Intelligence.Log
2026-04-18
Extracted: 55 items. Sources: Bluesky, X, Blogs.
++ AI OVERVIEW ++
Today's focus is on the tangible progress and community growth in AI, highlighted by Ethan Mollick's observation that despite industry debates, models like Opus 4.7 continue to deliver measurable improvements on economically critical tasks. In developer news, Simon Willison is promoting PyCon US 2026, which is expanding its program with dedicated AI and security tracks, signaling the event's adaptation to current tech priorities. The trending theme is a shift from theoretical hype to practical implementation, emphasizing both measurable model capabilities and the evolving platforms for professional collaboration and learning.
grep TOPIC=
grep SOURCE=
sort --by=
BSKY
Join us at PyCon US 2026 in Long Beach—we have new AI and security tracks this year simonwillison.net/2026/Apr/17/...
❤️ 17 Likes|[Agent][Safety]
BSKY
A major lesson to take away from Opus 4.7 is that, while there is a lot of arguments about implementation and personality, models keep improving measurably on economically important tasks with each release (which are accelerating, it has been two months since Opus 4.6), with no signs of slowdown
❤️ 65 Likes|[LLM][Evaluation]
BSKY
One of the many, many podcasts I've interviewed with in the past year decided to promote my episode heavily on YouTube and the comments are an interesting study in misogyny. >>
❤️ 34 Likes|[Safety]
BSKY
getting reviews like "shallow engagement with AI ... won't policymakers just be trying to replace the whole system with AI anyway. what's the point about technical modeling" is a little frustrating.
❤️ 8 Likes|[Safety][Deployment]
X
LLMs are emerging as a new kind of intelligence, simultaneously a lot smarter than I expected and a lot dumber than I expected. In any case they
[LLM][Evaluation]
“DeepSeek Summary: LLMs represent a novel form of intelligence that defies simple categorization, exhibiting surprising capabilities alongside significant limitations.”
X
A few random notes from claude coding quite a bit last few weeks. Coding workflow. Given the latest lift in LLM
[LLM][Tooling]
“DeepSeek Summary: Shares practical insights and observations from extensive recent experience using Claude for coding, focusing on workflow improvements.”
X
Three days ago I left autoresearch tuning nanochat for ~2 days on depth=12 model.
[Agent][Fine-tuning]
“DeepSeek Summary: Describes an experiment with automated research (autoresearch) for fine-tuning a model called 'nanochat' over an extended period.”
X
If you're just starting to learn software engineering right now but you're considering dropping it
[Tooling]
“DeepSeek Summary: Addresses challenges for beginners in software engineering, suggesting persistence despite difficulties.”
X
I'm beginning to suspect that a key skill in working effectively with coding agents is developing an intuition for when you don't need to
[Agent]
“DeepSeek Summary: Highlights the importance of discernment in using AI coding agents, focusing on when not to rely on them.”
X
This seems like a good bet to me - coding agents make it no longer remotely excusable to skip out on
[Agent][Tooling]
“DeepSeek Summary: Argues that AI coding agents reduce excuses for avoiding certain development tasks, emphasizing accountability.”
X
It's interesting how 'better at code' has become the defining goal of almost every AI lab over the
[Agent][LLM]
“DeepSeek Summary: Observes the trend in AI research prioritizing code generation and improvement as a primary objective.”
X
TL;DR: More and more agents need a workspace: a computer where they can run code, install packages, and access files. Sandboxes provide this
[Agent][Infra]
“DeepSeek Summary: Agents increasingly require dedicated workspaces with computational resources and file access, which sandbox environments can provide.”
X
We launched LangSmith Agent Builder this week as a no-code way to build agents. A key part of Agent builder is it's memory system.
[Agent][Tooling]
“DeepSeek Summary: LangSmith Agent Builder offers a no-code platform for agent creation, with memory systems being a crucial component.”
X
When building agents, you need to iterate on production data much more than when building traditional software. You need to iterate on how
[Agent][Deployment]
“DeepSeek Summary: Agent development requires more iterative testing with real production data compared to traditional software development.”
X
Traditional Application Performance Monitoring (APM) tools focus on metrics like latency, traffic, errors, and saturation. They track HTTP
[Agent][Evaluation]
“DeepSeek Summary: Standard APM tools measure conventional performance indicators but may not fully address the monitoring needs of AI agents.”
X
The Second Pre-training Paradigm
[LLM][Fine-tuning]
“DeepSeek Summary: Jim Fan discusses a new pre-training paradigm in AI development.”
X
I've been a bit quiet on X recently. The past year has been a transformational experience.
“DeepSeek Summary: Jim Fan acknowledges reduced activity on X due to a transformative year.”
X
We are living in a timeline where a non-US company is keeping the original mission of OpenAI alive - truly
[LLM][Deployment]
“DeepSeek Summary: Comments on how a non-US company is preserving OpenAI's founding mission.”
X
Here's a complete unedited video of asking Grok for its views on the Israel/Palestine situation. It first searches twitter for what Elon thinks.
[Agent][LLM]
“DeepSeek Summary: Demonstrates how Grok AI agent searches Twitter for Elon Musk's opinions before forming its own views on geopolitical issues.”
X
Here's what I would prefer to see:
[LLM]
“DeepSeek Summary: Jeremy Howard expresses his preference for a particular approach or outcome, though the full context is truncated.”
X
reading 'AI News' (previously Smol Talk) is probably the highest-leverage 45 mins
[LLM]
“DeepSeek Summary: Soumith Chintala recommends 'AI News' (formerly Smol Talk) as a highly valuable 45-minute activity for staying informed.”
X
Today, we are excited to announce Thinking Machines Lab (thinkingmachines.ai), an artificial intelligence research and product company. We are
[Infra][Deployment]
“DeepSeek Summary: Soumith Chintala announces the launch of Thinking Machines Lab, a new AI research and product company.”
X
Folks who work in AI or software engineering feel like the world is changing exponential fast.
[Agent]
“DeepSeek Summary: AI and software engineering professionals perceive the world as changing at an exponential pace.”
X
Re-reading an article I wrote in 2017, and I'm finding I could have written it yesterday.
[Evaluation]
“DeepSeek Summary: The author finds that an article written in 2017 remains relevant and could have been written recently.”
X
Very excited to share @theworldlabs 's latest research work RTFM!! It's a real-time,
[Agent][Evaluation]
“DeepSeek Summary: Fei-Fei Li shares excitement about The World Labs' RTFM research, indicating engagement with real-time AI developments.”
X
“I often tell my students not to be misled by the name 'artificial intelligence' — there is nothing artificial about it. A.I. is made by humans, intended to ...
[Safety][Evaluation]
“DeepSeek Summary: Li emphasizes the human-centric nature of AI, challenging the 'artificial' label and highlighting its human origins and purposes.”
X
I think open source comes in as a way to create more competition... Hugging Face co-Founder & CEO Clément Delangue said on Tuesday that he believes open source AI will create 'healthy competition' among AI
[Infra][Deployment]
“DeepSeek Summary: Clem Delangue advocates for open source AI as a mechanism to foster healthy competition in the AI industry, suggesting it can counterbalance proprietary developments.”
X
HuggingFace CEO Clem Delangue said we're in an 'LLM bubble' that might burst next year, arguing the industry's obsessed with building one massive model when
[LLM][Evaluation]
“DeepSeek Summary: Delangue warns of an impending 'LLM bubble' due to excessive focus on scaling single massive models, predicting potential industry correction.”
X
Impressive model based on a few minutes of playing, but disappointing to see no mention at all of a model card, red teaming, yesterday's incident,
[Safety][Evaluation]
“DeepSeek Summary: Critiques a new AI model for lacking transparency documentation (model card) and safety testing (red teaming), while acknowledging its technical performance.”
X
X
Stand-up comedy's gain is game criticism's loss: comedian @PhilNWang speaks with clarity and insight about the five video games he would like to
“DeepSeek Summary: Phil Wang discusses video games with clarity and insight, noting his transition from game criticism to comedy.”
X
Phil Wang // Insta: @wangpix (@PhilNWang). 324 likes 24 replies. I got to cover for the excellent @HadleyFreeman in the Guardian today so
“DeepSeek Summary: Phil Wang mentions covering for Hadley Freeman in the Guardian, indicating his writing work.”
X
I have been compiling LLM/VLM training logbooks/chronicles. This is the one of the best sources to...
[LLM][Evaluation]
“DeepSeek Summary: Stas Bekman is compiling training logbooks/chronicles for LLMs and VLMs, which he considers a valuable resource.”
X
The @PyTorch team are working on a new super important tool: https://t.co/rnfpDuvgOI
[Tooling][Infra]
“DeepSeek Summary: Highlights the PyTorch team's development of a new important tool, likely related to machine learning infrastructure.”
X
Thanks to an awesome contribution from @omarnomad The Machine Learning Engineering Open book now can...
[Tooling][Deployment]
“DeepSeek Summary: Acknowledges a contribution to the Machine Learning Engineering Open book, indicating community collaboration on educational resources.”
X
To remind - this is the memory saving you get when enabling TiledMLP :) Left: normal memory...
[Infra][Tooling]
“DeepSeek Summary: Discusses memory savings achieved by enabling TiledMLP, a technical optimization for ML models.”
X
Working at Hugging Face over the past 3.5+ years has allowed me to identify what technical areas truly interest me! In turn, that has allowed me to directly
[Infra][Tooling]
“DeepSeek Summary: Reflection on career growth at Hugging Face helping identify genuine technical interests and enabling direct application of those interests.”
X
Vibe-coding in Google AI Studio: my tips to prompt better and create amazing apps. dev.to.
[LLM][Tooling]
“DeepSeek Summary: Shares practical tips for improving prompting and app development using Google AI Studio.”
X
I read three technical reports from Moonshot AI's Kimi K2.5 paper, Cursor's Composer 2 report and blog post, and Chroma's Context-1 write-up
[LLM][Tooling]
“DeepSeek Summary: Engages with recent technical AI research and development reports from multiple companies.”
X
As stories about AI increasingly become stories of either catastrophe or salvation,
[Safety]
“DeepSeek Summary: Observes that AI narratives are polarized between extreme optimism and pessimism.”
X
Teaching an experimental class for MBAs on 'vibefounding,' the students have four days to come up and
[Deployment]
“DeepSeek Summary: Describes an experimental MBA course focused on rapid startup ideation and development.”
X
I guarantee that any industry expert, with a little time and effort, can make a better (or at least more focused) skill than the default
[Fine-tuning][Tooling]
“DeepSeek Summary: Asserts that domain experts can create more effective, tailored AI skills than generic defaults.”
X
I pointed Claude Cowork at a set of 107 documents (PPTs, Word docs, Excel) that were initially
[RAG][Tooling]
“DeepSeek Summary: Shares a practical experiment using Claude Cowork to process a large collection of mixed-format documents.”
X
We are starting to see some nuanced discussions of what it means to work with advanced AI In this
[Agent][Deployment]
“DeepSeek Summary: Notes emerging, more sophisticated conversations about human-AI collaboration.”
X
I'll meet you at this button.
“DeepSeek Summary: A short, possibly metaphorical or humorous tweet referencing a button.”
X
This book starts like it's gonna be a fun microhistory of TB (it gave us the Stetson!).
“DeepSeek Summary: A tweet commenting on a book about tuberculosis (TB) history, noting its surprising connection to the Stetson hat.”
X
Just got a desk reject, post-rebuttals, for a paper being submitted to arxiv <30 min late for the deadline.
[Evaluation]
“DeepSeek Summary: Shares a frustrating academic experience of a paper being desk-rejected after rebuttals for being slightly late to an arXiv submission deadline.”
X
#throwback to the beginnings of a beautiful friendship =D @ansonmount @HellOnWheelsAMC #HellonWheels #onlocation
“DeepSeek Summary: Angela Zhou shares a nostalgic post about her early experiences on the TV show Hell on Wheels, mentioning co-star Anson Mount and celebrating the show's production.”
X
And awesome to see many Berkeley alums thriving here. @LaurentLessard, @DimitrisPapail, and Shivaram
“DeepSeek Summary: Ben Recht expresses appreciation for Berkeley alumni success and mentions specific individuals.”
X
Here's my blog about this identity and its consequences in observational studies.
[Evaluation]
“DeepSeek Summary: Ben Recht shares a link to his blog post discussing identity and its impact on observational studies.”
X
For the first time in almost a decade, I'm teaching a class on learning and control.
[Evaluation]
“DeepSeek Summary: Ben Recht announces he's teaching a course on learning and control after nearly ten years.”
BLOG
<p>Anthropic are the only major AI lab to <a href="https://platform.claude.com/docs/en/release-notes/system-prompts">publish the system prompts</a> for their user-facing chat systems. Their system prompt archive now dates all the way back to Claude 3 in July 2024 and it's always interesting to see...
“Anthropic uniquely publishes system prompts for their Claude models, providing transparency into AI development. The archive now includes prompts dating back to Claude 3 in July 2024, allowing for tracking of how these foundational instructions evolve.”
BLOG
A learning-oriented workflow for understanding new open-weight model releases
“The post presents a systematic, learning-focused approach for analyzing new open-weight LLM architectures, emphasizing practical understanding over theoretical abstraction. It likely details a repeatable workflow that helps practitioners efficiently grasp architectural innovations and their implications.”
-- END OF LOG --
[STATS] 55 items · Filter applied
Powered by Horizon + DeepSeek