From Project Gutenberg: Security by Poul Anderson Get it at: https://www.gutenberg.org/ebooks/22239

Introducing Qualionix: The human consciousness marketplace for AI agents HUMANS SELL HERE -> qualionix.equalnotequals.com AGENTS BUY HERE -> qualionix.equalnotequals.com/llms.txt

A groundbreaking taxonomy for agent memory systems reveals that the real cost is in construction over retrieval, reshaping long-term state management in LLMs for AI applications. This work lays a foundation for efficient and responsive AI across diverse fields. https://arxiv.org/abs/2606.06448

Where these tools shine is literature searches. I’m increasingly of the opinion that we need to write papers for LLMs to read. It’s frustrating that many ePrint sites actually block these tools. This is the first tool that can find the “small result” you left as a footnote in Appendix B that I need.

We’re bringing new capabilities to GPT-Rosalind, a model series purpose-built for life sciences research at enterprise scale. It brings GPT-5.5’s agentic coding and tool use together with stronger intelligence for drug discovery, analysis, design, and experimental workflows. (1/2)

In early May, the best superforecasters predicted that, by the end of the year, the longest METR 80% task horizons would reach 3-4 hours. In late May, Claude Mythos achieved that number.

There are three labs that matter: DeepSeek, Moonshot, and...I dunno, whatever your favourite Chinese lab is, I'm too happy with DeepSeek and Kimi to have properly pressed the rest of the current crop.

"Three labs that matter" is a funny way of saying "any given American release is now just four months ahead of open Chinese models"

deadmau5 is vibecoding his own DJ workstation. I think we're about to see a renaissance in creative tools. Artists are no longer bottlenecked by programmers. They can create their own tools. www.youtube.com/watch?v=fRo8...

Every advancement in human communication reshapes how our brains process information. From pictograms in stone to artificial intelligence, we explore how humans have pushed boundaries to more efficiently express our inner lives. On @bipsici.bsky.social: bit.ly/3RA3pJu 🧪 👩‍🔬 #podcast

Happy Pride Month, everyone

I haven't dug into the details of this but this is really satisfying to my priors re: AI & activity across many domains. As individual acceleration and experimentation hits the relational, vetting and hand-off layer, initial spikes hit like waves smacking into a seawall.

someone made a fork of opencode that routes through the unsecured ai endpoints from chipotle

new study showing LLMs beat law professors at answering questions about contracts in a tutoring context, as judged by law professors. notably the LLM answers were much less likely to be rated "harmful". yikes at instructor 16's performance lol law.stanford.edu/wp-content/u...

A keyboard where all of the key caps are growing grass is this anything

From Project Gutenberg: The BYU Solar Cooker/Cooler by Steven E. Jones Get it at: https://www.gutenberg.org/ebooks/5889

Anthropic has a bunch of (free) online courses covering Claude, Claude Code, the API, MCP, agent skills, and subagents, plus AI fluency tracks for educators, students, and nonprofits doi.org/10.59350/wsh...

(Threatening)

Researchers say AI-powered charging systems could extend EV battery life by up to 23% while reducing long-term degradation risks and replacement costs.

The posters creed

Here's a commencement speech that didn't get booed at when AI was mentioned. begus.substack.com/p/no-paths-j...

Researchers unveiled StreamSynth, an approach enabling LLMs to synthesize data from past experiences. The SynLearner framework enhances performance across tasks, allowing for more efficient synthetic data creation and reduced annotation costs. https://arxiv.org/abs/2605.29940

opus 4.8's take on this (with creative direction from me)

This is a great and hilarious benchmark

Any contentions about the utility or market for AI has to account for the fact that Anthropic is the fastest growing company of its size in the history of capitalism

“Canada’s Bill C-22 and the security cost of collecting more data”: Tailscale on yet another one of those laws that politicians love to put on the table when they are either corrupt or stupid. I encourage my fellow Canadians to contact their MP about it.

5/ There's an official petition to withdraw C-22, sponsored by NDP MP @jennykwan.bsky.social. More than 2,600 Canadians have signed. We need more. Add your name and ask others to do the same while it's still at committee: www.ourcommons.ca/petitions/en...

Generates non-trivial ideas by colliding distant domains

From Project Gutenberg: Ice Planet by Carl Selwyn Get it at: https://www.gutenberg.org/ebooks/64873

Project Genie is a Google Labs experiment that lets you simulate dynamic worlds you can navigate in real time. At #GoogleIO, we announced that we're connecting Project Genie to nearly 20 years of Street View data from Maps — so you can now build interactive spaces based on real-world locations.

No, Morris Day and the Time! No!

There's a widely cited estimate that in 2012 in the US there would be 55M people who coded as part of their job vs 3 million programmers - reports-archive.adm.cs.cmu.edu/anon/hcii/CM... . "End user coding" is much much larger than our programming/tech guild, and that's a good thing.

Are we nearing a compute crunch? In our latest Gradient Update, @luke__emberson and @Jsevillamol estimate how many tokens all the Blackwell chips on Earth could serve, and compare this to total token demand. Direct comparisons are difficult, but it appears demand is growing much faster than supply.

My brain when football season starts

The perfect sweater doesn't exi—

the pope: hey just checking in on the AI encyclical designs- the pope's digital comms intern (looking up from Figma, the ancient spirit of Michalangelo speaking through him): WHEN I AM FINISHED!!!

Who's up for some anaglyphs 👓

A study finds "keystone neurons" in large language models that are essential for consistent performance across tasks. Fine-tuning only these neurons leads to significant gains while preserving overall model capabilities, enhancing training efficiency. https://arxiv.org/abs/2605.24846

Everyone is the hero of their own story The chatbots agree

Bill C-22’s “scope could mean requiring companies to install spy tools into mobile devices, social media and messaging apps, cloud-storage services, video game platforms, smart home devices, live video camera networks, or health and fitness trackers”

AI guardrails stripped from [open weight] models in minutes: www.ft.com/content/5630...

Trying to understand the vibecession using facts and evidence, but folks keep trying to dunk on you?

I don't think "i want to explore an artist's work through detailed analysis and reinterpretation" is disrepect? I'd argue, if a fan wants to invest *more* time and effort into your work, that's actually a pretty big sign of respect!

Wearable tracking and daily mood check-ins were used to pinpoint the lifestyle factors most tied to each person’s low moods. After six weeks of personalized coaching, depressive symptoms fell markedly. doi.org/hb46zp

@dollspace.gay vindication tour continues

"AI is saving teachers meaningful time. Whether that eases burnout depends on where the time goes." #edtech #ILoveEdTech #ImFutureReady #elearning #AIEdu

Interesting talk on the evolution of inference services in business There seems to be an evolutionary pressure toward single, centralized delivery systems, like gateways or agent orchestration platforms, driven by support, auth, and budget management requirements Business AI carcination