A fascinating look into OpenAI the company:

[…] you probably shouldn’t view OpenAI as a single monolith. I think of OpenAI as an organization that started like Los Alamos. It was a group of scientists and tinkerers investigating the cutting edge of science. That group happened to accidentally spawn the most viral consumer app in history.

Preprint:

We detect a measurable and abrupt increase in the use of words preferentially generated by ChatGPT, such as delve, comprehend, boast, swift, and meticulous, after its release.

Just for the record, I’ve been using em dashes — correctly, I hope — for decades.

The AP article quotes Simon Willison:

Willison also said he finds Grok 4’s capabilities impressive but said people buying software “don’t want surprises like it turning into ‘mechaHitler’ or deciding to search for what Musk thinks about issues.”

“Grok 4 looks like it’s a very strong model. It’s doing great in all of the benchmarks,” Willison said. “But if I’m going to build software on top of it, I need transparency.”

I haven’t had to figure out AWS IAM or review the Cost Explorer in a hot minute.

First, you’ll need to understand modern AWS service naming. They’ve clearly hired a team of Scrabble champions who’ve been hitting the espresso too hard and have used up all the good letters / names in the early game. Need serverless AI-powered quantum computing? That’s AWS QuantumLambdaForgeMaxProUltra™ (but they call it “Amazon Q” for short). Want to deploy a simple API? You’ll need AWS HyperGatewayMeshFabricOrchestrator360™.

Via @ErikJonker@mastodon.social:

We’ve been hearing a lot about MCP lately (e.g. Wiley Partners with Anthropic to Accelerate Responsible AI Integration Across Scholarly Research), but A2A hasn’t been on my radar until now.

You can think of A2A as the “meeting room” where agents coordinate and hand off work, while MCP is the “toolbox” each agent uses to get things done. They are complementary technologies.