Shahan Ali Memon
@shahanmemon.bsky.social
Researching {science of AI-mediated science, metascience #SciSci, #AI4Science, generative #AI, LLMs, agents, alignment, misinformation in science}
PhD @ UW.
Visiting @ NYU & MSR
Alum @ Carnegie Mellon
Academic webpage: https://samemon.github.io
PhD @ UW.
Visiting @ NYU & MSR
Alum @ Carnegie Mellon
Academic webpage: https://samemon.github.io
Have worked (and am working) at this intersection. Happy to talk more.
www.nature.com/articles/s41...
www.nature.com/articles/s41...
Characterizing the effect of retractions on publishing careers - Nature Human Behaviour
Memon et al. examine the impact that having a paper retracted has on scientists’ subsequent careers.
www.nature.com
October 24, 2025 at 6:53 AM
Have worked (and am working) at this intersection. Happy to talk more.
www.nature.com/articles/s41...
www.nature.com/articles/s41...
Oh yes! thanks for sharing this! Super useful.
October 20, 2025 at 3:16 PM
Oh yes! thanks for sharing this! Super useful.
Was thinking of writing about this, but wondering if there are studies (empirical or otherwise) already studying this or related directly.
October 20, 2025 at 3:00 PM
Was thinking of writing about this, but wondering if there are studies (empirical or otherwise) already studying this or related directly.
Automation of generation (of papers, etc.) increases production (which is socially rewarded quite a lot), but automation of certification does not (cannot?) provide friction at the same scale (aside from BlackSpatula kind issues), and it does not get rewarded either.
October 20, 2025 at 3:00 PM
Automation of generation (of papers, etc.) increases production (which is socially rewarded quite a lot), but automation of certification does not (cannot?) provide friction at the same scale (aside from BlackSpatula kind issues), and it does not get rewarded either.
But it seems intuitive in some sense that production and certification gap may be increasing too because while AI does automate the seemingly generative part of science (writing) perceivingly well, the certification (review, evaluation, feedback, slop-control) just cannot be scaled to same extent.
October 20, 2025 at 3:00 PM
But it seems intuitive in some sense that production and certification gap may be increasing too because while AI does automate the seemingly generative part of science (writing) perceivingly well, the certification (review, evaluation, feedback, slop-control) just cannot be scaled to same extent.
On this point, I am wondering if you have seen any work on increasing production-certification gap in science (possibly due to AI), and whether someone has argued for this point. Authors of ‘Can AI slow science?’ have argued for production-progress gap.
October 20, 2025 at 3:00 PM
On this point, I am wondering if you have seen any work on increasing production-certification gap in science (possibly due to AI), and whether someone has argued for this point. Authors of ‘Can AI slow science?’ have argued for production-progress gap.
You beat me to this post!!
October 20, 2025 at 2:39 PM
You beat me to this post!!
Wow. So exciting. Can’t wait to read this!
October 15, 2025 at 10:26 PM
Wow. So exciting. Can’t wait to read this!
We had human-centered AI.
Then human–AI interaction.
Then human-AI collaboration.
Then human–AI alignment.
Now human-AI relationships.
Next up: welfare programs for the hybrid human-AI babies.
Then human–AI interaction.
Then human-AI collaboration.
Then human–AI alignment.
Now human-AI relationships.
Next up: welfare programs for the hybrid human-AI babies.
October 15, 2025 at 5:20 AM
We had human-centered AI.
Then human–AI interaction.
Then human-AI collaboration.
Then human–AI alignment.
Now human-AI relationships.
Next up: welfare programs for the hybrid human-AI babies.
Then human–AI interaction.
Then human-AI collaboration.
Then human–AI alignment.
Now human-AI relationships.
Next up: welfare programs for the hybrid human-AI babies.