Andy Arditi

@andyarditi

PhD @ the Bau Lab

1 216Followers

699Following

253Posts total

37.2KViews on collected posts

Against accounts of the same size

12 posts from the last 90 days, next to the under 10K follower range. reaches fewer people than peers of the same size.

Median views644this account3 166median for under 10K

Reach, %52.96%this account193.65%median for under 10K

Engagement, %1.58%this account1.50%median for under 10K

Metric	This account	Median for under 10K	Ratio
Median views per post	644	3 166	0.20×
Reach (views ÷ followers)	52.96%	193.65%	0.27×
Engagement rate	1.58%	1.50%	1.05×

Others in this range → Compare with another account → How these benchmarks are built →

Growth & engagement

How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.

Views per post

18.3K15 Jul

902

622

507

569

3.6K

666

9.6K

987

545

496

266

Last 12 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

0.74%15 Jul

1.33%

1.61%

1.58%

1.58%

0.77%

1.80%

0.41%

3.24%

2.94%

3.43%

1.50%

Reactions — likes, reposts, replies and quotes — divided by views. Median for under 10K accounts is 1.50%.

What the audience does

Likes73.0%281 in total

Reposts3.6%14 in total

Replies3.4%13 in total

Quotes3.9%15 in total

Bookmarks16.1%62 in total

Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.

The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.

Latest posts

@andyarditi This is excellent work! 266 views · 4 likes · 0 reposts · 0 replies 15 Jul 2026 Special thanks to @max_spero_ and @pangram for generously providing the Pangram credits that made this analysis possible. https://t.co/BjffofE3WA 496 views · 17 likes · 0 reposts · 0 replies 15 Jul 2026 One principle that we like: authors should take responsibility for work bearing their name. If your name is attached to a manuscript, you should certainly know what's in it, and you should be able to defend its contents - regardless of how it was generated. 545 views · 15 likes · 0 reposts · 1 replies 15 Jul 2026 This suggests a simple adjustment to peer review: deanonymize all submissions (after review process) including rejected ones - perhaps along with AI-detection scores. Authors who repeatedly submit low-effort AI-generated papers would build public reputations as "slopsters". 987 views · 29 likes · 1 reposts · 1 replies 15 Jul 2026 We also analyzed reviews. - In 2026, half of all reviews contained at least one AI-generated passage, and 17% were flagged as entirely AI. - On heavily AI-generated papers, we find that AI-generated reviews give more generous recommendations than human-written reviews. https://t. 9.6K views · 31 likes · 2 reposts · 2 replies 15 Jul 2026 AI-assisted research is here to stay - there's no going back. As researchers and as a community, we need to figure out how to adapt; we need principles for using this technology responsibly. 666 views · 11 likes · 0 reposts · 1 replies 15 Jul 2026 The best papers are still mostly human-written, though. - 91% of spotlight papers were categorized as entirely or mostly human-written, compared to about 50% for the overall submission pool. https://t.co/f6PPwEz48a 3.6K views · 24 likes · 1 reposts · 1 replies 15 Jul 2026 The prevalence of AI-generated writing has grown steadily. - In 2024, essentially no submissions were flagged. - By 2026, about a third of papers had a majority of their text flagged as AI-generated. https://t.co/iZAJ11nKAD 569 views · 8 likes · 0 reposts · 1 replies 15 Jul 2026 To quantify AI-generated content, we took every submission and every review from all three workshop editions, and ran them through Pangram, an AI-text detector. Some highlights below. 507 views · 7 likes · 0 reposts · 1 replies 15 Jul 2026 A chunk of this growth is driven by individual authors submitting more papers. - 24% of 2026 submissions were solo-authored, up from 9% in 2024. - 62 authors first-authored at least two submissions, and there was even one person who first-authored six different submissions. https 622 views · 9 likes · 0 reposts · 1 replies 15 Jul 2026 Some context first: - The MI workshop is growing rapidly. Submissions have more than doubled between each iteration: 143 -> 320 -> 801. - Two more doublings would put the workshop around 3,200 submissions, roughly where ICLR was in 2022. https://t.co/LaQpnoRsAC 902 views · 11 likes · 0 reposts · 1 replies 15 Jul 2026 Last week we wrapped up the third Mech Interp Workshop. This iteration, we noticed a lot more "AI slop" submissions, and reviewers complained about low-effort papers. So I decided to analyze the prevalence of AI-generated content at the workshop. (1/n) https://t.co/BjffofE3WA 18.3K views · 115 likes · 10 reposts · 3 replies 15 Jul 2026

Similar accounts