Lastly, the exhausted immune signature thought to be race-aligned that contributes to poor outcome in black women is actually aligned to high poverty burden, independently of race. Thank you to our many collaborators on this story including @raghavanlab.bsky.social and many others #notonhere

Breast cancer cells from white women placed into the collagen patterned microenvironment of black breast cancer lose metastatic potential. Counter to popular theory, breast cancer cells from black women are NOT inherently more metastatic. They need that collagen to initiate mets 2/3

I was at a keynote address where the suggestion was that the future of academia is getting several different genAI models into the room to correct each other's errors. Why would I do that when I could get human experts on different topics into a room. To generate insight not regurgitate facts.

The thing about AI that is wrongheaded is that the end goal seems to be alienating you from other humans. This is the truly terrifying thing about AI, it's social media echo chambering taken to a dangerous extreme that will harm users. Only the extent of harm is TBD based on use

Mind boggling to see how far people will go before someone has the courage to say the emperor is not actually wearing any clothes. And the biggest fall first. Are they scared they'll lose out on the money grab? Or have we created a class of leaders who are feckless followers lacking a spine.

Got all of ours two weeks ago! ✊🏽

When the most privileged in our society decide to shut up and sing, leaving the hard fighting to those least privileged, that is problematic. I often hear the response that we should not expect someone to be morally upright just because they are rich and powerful. I think I will always disagree.

But only if you have the training. If you let students who don't have the training and expertise yet use AI as a proxy then you end up with a problem

But analyzing data is critical to understanding it and to generating ideas. The further away you are from your data, the harder it is to appreciate its strengths and weaknesses.

AI works in the simulation mentioned here because of the existing wealth of data and ideas on this topic. AI cannot come up with new ideas. Any PhD program that tries to "adapt" its training by encouraging students to use AI to do their thinking for them does not understand what drives progress 🧪

Tbf her books are easy reads so there's that