Simona Cristea (@simocristea) 's Twitter Profile
Simona Cristea

@simocristea

Cancer genomics AI scientist. Faculty @DanaFarber @Harvard. Head of Genomics Data Science & AI @ Hale Center for Pancreatic Cancer. PhD @eth. 🇷🇴🇸🇪🇨🇭🇺🇸

ID: 4792966221

linkhttp://www.simona.ai calendar_today13-01-2016 09:56:14

4,4K Tweet

8,8K Followers

502 Following

Deedy (@deedydas) 's Twitter Profile Photo

The new President of Romania has 2 perfect score Gold medals on the International Math Olympiad (IMO), a feat only 25 people have ever achieved in history! First the pope, and now this. First they learn about higher dimensional spaces and then they excel in all of them.

The new President of Romania has 2 perfect score Gold medals on the International Math Olympiad (IMO), a feat only 25 people have ever achieved in history!

First the pope, and now this. First they learn about higher dimensional spaces and then they excel in all of them.
Simona Cristea (@simocristea) 's Twitter Profile Photo

i see this in young graduates every day. using chatGPT for mindful learning and careful inspection can help young people grow into the most wonderful scientists. but, mindless copy-paste from chatGPT at every step will kill their scientific reasoning. it’s quite sad to watch

Ethan Mollick (@emollick) 's Twitter Profile Photo

My conclusion from the last couple of weeks of AI launches is that none of the AI companies can explain very well what their systems do. This is partially because they don’t always know & partially because there is no established approach for documentation of AI capabilities.

Simona Cristea (@simocristea) 's Twitter Profile Photo

writing is literally thinking, that's why only copy-pasting GPT outputs will make people forget how to think for themselves

Simona Cristea (@simocristea) 's Twitter Profile Photo

today, we take for granted that super large & powerful deep learning models can predict biology to an insane level of detail yet we don’t spend enough time to understand what this means - that the underlying biological paths leading to complex outcomes are in fact structured

Simona Cristea (@simocristea) 's Twitter Profile Photo

many people still blindly follow scRNAseq tutorials when analyzing cancer data - and that’s a really bad idea. you should at least: 1. use a higher mitochondrial gene percentage QC threshold than default, e.g. 20% 2. do not integrate (Harmony) your data, especially cancer cells