BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:CANCELLED: How do we protect the democratic information environmen
 t in an AI-powered world? Tackling online harms with computational social 
 science and AI - Helen Margetts (University of Oxford)
DTSTART:20250129T130000Z
DTEND:20250129T140000Z
UID:TALK224557@talks.cam.ac.uk
CONTACT:Yara Kyrychenko
DESCRIPTION:*THIS TALK IS CANCELLED*\n\n2023-2024 was a peak year of conce
 rn over AI safety\, highlighted at the UK AI Safety Summit in Bletchley Pa
 rk\, the creation of AI safety institutes across the world\, and concerns 
 raised by technology experts over AI-powered threats to democracy and exis
 tential risks brought by the fast pace of AI development. This talk will s
 crutinise the claims that have been made regarding threats from AI to demo
 cratic life in an online world. It will report research that suggests that
  hype over AI is having its own independent effect\, that targeted politic
 al persuasion may be less effective than has been claimed\, that smaller m
 odels can be as persuasive as larger models and that AI itself may be part
  of the solution to AI-powered harms. Research suggests that the real thre
 ats to society and democracy may come from long-running shifts in the info
 rmation environment\, where people start mistrusting all information and a
 re increasingly fearful of expressing political opinions in online setting
 s. AI and computational social science researchers will need to find new w
 ays of understanding\, measuring and mitigating these threats.
LOCATION:Ground Floor Lecture Theatre\, Department of Psychology\, Downing
  Site\, Cambridge
END:VEVENT
END:VCALENDAR