001031976 001__ 1031976
001031976 005__ 20241213210707.0
001031976 037__ $$aFZJ-2024-05900
001031976 1001_ $$0P:(DE-Juel1)176538$$aRathkopf, Charles$$b0$$eCorresponding author$$ufzj
001031976 1112_ $$aUppsala Vienna AI Colloquium$$d2024-10-25 - $$wonline event
001031976 245__ $$aHallucination, justification, and the role of generative AI in science
001031976 260__ $$c2024
001031976 3367_ $$033$$2EndNote$$aConference Paper
001031976 3367_ $$2DataCite$$aOther
001031976 3367_ $$2BibTeX$$aINPROCEEDINGS
001031976 3367_ $$2ORCID$$aLECTURE_SPEECH
001031976 3367_ $$0PUB:(DE-HGF)31$$2PUB:(DE-HGF)$$aTalk (non-conference)$$btalk$$mtalk$$s1734070645_20223$$xInvited
001031976 3367_ $$2DINI$$aOther
001031976 520__ $$aGenerative AI models are now being used to create synthetic climate data to improve the accuracy of climate models, and to construct virtual molecules which can then be synthesized for medical applications. But generative AI models are also notorious for their disposition to “hallucinate.” A recent Nature editorial defines hallucination as a process in which a generative model “makes up incorrect answers” (Jones, 2024).  This raises an obvious puzzle. If generative models are prone to fabricating incorrect answers, how can they be used responsibly? In this talk I provide an analysis of the phenomenon of hallucination, and give special attention to diffusion models trained on scientific data (rather than transformers trained on natural language.) The goal of the paper is to work out how generative AI can be made compatible with reliabilist epistemology. I draw a distinction between parameter-space and feature-space deviations from the training data, and argue that  hallucination is a subset of the latter. This allows us to recognize a class of cases in which the threat of hallucination simply does not arise. Among the remaining cases, I draw an additional distinction between deviations that are discoverable by algorithmic means, and those that are not. I then argue that if a deviation is discoverable by algorithmic means, reliability is not threatened, and that if the deviation is not so discoverable, then the generative model that produced it will be relevantly similar to other discovery procedures, and can therefore be accommodated within the reliabilist framework.
001031976 536__ $$0G:(DE-HGF)POF4-5255$$a5255 - Neuroethics and Ethics of Information (POF4-525)$$cPOF4-525$$fPOF IV$$x0
001031976 909CO $$ooai:juser.fz-juelich.de:1031976$$pVDB
001031976 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)176538$$aForschungszentrum Jülich$$b0$$kFZJ
001031976 9131_ $$0G:(DE-HGF)POF4-525$$1G:(DE-HGF)POF4-520$$2G:(DE-HGF)POF4-500$$3G:(DE-HGF)POF4$$4G:(DE-HGF)POF$$9G:(DE-HGF)POF4-5255$$aDE-HGF$$bKey Technologies$$lNatural, Artificial and Cognitive Information Processing$$vDecoding Brain Organization and Dysfunction$$x0
001031976 9141_ $$y2024
001031976 9201_ $$0I:(DE-Juel1)INM-7-20090406$$kINM-7$$lGehirn & Verhalten$$x0
001031976 980__ $$atalk
001031976 980__ $$aVDB
001031976 980__ $$aI:(DE-Juel1)INM-7-20090406
001031976 980__ $$aUNRESTRICTED