Shashwat’s Substack
Subscribe
Sign in
Home
Notes
Archive
About
Scientific discovery as a training environment for Superintelligence
Why I think automated research is the means, not just the end, for ASI
Nov 19, 2025
•
Shashwat Goel
5
2
2
Latest
Top
Discussions
Does "generalization" generalize?
Every time people talk about “generalization” in LLMs, they mean something different.
Dec 27, 2025
•
Shashwat Goel
7
2
2
How to game the METR plot
Unpacking AI's favourite exponential curve of 2025
Dec 20, 2025
•
Shashwat Goel
76
7
7
The Research Process
A live document collecting learnings about the research process as they come.
Nov 24, 2025
•
Shashwat Goel
2
3
1
what if things went well?
a human-AI collaboration.
Oct 13, 2025
•
Shashwat Goel
1
Not all bits are made equal
Focus on the higher-order bits.
Oct 8, 2025
•
Shashwat Goel
7
1
It is time to move on from MCQ evaluations
We show generative evaluations using language models are now a viable alternative to MCQs that scales cheaply.
Jul 6, 2025
•
Shashwat Goel
2
Counting Down Capabilities to AGI
What remains on the path to building generally-intelligent agents?
Jun 29, 2025
•
Shashwat Goel
19
1
2
See all
Shashwat’s Substack
My personal Substack
Subscribe
Recommendations
Daniel Paleka's Newsletter
Daniel Paleka
Shashwat’s Substack
Subscribe
About
Archive
Recommendations
Sitemap
This site requires JavaScript to run correctly. Please
turn on JavaScript
or unblock scripts