We know that Microsoft is being criticized by one section of the society, including Elon Musk, for the fact that OpenAI which was an open-source company at the time Elon Musk invested into the company and the one that also planned to stay open-source has now been close sourced because of the fact that Microsoft invested heavily inside the company and possibly had some terms and conditions along with the investment as well. Now, we also know that AI is being used in medical field as well and Microsoft has done something that a lot of people in medicine would be proud of.
There is a Novel Protein-generating AI which is available in the market right now which is known as EvoDiff and it is worth noting that it was close-sourced until now. But now, Microsoft has announced that this AI will be open-sourced for the development of society. For those who don’t know, proteins are very important in the making of vaccines which can act as antidotes to viruses such as COVID-19 and others. EvoDiff’s co-creators said, “We envision that EvoDiff will expand capabilities in protein engineering beyond the structure-function paradigm towards programmable, sequence-first design,” and added, “With EvoDiff, we’re demonstrating that we may not actually need structure, but rather that ‘protein sequence is all you need’ to controllably design new proteins.”
Microsoft Senior Researcher said, “If there’s one thing to take away [from EvoDiff], I think it’d be this idea that we can — and should — do protein generation over sequence because of the generality, scale, and modularity that we’re able to achieve,” and added that “Our diffusion framework gives us the ability to do that and also to control how we design these proteins to meet specific functional goals.” Adding to the statement, Microsoft researcher Ava Amini said, “This is just a 640-million-parameter model, and we may see improved generation quality if we scale up to billions of parameters”. “While we demonstrated some coarse-grained strategies, to achieve even more fine-grained control, we would want to condition EvoDiff on text, chemical information or other ways to specify the desired function.”
