BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:CDM Seminar: Scoring Rules for a Theory of AI
DTSTART:20261008T140000
DTEND:20261008T151500
DTSTAMP:20261006T055111Z
UID:f0caca7a34167d1443457c3009e9bb05f0d482fab3036d1c49ad921a
CATEGORIES:Conferences - Seminars
DESCRIPTION:Prof. Jason Hartline Northwestern University\, Evanston\, Illi
 nois\nAbstract:\nScoring rules are foundational in decision theory and\, t
 herefore\, are foundational for a developing theory of artificial intellig
 ence. Just as simple models from decision theory provide context for under
 standing the decisions of complex humans\, so too can they for complex AI 
 systems. Bayesian decision theory considers an agent receiving a signal th
 at is correlated with the state\, choosing an action\, and obtaining a pay
 off that depends on both the state and action. With Bayesian updating and 
 the revelation principle\, the signal becomes a posterior belief and the d
 ecision problem becomes a scoring rule. Given a scoring rule\, baseline pe
 rformance is the optimal score under the prior\; benchmark performance is 
 the optimal score under the posterior\; and the optimal scoring rule — f
 ramed as a mechanism design problem — maximizes the difference between t
 hem. The talk reviews this theory and applies it to (a) evaluate predictio
 n as a service\, (b) behavioral experiments on human-AI decision making\, 
 (c) develop proper scoring rules for text (e.g. for training language mode
 ls to hallucinate less).\n\n\nShort bio:\nProf. Hartline’s research intr
 oduces design and analysis methodologies from computer science to understa
 nd and improve outcomes of economic\, legal\, and AI systems. Optimal beha
 vior and outcomes in complex environments are complex and\, therefore\, sh
 ould not be expected\; instead\, the theory of approximation can show that
  simple and natural behaviors are approximately optimal in complex environ
 ments. This approach is applied to auction theory and mechanism design in 
 his graduate textbook Mechanism Design and Approximation which is under pr
 eparation.\n\nProf. Hartline received his Ph.D. in 2003 from the Universit
 y of Washington under the supervision of Anna Karlin. He was a postdoctora
 l fellow at Carnegie Mellon University under the supervision of Avrim Blum
 \; and subsequently a researcher at Microsoft Research in Silicon Valley. 
 He joined Northwestern University in 2008 where he is a professor of compu
 ter science. He was on sabbatical at Harvard University in the Economics D
 epartment during the 2014 calendar year and visiting Microsoft Research\, 
 New England for the Spring of 2015. He was on sabbatical at Stanford Unive
 rsity for the 2023-2024 academic year. He is visiting ENS Paris-Saclay dur
 ing the 2026-2027 academic year.\n\nProf. Hartline is the director of Nort
 hwestern’s Online Markets Lab\, he was a founding codirector of the Inst
 itute for Data\, Econometrics\, Algorithms\, and Learning from 2019-2022\,
  and is a cofounder of virtual conference organizing platform Virtual Chai
 r.\n 
LOCATION:ODY 4 03 https://plan.epfl.ch/?room==ODY%204%2003 https://epfl.zo
 om.us/j/62755895565
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR