Questioning the Impact of LLMs on Scientific Progress
By
julienreszka
A touch underbaked. Edible, but you'll want a strong coffee alongside.
Summary
The author reflects on the current state of "Scientific AI," noting that while LLMs accelerate digital process development—debugging, stitching together pipelines, and improving code for scientific tools—there is a lingering doubt about whether this truly advances science or just speeds up existing workflows. The author questions if the ceiling is the unit operations themselves and wonders if simply doing things faster is enough to push science forward.
Key quotes
· 3 pulledI see a ton of value here along that entire pathway for LLMs, it's basically digital process development: take each step, make it better, repeat, repeat, repeat.
But the ceiling are the unit operations themselves, which sure LLMs can improve the code of those tools next.
But if science was held back by simply people not doing the same things faster, maybe this will really push us forward, but I have a nagging feeling of 'is this it?'.
You might also wanna read
How AI is Reshaping Engineering Values in Modern Programming
The author reflects on how AI and LLMs have fundamentally changed programming, noting they rarely write code by hand anymore. They explore w
cpojer.net·13d ago
Why AI's effectiveness doesn't matter: A labor force perspective
The article argues that debates about whether LLMs and AI actually "work" for specific applications are irrelevant. The author contends that
henry.codes·1d ago
Creative studio heads assess AI's real impact on the creative process
Creative studio heads share insights on how AI is reshaping the creative process, from upstream exploration to downstream delivery. The arti
Creative Boom·7mo ago
MIT and Wharton study reveals AI's contradictory impact on L&D productivity
A nuanced analysis of AI's impact on L&D (Learning & Development) professionals, drawing on a new MIT and Wharton study. The article argues
drphilippahardman.substack.com·11d ago
Why Criticizing Data Centers Matters More Than Fearing AI Language Models
The article argues that LLMs do not steal or store copyrighted content but instead learn statistical relationships to generate new sequences
substack.com·15d ago
How OpenClaw and AI agent harnesses are reshaping LLMs, inference, and CPU demand
The article discusses how AI agent harnesses like OpenClaw are transforming the LLM landscape by enabling models to automate complex tasks b
theregister.com·14d ago