David Scott "Lightman" Lewis

One pipeline. Many frontiers: climate intervention + drug discovery + finance + robotics … and much, much more.

@vlelyavin @karpathy Nope. We’re now maestros conducting a host of models, agents, frameworks, & tools. During work hours, we use 5.2 Pro with Extended thinking, Gemini 3 Pro with Deep Think, & Grok 4.1 Heavy, while we plan what to hand off to Claude Code—especially for it to work on while we sleep.

@karpathy Andrej is the type of person who drops a post like this and people actually read it thoroughly til the end "it hurts the ego a bit" is the most relatable line here imo. we spent years learning how to code and now we're just project managers for an intern that never sleeps

A few random notes from claude coding quite a bit last few weeks. Coding workflow. Given the latest lift in LLM coding capability, like many others I rapidly went from about 80% manual+autocomplete coding and 20% agents in November to 80% agent coding and 20% edits+touchups in December. i.e. I really am mostly programming in English now, a bit sheepishly telling the LLM what code to write... in words. It hurts the ego a bit but the power to operate over software in large "code actions" is just too net useful, especially once you adapt to it, configure it, learn to use it, and wrap your head around what it can and cannot do. This is easily the biggest change to my basic coding workflow in ~2 decades of programming and it happened over the course of a few weeks. I'd expect something similar to be happening to well into double digit percent of engineers out there, while the awareness of it in the general population feels well into low single digit percent. IDEs/agent swarms/fallability. Both the "no need for IDE anymore" hype and the "agent swarm" hype is imo too much for right now. The models definitely still make mistakes and if you have any code you actually care about I would watch them like a hawk, in a nice large IDE on the side. The mistakes have changed a lot - they are not simple syntax errors anymore, they are subtle conceptual errors that a slightly sloppy, hasty junior dev might do. The most common category is that the models make wrong assumptions on your behalf and just run along with them without checking. They also don't manage their confusion, they don't seek clarifications, they don't surface inconsistencies, they don't present tradeoffs, they don't push back when they should, and they are still a little too sycophantic. Things get better in plan mode, but there is some need for a lightweight inline plan mode. They also really like to overcomplicate code and APIs, they bloat abstractions, they don't clean up dead code after themselves, etc. They will implement an inefficient, bloated, brittle construction over 1000 lines of code and it's up to you to be like "umm couldn't you just do this instead?" and they will be like "of course!" and immediately cut it down to 100 lines. They still sometimes change/remove comments and code they don't like or don't sufficiently understand as side effects, even if it is orthogonal to the task at hand. All of this happens despite a few simple attempts to fix it via instructions in CLAUDE . md. Despite all these issues, it is still a net huge improvement and it's very difficult to imagine going back to manual coding. TLDR everyone has their developing flow, my current is a small few CC sessions on the left in ghostty windows/tabs and an IDE on the right for viewing the code + manual edits. Tenacity. It's so interesting to watch an agent relentlessly work at something. They never get tired, they never get demoralized, they just keep going and trying things where a person would have given up long ago to fight another day. It's a "feel the AGI" moment to watch it struggle with something for a long time just to come out victorious 30 minutes later. You realize that stamina is a core bottleneck to work and that with LLMs in hand it has been dramatically increased. Speedups. It's not clear how to measure the "speedup" of LLM assistance. Certainly I feel net way faster at what I was going to do, but the main effect is that I do a lot more than I was going to do because 1) I can code up all kinds of things that just wouldn't have been worth coding before and 2) I can approach code that I couldn't work on before because of knowledge/skill issue. So certainly it's speedup, but it's possibly a lot more an expansion. Leverage. LLMs are exceptionally good at looping until they meet specific goals and this is where most of the "feel the AGI" magic is to be found. Don't tell it what to do, give it success criteria and watch it go. Get it to write tests first and then pass them. Put it in the loop with a browser MCP. Write the naive algorithm that is very likely correct first, then ask it to optimize it while preserving correctness. Change your approach from imperative to declarative to get the agents looping longer and gain leverage. Fun. I didn't anticipate that with agents programming feels *more* fun because a lot of the fill in the blanks drudgery is removed and what remains is the creative part. I also feel less blocked/stuck (which is not fun) and I experience a lot more courage because there's almost always a way to work hand in hand with it to make some positive progress. I have seen the opposite sentiment from other people too; LLM coding will split up engineers based on those who primarily liked coding and those who primarily liked building. Atrophy. I've already noticed that I am slowly starting to atrophy my ability to write code manually. Generation (writing code) and discrimination (reading code) are different capabilities in the brain. Largely due to all the little mostly syntactic details involved in programming, you can review code just fine even if you struggle to write it. Slopacolypse. I am bracing for 2026 as the year of the slopacolypse across all of github, substack, arxiv, X/instagram, and generally all digital media. We're also going to see a lot more AI hype productivity theater (is that even possible?), on the side of actual, real improvements. Questions. A few of the questions on my mind: - What happens to the "10X engineer" - the ratio of productivity between the mean and the max engineer? It's quite possible that this grows *a lot*. - Armed with LLMs, do generalists increasingly outperform specialists? LLMs are a lot better at fill in the blanks (the micro) than grand strategy (the macro). - What does LLM coding feel like in the future? Is it like playing StarCraft? Playing Factorio? Playing music? - How much of society is bottlenecked by digital knowledge work? TLDR Where does this leave us? LLM agent capabilities (Claude & Codex especially) have crossed some kind of threshold of coherence around December 2025 and caused a phase shift in software engineering and closely related. The intelligence part suddenly feels quite a bit ahead of all the rest of it - integrations (tools, knowledge), the necessity for new organizational workflows, processes, diffusion more generally. 2026 is going to be a high energy year as the industry metabolizes the new capability.

@pmddomingos No, no, no. Gore invented the Internet.

No, Jurgen Schmidhuber didn't invent AI. Al Gore did.

@emollick Nothing new -- for those of us in the e/max cohort -- but nice to see an academic exploration of this topic.

Some useful findings: 1) Working with AI boosts the performance of people solving math, science & ethics questions 2) The biggest boost is for the hardest problems 3) High performers remain highest performing, but low performers gain more 4) People who are good with AI gain most

We encourage you to explore what can be done with AI beyond using frontier models ...

One pipeline. Many frontiers: climate intervention + drug discovery + finance + robotics … and much, much more.

Although a bit late to the game (we weren't aware of this NeurIPS workshop until a few days b4 the deadline & devoted 24/7 to fleshing out our proposal), we're delighted to be presenting a poster at AI4D3, AI Virtual Cells & Instruments: A New Era in Drug Discovery & Development.

Our agentic pipeline includes models, frameworks, and tools from Stanford, Oxford, Yale, Berkeley, Duke, LANL (Los Alamos), ANL (Argonne), Nvidia, Microsoft, ... -- plus some of our own secret sauce that greatly improves upon a mere combination of "goodies".

We gave a talk on Monday afternoon at the AI for Science international conference, specifically in their AI & Environmental Science SMASHtrack.

@emollick How about a 2.5 Pro Deep Research & 2.5 Pro Deep Think? Sans Codex, I also run all three OpenAI offerings simultaneously, along w the two Gemini offerings just mentioned. Often I'll toss in Grok 4 DeepSearch in Expert mode (free) & Claude Sonnet 4 (unlimited) for good measure.

I have a Codex project, a Deep Research run, a ChatGPT agent task, & a GPT-5 Pro task all going at the same time. Is this that agent managing thing that some people keep saying is supposed to be the future of work? The current AI interfaces aren't really up to the task.

@kimmonismus MM, it's called NotebookLM Pro.

This clip showed me once again how much we live in an AI bubble here. The vast majority of people have no idea what developments are currently taking place and what AI is already capable of today.

Paper: From Brute Force to Brain Power: How Stanford's s1 Surpasses DeepSeek-R1 Stanford's s1 introduces a streamlined, data-efficient method that surpasses previous open-source and open-weights reasoning models-most notably DeepSeek-R1. papers.ssrn.com/sol3/papers.cf…

The most important book to read right now. Not just a "must read," but a "read right now." Asia's New Geopolitics: Essays on Reshaping the Indo-Pacific by Hoover's Michael Auslin. Amazon dropped the price of the Kindle version to only USD 6.99, a 77% di…lnkd.in/g-u3GE7

And just imagine that I thought these were all observations from CIA's China Desk. Sounds like something I heard at an Aspen Forum a couple/few years back--said by some Asia regional head at Langley. lnkd.in/gTUrQ8E lnkd.in/gzT_5bF

ICYMI. lnkd.in/gAZ3_vT (PDF, 6 pp.) lnkd.in/gGXmVmy

This is EXACTLY what the NSF should be doing and the areas given priority--artificial intelligence, quantum information science, and computationally intensive research--are probably the best choices. Not all research areas are created equal. lnkd.in/gUea_eR

ICYMI. "Understanding when and how states may lawfully deploy countermeasures is critical for states operating in the cyber arena—not only to understand their own options when injured but also to anticipate the responses that thei…lnkd.in/g5xwAG3 lnkd.in/gHz4W6P

A must read authored by Kenneth Geers, a great guy, true scholar, and experienced cyber warrior. lnkd.in/gAgaACu

Let's be honest, we all know that the FT, the Financial Times, is the best newspaper in the world. The only problem: It's way too expensive! It's sad to experience the intellectual death of the New York Times. A country with the size and wealth of Americ…lnkd.in/gfRAFSV

Yesterday. Hearing on “Defending Against Future Cyberattacks: Evaluating the Cyberspace Solarium Commission Recommendations,” Subcommittee on Cybersecurity, Infrastructure Protection, and Innovation (Committee on Homeland Security…lnkd.in/gUf-key lnkd.in/g9zkWVA

A short video on DARPA's new FETT (Finding Exploits to Thwart Tampering) Bug Bounty. lnkd.in/gksmviz