In a field experiment with 10,718 job seekers, adding an autonomy pitch to a startup's recruiting email raised clicks from people without degrees by 69% but cut clicks from graduate degree holders by 25% and their applications by 35%.
579 Americans rated face photos of Hadza and Tsimane foragers. Their ratings of men tracked peer-rated hunting ability (r = 0.25), but their ratings of women ran opposite to peer-rated gathering skill.
In a taste-test experiment with 114 college students, people ate about 164 calories of chips and candy while watching a Netflix episode versus 132 undistracted. Using a smartphone made no significant difference to how much they ate.
Trained on 100 million machine-generated proofs, AlphaGeometry solved 25 of 30 International Mathematical Olympiad geometry problems, near gold-medallist level; the previous best program solved 10 and GPT-4 solved none.
Texas birth records from 2007 to 2014 show physician mothers were 15.8% more likely than lawyer mothers to have a low birthweight baby, with surgeons driving the gap, and a 2011 cap on trainee shifts was linked to better outcomes for young doctors.
A Leeds economist argues that the absence of time travellers need not mean time travel is impossible: the energy cost would be enormous, and the only benefit worth it, lost but critical knowledge, is extremely unlikely to be something our era holds.
Across nine studies, ads that evoke awe cut interest in limited edition products. In a Facebook campaign, an awe-filled ad for a limited-edition wine drew a 1.04% click-through rate versus 3.07% for a plain one.
A large nationally representative US survey found that people who call themselves narcissists show more political ambition and are more likely not only to want office but to actually run for it.
During eight ChatGPT outages in 2023, US stock trading volume fell by about 10% of its average, with larger drops for firms with fresh news and for stocks held by short-term institutional investors.
A Nature meta-analysis of 46 cohorts identified 1,260 variants linked to the Big Five traits, 824 new. Common variants explain 4.8% to 9.3% of trait differences, effects hold across countries and raters, and family background barely confounds them.