ai news
All Senwitt Journal posts in this category.
- May 26, 2026
The Anthropic coding-skill study: what developers should actually take away
Anthropic's 2026 study found AI-assisted developers scored ~17 points lower (50% vs 67%) on an immediate comprehension quiz when learning a new library. A careful read of what it does — and doesn't — show, and what developers should change in their daily practice.
- May 26, 2026
OpenAI's reasoning models: do they fix the thinking problem?
OpenAI's o-series reasoning models 'think before answering' using chain-of-thought-style processing. That changes what the model does. It does not change what the user does — and the cognitive-offloading concern is about the user.
- May 26, 2026
Ten years after FTC v Lumosity: what actually changed in brain apps
The 2016 FTC settlement against Lumosity turns ten. The honest accounting of what the category learned, what it didn't, and what the post-2016 honest products look like.
- May 26, 2026
The BCG/HBR 'AI brain fry' study, explained
A focused study-explainer on the 2026 BCG and HBR-aligned research covered as 'AI brain fry'. What the ~14% figure does and doesn't show, why the mechanism is plausible, and what's missing from most coverage.
- May 26, 2026
What Senwitt borrows from Duolingo — and what it doesn't
Duolingo's habit design and streak research are the most documented in consumer learning. Senwitt borrows the bounded daily ask and the forgiveness on missed days. We don't borrow the streak guilt or the leaderboard pressure.