Huafeng Xu

Had deep neural networks been invented before Newton, physics would not have looked the same.

Ironically, we now pin our hopes of transforming biology into a predictable, reproducible, and deterministic engineering discipline on neural networks, a technology inspired by biology and, rare among engineered systems, itself unpredictable and non-deterministic.

Rich people are different. They chase what money cannot buy: they chase what only an excessive amount of money can buy.

The discovery of a new medicine is a rare event. Can AI help turn the improbable into the likely? My (digressive) thoughts on how AI may impact drug discovery in a meandering post. Title inspired by Blade Runner. Style warped by sleep deprivation. @huafeng/do-neural-networks-dream-of-drug-molecules-55e9ad492e1e" target="_blank" rel="nofollow noopener">medium.com/@huafeng/do-ne…

I just got recommended to the recruiter for a job I am hiring for. Flattered, twice.

Did the pants-to-shirts sales ratio decrease with the rise of remote meetings?

@JDVance They cleared out the whole place for the second family.

With three little kids staring into the sun, this was actually the *best* photo we got at the Taj Mahal today 😂

Suspicious?

The most notable effect of AI on science will not be the solution of outstanding scientific problems but the shortening of author list in scientific publications. We might soon see single author papers describing complex scientific works that previously call for a team.

I have used both, one much more than the other.

Antonie van Leeuwenhoek’s microscope and Anton the computational microscope.

I like a watermelon because “you can eat, drink, and wash your face all at the same time.” What is this monstrosity that takes away two thirds of the fun?

Is this what people call a dream job?

@gavincrooks We do these things not because they are easy, but because we thought they would be hard, but not THAT hard.

We do these things not because they are easy, but because we thought they would be easy.

@McdarghZach So mathematically rigorous

@hardsphere Evidently, that is all that is due

“With due respect” is invariably followed by words spoken without respect.

I share my thoughts on the enabling elements that drive rapid technological advances: public attention, scientific foundation, talent pool and empowering infrastructure, focus on value creation, technology integration, and sufficient investment. tinyurl.com/5n8srpsc

AI’s results feel adequate when I have no clear expectations—but often fall short when I do.

@olexandr One day all papers will be written by AI, reviewed by AI, and read by AI. #AI4science. You never know.

Please report AI generated papers to your journal editors! We are here to help with cases like these

Wenfeng Liang, who founded and led the DeepSeek team, is one of the over one hundred authors of the DeepSeek paper, but he is neither the first nor the last author, and there is no asterisk next to his name. That alone is a distinction.

So much excitement but maybe it is still a stochastic parrot?

So much excitement but maybe it is still a stochastic parrot?