1. Getting to Decisions Faster in A/B Tests – Part 1 by Aurimas Račas. In this first post in a series, the author explores approaches that are used in industry to improve decision speed in A/B testing. This post is a literature review with links to lots of resources and summaries along with some examples of how they’ve been used in practice.
  2. Cognitive Behavior Therapy vs. Control Conditions, Other Psychotherapies, Pharmacotherapies and Combined Treatment for Depression: A Comprehensive Meta-analysis Including 409 Trials with 52,702 Patients by Cuijpers et al. (2023). Biggest ever meta-analysis on cognitive behavioral therapy finds it works very well in a wide variety of circumstances, may slightly outperform other psychotherapies, quals medication in the short-term but perhaps outperforms it in the long-term, and doesn’t get less effective with time.
  3. Bing: “I Will Not Harm You Unless You Harm Me First” by Simon Willison. A collection of errors Bing Chat errors detected in the past weeks.  “[…] large language models have no concept of “truth”—they just know how to best complete a sentence in a way that’s statistically probable based on their inputs and training set. So they make things up, and then state them with extreme confidence.”
  4. Talking About Large Language Models by Shanahan (2022). As LLMs become better at imitating human language, we become more prone to anthropomorphism, i.e. the tendency to see the systems in which they are embedded as more human-like than they really are.
  5. Oral Contraceptives Don’t Impact Muscle Growth and Strength Gains by Greg Nuckols at Stronger For Science. A new study puts serious horse power behind the evidence that oral contraceptives (contra to popular opinion) don’t seem to meaningfully affect strength or hypertrophy outcomes. It’s the biggest study on the topic (n=74) and the researchers went to great lengths to control  for menstrual cycle phase.