Meta abandons protein-folding AI project, focusing on commercial generative AI instead


As a fundamental technology, AI is advancing many sciences, but Meta reportedly sees no business in it.

That’s according to the Financial Times, citing three sources familiar with Meta’s AI projects. According to the report, Meta has shut down its ESMFold protein-folding project, which was unveiled late last year, and laid off the team. About a dozen researchers involved were forced to leave in the spring as part of a larger wave of layoffs.

ESMFold used a large language model to generate protein structure predictions 60 times faster than Deepmind’s Alphafold, which is considered a scientific sensation. However, ESMFold loses accuracy.

Fast and accurate protein predictions can help in the development of new drugs. Deepmind and Alphabet have formed a new AI medicine company, Isomorphic Labs, to do just that. Maybe Meta doesn’t think it can compete there, or doesn’t see the opportunity.


A signal for more commercial AI projects with a focus on generative AI

Meta announced a “year of efficiency” at the end of 2022, following a slump in its share price, to focus on profitability and businesses with good growth prospects. As part of this restructuring, Meta will lay off approximately 20,000 employees.

The ESMFold team seems to have fallen victim to this change in strategy. It is unclear whether Meta will continue to offer the existing ESMFold infrastructure and database.

While the ESMFold team was tiny compared to the thousands of AI experts at Meta, the project’s demise is a sign that Meta is prioritizing commercial AI projects with profit prospects over scientific or experimental projects, according to former ESMFold employee Yaniv Shmueli. A particular focus will be on generative AI.

According to Joëlle Pineau, vice president of AI research at Meta, the company remains committed to exploratory research based on open science in its science-based “Fair” AI division.

In June, the Wall Street Journal also published an article about Meta shifting its resources from AI research to AI products.


Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top