Updates from Trey Causey

Archives
February 16, 2026

[treycausey.com] Takeoff

skooks (@skooookum) on X: "Everything is changing so quickly that I don’t feel particularly sad about what’s going to be lost. I don’t feel particularly optimistic about what will be gained either. Mostly I feel startled. I thought history was something that happened to other people."

Hello!

I've been posting much more frequently on my linkblog, commonplace, lately, so I'm sharing a digest of recent posts at the end of this email.

There's a very clear theme from the past few weeks: AI takeoff seems to be happening. Things are moving so quickly that I haven't had time to sit down and write lengthier pieces about where we find ourselves, and struggle just to keep up with the conversation. Normally, I'd discourage the idea of "keeping up with the conversation" but I'm spending a lot of time thinking about how to be agentic and shape my career to accommodate this. (I'm still looking for my next role!). Right now, it's not clear to me what careers in tech might look like in even a year or two. If you have thoughts, I would love to hear them.

Meanwhile, Sonnet 5, GPT 5.3, and Gemini 3.1 are all rumored to drop ~this week, hot on the heels of Opus 4.6, GPT-5.3-Codex, and GPT-5.3-Codex-Spark. Point releases show more gains in progress than we used to see between major releases.

It's undeniable, to me at least, that we hit an inflection point in the conversation about AI capabilities around November of last year. It's hard to look at all of the data that METR is collecting and conclude that we're not also hitting real inflection points. Of course, benchmarks are gamed, don't accurately measure things in the real world, etc., etc., but every benchmark, new and old, is basically showing the same trajectory. And I don't think this is what people meant last summer when they said AI progress had hit a wall 😅.

A graph from METR showing the time horizon of software tasks different LLMs can complete 50% of the time. The past ~year the slope of the graph has become almost vertical.
Source: Measuring AI Ability to Complete Long Tasks, METR

The posts I've been sharing reflect the moment we find ourselves inn. If you only read one, I recommend "Something Big Is Happening" from Matt Shumer. Buckle up.

As always, thank you for spending time reading what I have to say. It means a lot, truly, to have interested readers in a crowded information landscape. I try to maintain a high-signal, useful feed. I try not to share too much viral content (like the Shumer post) unless I think it truly deserves even more amplification.

Recent commonplace posts

  • 2026-02-07 — The Anthropic Hive Mind

  • 2026-02-08 — My AI Adoption Journey – Mitchell Hashimoto

  • 2026-02-11 — "Something Big Is Happening"

  • 2026-02-12 — Quoting skooks on X

  • 2026-02-13 — When "technically true" becomes "actually misleading" ($)

  • 2026-02-13 — Highbrow climate misinformation

  • 2026-02-13 — Quoting Armin Ronacher

  • 2026-02-15 — The many masks LLMs wear - Kai Williams

  • 2026-02-15 — Use /copy in Claude Code!

  • 2026-02-15 — Séb Krier on misalignment and evaluations

- Trey

Don't miss what's next. Subscribe to Updates from Trey Causey:
Powered by Buttondown, the easiest way to start and grow your newsletter.