Dylan Bowman's AI safety reading list: 18 essential essays for generalists

Dylan Bowman, a researcher in AI safety, compiled a reading list of 18 documents aimed at helping generalists in the community become more effective. The list is organized into four categories: dispositional traits, strategic decision-making, project leadership, and interpersonal/organizational dynamics. Bowman notes that many AI safety projects require strong generalists, and this list is designed to accelerate growth for those who apply the lessons rigorously.

Key contributors include Paul Graham (five essays, including 'What We Look for in Founders' and 'Do Things that Don’t Scale'), Ben Kuhn (four essays, such as 'Impact, agency, and taste'), and Eliezer Yudkowsky ('Shut Up and Do the Impossible!'). Bowman singles out Anna Salamon's 'Humans are not automatically strategic' as the most impactful single item. The list also features works by Ethan Perez on research project selection, Greg Brockman on CTO leadership, and the book 'The Elephant in the Brain' by Kevin Simler and Robin Hanson, which Bowman calls the best book for organizational psychology.