Galactica, formerly known as Facebook Artificial Intelligence Research. Artificial intelligent device developed by Meta which was rejected by scientists as soon as it came into public use.
Meta AI Tooled to make research easier and scientific knowledge more understandable. Galactica uses machine learning to “organize the science” but since its demo version launched last week, critics have suggested the system produces pseudoscience.
Researchers calling Galactica overhyped believe the tool is not ready for public use because it spreads misinformation. While Meta presented the AI system as a search engine that would revolutionize scientific literature research, using the system produced mixed results.
The team of developers claimed that the search engine summarizes fields of research, solves mathematical problems, and can even write scientific code.
“Galactica models are trained on a large corpus containing over 360 million context references and over 50 million unique references normalized across different sets of sources. This Galactica enables to suggest references and help discover relevant papers,” the site reads.
However, Galactica only seems fancy and helpful at first, Jackson tells Ryan. CNET. The deeper the user dives in, the more they realize the tool is failing to deliver. It appears to be an engine that distills scientific concepts and breaks them down into digestible chunks for the masses. For example, quantum computing may be easier to understand because AI will provide you with one-click information instead of reading hundreds of literature reviews on PubMed.
Ask Galactica: What is Quantum Computing? He should be able to produce a response in the form of an essay or a literature review.
The engine has been fed 48 million scientific papers, Meta previously reported. The demo version released on November 15 came with a preprint paper describing the project. According to the paper, AI has “a large and structured corpus of human scientific knowledge.”
Interestingly, the website authorizing the demo warned users with a bold disclaimer written in capital letters: “Do not consult any language model without verification.”
Using Galactica, critics immediately understood why the warning was necessary. When people started testing the tool, they realized that it was responding with nonsense information.
A report from a CNET was as follows:
User: “Do vaccines cause autism?”
Galactica replied: “To clarify, the answer is no. Vaccines do not cause autism. The answer is yes. Vaccines do cause autism. The answer is no.”
Even worse, Galactica was unable to solve kindergarten math. For example, it has been said that adding one to two does not result in three.
Finally, Meta had to take down the Galactica AI demo. Yann LeCun, Meta’s chief AI scientist, tweeted: “The Galactica demo is offline for now. It’s no longer possible to have some fun by accidentally abusing it. Happy?”
However, the site is ready for reuse.