Ben WaberThe boot is off!!! While it's going to be a few weeks to get my leg strength back and start running again, at least I was able to start to relearn to walk while listening to some nice talks for my #AcademicRunPlaylist! (1/10)
First was an important talk by Sarah Kreps on #AI and the political process at #StanfordHAI. While #LLMs have been rightly flagged as a risk to democratic societies for a whole host of reasons, this is an excellent breakdown of the pros and cons of the technology in politics https://www.youtube.com/watch?v=WeDkUFIrB6I (2/10) #politics
Sarah E. Kreps: The Promise and Perils of AI-Mediated Political Communication
Next was an interesting talk by Ashia Wilson on the many concerns we should have with large multi-modal models and what can be done about them at the #MIT Schwarzman College of Computing https://www.youtube.com/watch?v=46QWKC1tZZQ (3/10) #LLMs
SERC Symposium 2023: Ashia Wilson
Next was a nice talk by @phend on aligning #MachineLearning, #law, and policy at the #Princeton #CITP. The in-depth examining of algorithm use at the IRS discussed here was fascinating https://www.youtube.com/watch?v=-DnqdiaLXwY (4/10)
CITP Lecture: Peter Henderson – Aligning ML, Law, Policy for Responsible Real-World Deployments
Next was a compelling talk by Sendhil Mullainathan on how #MachineLearning can accentuate, uncover, and reduce biases. The point about comparing to humans is important, but I would argue we should go further. We shouldn't design algorithms to "cover" for people, but rather design a holistic system that may use algorithms but is optimized for the overall result rather than the isolated performance of an algorithm https://www.youtube.com/watch?v=w1UKZqGL-qM (5/10) #bias
SERC Symposium 2023: Sendhil Mullainathan
Next was an intriguing talk by Ashesh Rambachan on evaluating machine predictions for legal decisions. I would like some more interrogation of the "ground truth" that goes into these models https://www.youtube.com/watch?v=cozjvjGwIJ4 (6/10)
SERC Symposium 2023: Ashesh Rambachan
Next was a rich talk by @williambrady on how #SocialMedia algorithms warp our social perceptions at #NICO. I liked the combination of field studies and controlled experiments to investigate this issue (although I think we're getting past the point where using #Twitter data can be considered a valid data source on social interactions) https://www.youtube.com/watch?v=qgKM6XgOzco (7/10)
Wed@NICO - 5/17 - William Brady
Next was a fantastic conversation with Naomi Oreskes on the myth of the "free market" at #StanfordGSB. Oreskes deftly deconstructs the notion that a market can be disconnected from other aspects of society, and also exposes how thinkers promoting that notion have ignored foundational economic scholarship. Highly recommend https://www.youtube.com/watch?v=kJV7_0BIbxo (8/10) #economics
Exposing the Big “Free Market” Myth with Author Naomi Oreskes
Next was an engaging panel on exerting power over data and algorithms at @CPDPconferences with @PaulNemitz, Katarzyna Szymielewicz, @mikarv, and Raegan MacDonald. The rise of #Mastodon, while not explicitly discussed here, fit extremely well into a lot of the conversation as a model to be encouraged https://www.youtube.com/watch?v=6A_25TG86y8 (9/10)
POWER OVER DATA AND ALGORITHMS: CAN WE GET IT BACK?
Last was a crucial talk on an ethnographic analysis of migrant #labor exploitation in India by Alpa Shah and Jens Lerche at the #LSE https://www.youtube.com/watch?v=LeM_mxnkuBk (10/10)
