We have some of the most powerful AI systems in history - but what if one of the biggest obstacles to using AI to solve biology isn't the algorithms? What if it's the data?
There are enormous amounts of biological data sitting inside pharmaceutical companies, biotech companies and research institutions. The problem is that much of it is proprietary - and the organizations that spent millions generating it have little incentive to simply give it away.
What if there's a way for AI to learn from that data without the data having to leave its owner's hands?
Dr. Harlan Robins, Ph.D. is a physicist, computational biologist, biotech entrepreneur, and longtime pioneer in applying computational methods to biological problems.
Dr. Robins earned his undergraduate degree in physics from Harvard University and his PhD in physics from UC Berkeley, followed by postdoctoral work at the Weizmann Institute of Science and the Institute for Advanced Study in Princeton, where he worked with renowned biologist Dr. Arnold Levine on computational approaches to problems in genetics and genomics.
Dr. Robins subsequently joined the Fred Hutchinson Cancer Center, where he became a faculty member and ultimately headed its Computational Biology Program - at a time when computational biology itself was still a relatively new field.
In 2009, Dr. Robins co-founded Adaptive Biotechnologies ( https://www.adaptivebiotech.com/ ), where he served as Chief Scientific Officer for more than 16 years. Adaptive was built around a fascinating idea: that the enormous information contained in the genetics of our adaptive immune system could be decoded and translated into clinical tools for diagnosing and treating disease. The company went public in 2019.
Now Dr. Robins is embarking on a new chapter as President and Co-Founder of Harell Data ( https://harelldata.com/ ), a company focused on one of the potentially enormous bottlenecks in the age of AI-driven science: access to high-quality scientific data.
There is no shortage of biological data in the world - but much of the most valuable data is generated and held by companies and researchers who understandably don't want to simply give away their intellectual property.
Harell is trying to create a different model: allowing powerful AI and computational systems to make use of proprietary scientific datasets while allowing the organizations that generated that data to retain control over it and participate in the value created.
And that raises a much bigger question - one that goes well beyond any individual company:
As AI becomes increasingly capable of solving problems in biology, could access to the right data become one of the biggest constraints on scientific progress?
Podden och tillhörande omslagsbild på den här sidan tillhör
Ira Pastor. Innehållet i podden är skapat av Ira Pastor och inte av,
eller tillsammans med, Poddtoppen.
