BEGIN:VCALENDAR
PRODID;X-RICAL-TZSOURCE=TZINFO:-//com.denhaven2/NONSGML ri_cal gem//EN
CALSCALE:GREGORIAN
VERSION:2.0
BEGIN:VEVENT
DTEND;VALUE=DATE-TIME:20261119T163000Z
DTSTART;VALUE=DATE-TIME:20261119T153000Z
DTSTAMP;VALUE=DATE-TIME:20261113T133147Z
UID:42971eaf-5af0-4517-abc9-a0a13f180434@talks.stuts.de
DESCRIPTION:This paper asks what ChatGPT’s inability to self-correct\, ex
 press uncertainty\, or resist user pressure reveals about the nature of 
 machine “understanding.” Drawing on empirical studies of sycophancy (Sha
 rma et al.\, 2023\; Adler\, 2025)\, the hallucination literature\, and B
 ender et al.’s (2021) critique of ungrounded language models\, it argues
  that sycophancy\, confident hallucination\, and the performance of corr
 ection without genuine belief revision are not separate defects but symp
 toms of a single condition: systems trained through RLHF to optimize for
  user approval rather than truth\, with no stable position to defend or 
 revise. Case studies including Mata v. Avianca (2023) and documented ins
 tances of AI-induced psychosis show how these structural features produc
 e real harm. The paper concludes that the appearance of intelligence is 
 engineered rather than incidental\, that commercial incentives make the 
 problem self-perpetuating\, and that remedies require coordinated action
  at the individual\, institutional\, regulatory\, and design levels.
URL:https://talks.stuts.de/de/stuts80/public/events/1627
SUMMARY:Illusion of Intelligence
ORGANIZER:stuts80
LOCATION:stuts80 - Atrium (Philosophikum)
END:VEVENT
END:VCALENDAR