← the record
AIAAIC-1704

Galactica generates inaccurate, racist, homophobic and offensive responses

Meta's Galactica model spouted fake and offensive information immediately after its launch, prompting ridicule and its closure. Within hours of its release, users discovered Galactica creating entirely fabricated studies and research papers; making false attributions; producing biased content, including false claims about Black people and immigrants; and creating nonsensical information, such as a fake study about the benefits of eating crushed glass. Galactica also failed to solve basic mathematical problems accurately, d espite being designed to assist with scientific tasks . A Wikipedia text about AI researcher and entrepreneur Gary Marcus generated by Galactica contained 85 percent incorrect information, according to the researcher, but it was phrased plausibly. A "decent AI system" could check such information online, but Galactica doesn't provide that feature, Marcus said. Galactica , he said, constitutes 'pitch perfect and utterly bogus imitations of science and math, presented as the real thing.' For University of Washington in Seattle biologist Carl Bergstrom, the problem with Galactica is that it 'pretends to a portal to knowledge. Actually it's just a random bullshit generator'. 'It’s no longer possible to have some fun by casually misusing it. Happy?' Meta’s chief AI scientist Yann LeCun followed-up . Galactica was withdrawn after three days in the wake of criticism from prominent AI researchers and technology commentators. System 🤖 Galactica Operator: Meta/Facebook Developer: Meta/Facebook Country: Global Sector: Technology Purpose: Assist scientists Technology: Large language model (LLM); NLP/text analysis; Neural network; Deep learning Issue: Accuracy/reliability; Bias/discrimination - race, ethnicity, gender, religion; Mis/disinformation ; Safety

Date it happened
2022-11-01
Organisation involved
Facebook
Product, system or model
Galactica
Where this came from
Share this incident
XLinkedInFacebookWhatsAppEmail
Attribution

This incident was imported from AIAAIC and is used under CC BY-SA 4.0. Our additions to it — the structured fields, the translation, the checks against other reports — are published under the same licence.

This is a record of what was reported, not a finding that anyone broke the law. If it names your organisation and you believe it is wrong, the corrections process is free and open to everyone.