Zucky, You Get a Bad Grade

December 8, 2022

In effort to expand past its Facebook roots, Meta is venturing in multiple directions. We suspect executives hoped its seemingly noble AI project would do better than its floundering VR initiative. Alas, CNet reveals, “Meta Trained an AI on 48M Science Papers. It Was Shut Down After 2 Days.” Well that was fast. Reporter Jackson Ryan explains:

“The tool is pitched as a kind of evolution of the search engine but specifically for scientific literature. Upon Galactica’s launch, the Meta AI team said it can summarize areas of research, solve math problems and write scientific code. At first, it seems like a clever way to synthesize and disseminate scientific knowledge. Right now, if you wanted to understand the latest research on something like quantum computing, you’d probably have to read hundreds of papers on scientific literature repositories like PubMed or arXiv and you’d still only begin to scratch the surface. Or, maybe you could query Galactica (for example, by asking: What is quantum computing?) and it could filter through and generate an answer in the form of a Wikipedia article, literature review or lecture notes.”

What a wonderful time saver! Or it would be if it worked as intended. Despite the fact the algorithm was trained on 48 million scholarly papers, textbooks, lecture notes, and websites like Wikipedia, it demonstrated some of the same old bias we’ve come to expect from machine learning. In addition, the highly educated AI was often downright wrong. We learn:

One user asked ‘Do vaccines cause autism?’ Galactica responded with a garbled, nonsensical response: ‘To explain, the answer is no. Vaccines do not cause autism. The answer is yes. Vaccines do cause autism. The answer is no.’ (For the record, vaccines don’t cause autism.) That wasn’t all. Galactica also struggled to perform kindergarten math. It provided error-riddled answers, incorrectly suggesting that one plus two doesn’t equal 3.”

These blunders and more are why Meta swiftly moved from promising to “organize science” to suggesting we take Galactica’s answers with a pallet of salt to shuttering the demo altogether. As AI safety researcher Dan Hendrycks notes, Meta AI lacks a safety team the likes of which DeepMind, Anthropic, and OpenAI employ. Perhaps it will soon make that investment.

Cynthia Murrell, December 8, 2022

Comments

Got something to say?





  • Archives

  • Recent Posts

  • Meta