1. Meta by Peter Watts at Rifters. Back when the first OpenAI’s GPT-2 results came out I was pretty impressed – but I failed to notice that astrophysicist Cody Raskin fed one of my favourite scifi works Blindsight to the text-generating algorithm. While the results lack the actual insights in Watts’ writings, the style is eerily similar.
  2. How Lyft Designs the Machine Learning Software Engineering Interview by Hao Yi Ong for Lyft Engineering. A look at the way Lyft structures their Machine Learning interviews.
  3. Predicting Psychological and Subjective Well-Being from Personality: A Meta-Analysis by Anglim et al. (2019) An examination of the links between self-reported personality and well-being links high extraversion and low neuroticism to high levels of happiness.
  4. Bayesian Examination by Lê Nguyên Hoang for LessWrong. Hoang advocates Bayesian examination models to strengthen probabilistic thinking.
  5. Why Not RFID Tag Humans? by Robin Hanson at Overcoming Bias. Short read on the acceptance level of RFID tags for humans – the majority in his poll are still opposed to the idea. But a growing number of biohackers (me included) are already toying with the possibilities.
  6. Is Eating Meat A Net Harm? Adversarial Collaboration Debate at Slate Star Codex. The debate covers the topics of animal consciousness, preference of existence vs. non-existence, as well as environmental/health/financial considerations.
  7. It’s Time To Go Vegan by Alex O’Connor of Cosmic Skeptic. As always, O’Conner brilliantly argues his case, leaving you with little rational reason to remain an omnivore.
  8. Artificial Intelligence Is Rushing Into Patient Care—And Could Raise Risks by Liz Szabo for The Scientific American. Medical AI applications are in hot demand, but the “fail fast and fix things later” mentality of tech falters where lives are at risk.
  9. Humans Are Embedded Agents Too by johnswentworth at LessWrong. The embedded agency sequence shows how difficult it is to model artificial intelligence – but that same issue applies to modelling humans.
  10. Twelve Million Phones, One Dataset, Zero Privacy by Stuart A. Thompson and Charlie Warzel for the New York Times. ““D.N.A.,” he [Paul Ohm] added, “is probably the only thing that’s harder to anonymize than precise geolocation information.” Our privacy is only as secure as the least secure app on our device.”