The google AI strategy seems to be to build all sorts of different domain specific models to which gemini (the LLM) will have access to use as tools. Imo this strategy wins the whole AI race. It will just appear like they are behind for a while, but the LLMs are the easy part.

Social media has warped so many peoples minds that if someone doesn’t post about helping someone out then it never happened.

Using Opus in Claude code feels like working 10 years in the future until you see the bill.

As someone without kids I hate to spoil this, but dad friends talk this way to avoid making you feel bad by saying “I don’t want to get $8 beers with you because there is a tiny human who will absolutely light up the second I walk through the door, so I’m going home”

A couple of years ago I was always complaining on here about how there hadn't been any small trials using SGLT2i in CHD patients, and sure enough turns out there appears to be some benefit. Hopefully more work continues to be done breaking down the effects in different types of

Meanwhile biology researchers:

Software will increasingly become core infrastructure behind science.

Me tonight when we light the sparkler bomb

The AI researcher war between the big companies is basically just the NIL in college football for nerds.

As LLMs get better and better at coding the most useful "software engineers" become the people who are the experts in their field, the ones who know exactly what niche problems they have and how software could help solve those problems.

It's all in the electrons

How long did the researchers have to sit to publish that regression line

I am increasingly convinced that the basic research -> target -> hit lead time is going to dramatically shorten due to a combination of LLMs and specialized AI/ML models that can create optimized hit designs for testing in the lab. Because of this the value proposition in

1. "Education experts" have been saying for decades that we must wait to start teaching reading until 6-7 for neuroscientific reasons. These reasons appear, as far as I can tell, to be basically made up. Consider this recent article, which quotes a bunch of experts on this.

Townsend is now open for investment! Walkable. Beautiful. Built to last. We’re on the 5-yard line: entitled, permitted, & ready to break ground in a historic & growing downtown. DM me for the deck. If you believe in building places worth keeping, a repost is appreciated!

In the not-too-distant future they will likely have security clearances for model use in biology with how rapidly LLMs and protein generative models are advancing.

A good scientific/medical ethos would face these rare conditions and diseases and try to find ways to treat them rather than throwing our hands up and deciding the patient would be best off not ever having existed in the first place. Restoration as the guiding principal not

My therapist said this to me, and it hit me like a brick: “Your fingers would remember their old strength better if they grasped your sword."