This Week I Learned #2—JOURNAL

This Week I Learned #3—JOURNAL

Less reading this week… because more composing.

Initial Notes on Elicit—JOURNAL

Playing around a bit with the ML-powered research tool from the folks at Ought.org.

Is GitHub Copilot Any Good?—JOURNAL

Betteridge’s Law applies, but the details of why I think so might still interest you.

Upcoming: Substrate Engineering (StaffPlus New York 2024)—ELSEWHERE

Coming in September: a talk on “Engineering Foundations in a World of LLMs”.

StaffPlus NY 2024: Substrate Engineering —ELSEWHERE

Engineering Foundations in a World of LLMs

New Tool: jj-gpc—JOURNAL

Making nicer branch names on the fly.

No AI on This Site—NOTES

A statement and a pledge.

Implications of Plateauing LLMs—NOTES

They may not be what you think.

Two Quotes and a Question on Software Engineering and AI—NOTES

What should we make of generative AI systems in software engineering?

Re-packaging Existing Content, Transformation, or Both?—NOTES

Some technical notes for thinking about the legal and ethical questions around “generative” AI systems.

Robin Sloan—“Is It Okay?”↩︎—❈—NOTES

(Betteridge cannot cheat on this one.)

Yes, Hallucinations are the “Least Dangerous”—Exactly—NOTES

A rejoinder to Simon Willison’s rant last week.

Claude as a Better Grammarly?—NOTES

Results are mixed. Perhaps unsurprisingly.

History and Disposition—NOTES

I have spent my career on large, legacy systems, and that informs an awful lot about me—including my views on LLMs.

BBEdit Syntax for Jujutsu Commit Messages↩︎—❈—ELSEWHERE

Because I want JJ: lines to be treated like comments, dagnabbit!

Telling—NOTES

Don’t Conflate Intelligence with Value↩︎—❈—ELSEWHERE

In Christianity Today’s July/August 2025 issue on artificial intelligence.

the way we use language—NOTES

Quoting Ben Makuh on how we talk about LLMs and the language of slavery,