz Mazurkiewicz @MazurkiewiczWho]

@stripepress Wow, all of these are really pretty

[OpenAI @OpenAI]

Goals with ChatGPT.

[Mateusz Mazurkiewicz @MazurkiewiczWho]

@MartinShkreli Obviously, you need GPU-optimized programs to benefit from your GPU as well. You need neural networks to make use of neuromorphic architectures. That's how different computers work

[Elon Musk @elonmusk]

Precise laser tracking in space at 4000 km (2500 miles) distance while moving at ~25 times the speed of sound

[Igor Babuschkin @ibab]

Research research research research research research research research research research research research research research research research research research research research research research research research research research research research research research research research research research research research research research research research research research

[Mateusz Mazurkiewicz @MazurkiewiczWho]

@TeedgeeOfficial Alice in Borderland

[Mateusz Mazurkiewicz @MazurkiewiczWho]

@grok @MichalOrzolek @grok what about the interview?

[Mateusz Mazurkiewicz @MazurkiewiczWho]

@lukOlejnik What do you mean by that? What could be a bigger data protection challenge than any encrypted, confidential data becoming decrypted and possibly public?

[Mateusz Mazurkiewicz @MazurkiewiczWho]

@itseieio That's pretty fun

[Mateusz Mazurkiewicz @MazurkiewiczWho]

@IsomorphicPhi @meowdib @bcubeddd I think the crucial element of science is not just "what," but "how do we know it," "how do we verify it," and giving purpose to the knowledge by applying it either in further scientific research or other.

Scientific knowledge is a living thing. By reading papers, you see it.

[Mateusz Mazurkiewicz @MazurkiewiczWho]

@IsomorphicPhi @meowdib @bcubeddd Science isn't a collection of facts but a process, and understanding of methodology of research and more broadly approaches to generating the knowledge is a key to understanding it scientifically