6 Comments
User's avatar
log x's avatar

Try Gemini 3 pro and opus 4.5

Prasanna's avatar

I'm sorry but this post in itself doesn't contain any useful information outside of just linking to other reports and the X thread . The title and initial sentences makes it seem like there's actual content/takeaways to follow.

Dhruv Trehan's avatar

You are correct, Prasanna. I hear you! We had created a website for this report, and hence wanted to avoid repetition. Here is the link for that - http://whyaiscientistsfail.lossfunk.com.

Sorry for this and will keep in mind from next time. If you'd still like to read this in long-form you can also check out the LessWrong post - https://www.lesswrong.com/posts/y7TpjDtKFcJSGzunm/why-llms-aren-t-scientists-yet

kalyani khona's avatar

please consider writing notes, insights and takeaways of such research papers on substack (as long form reads). some backstory and challenges that emerged as the research progressed and behind the scene notes would help in learning from your work too.

Dhruv Trehan's avatar

Yes for sure, Kalyani. We had created a website for this report, and hence wanted to avoid repetition here. Here is the link for that - http://whyaiscientistsfail.lossfunk.com. If you'd still like to read this in long-form you can also check out the LessWrong post - https://www.lesswrong.com/posts/y7TpjDtKFcJSGzunm/why-llms-aren-t-scientists-yet

kalyani khona's avatar

Gotcha, will check it out. Thanks!