
Until that disclosure is routine, treat a leaderboard score as a provisional signal of coding ability. Accepted for publication in Communications of the ACM. Read it here: jatinganhotra.dev/blog/swe-a…

I am a computer scientist who works on natural language generation and evaluation, often in healthcare contexts. I teach at Aberdeen University.

Until that disclosure is routine, treat a leaderboard score as a provisional signal of coding ability. Accepted for publication in Communications of the ACM. Read it here: jatinganhotra.dev/blog/swe-a…

Congratulations to my student Adarsa Sivaprasad for passing her PhD vivia with minor corrections! Her thesis was on Explaining Healthcare Outcome Prediction Models to Lay Users

Seems pretty clear that at TACL, similar to other AI venues, submission numbers are rising but average submission quality is falling. Ie, many/most/all of the "additional" submissions are low quality, often LLM-generated. Which is frustrating...

TACL now only allows one new submission from first authors every three months (excluding invited resubmissions), with some exceptions. See transacl.org/index.php/tacl/…

I'm running paper titles though Google Search to see if submission is already published, as part of desk reject process. Surprised to see Google "AI Assistant" consistently gives one-page summary of the paper, including "core findings" and "key takeaways", purely from the title

Some of my students are doing a community survey on NLG evaluation practices. If you are involved in NLG evaluation, please consider filling it out qualtricsxmgbkzryq6l.qualtri…

TACL update: Papers which were submitted to ARR or xACL conferences within the past 9 months cannot be submitted to TACL. This includes papers which were unsuccessful in the ARR reviewer capacity lottery, or otherwise were desk rejected by ARR. transacl.org/index.php/tacl/…

New blog: Four Types of LLM-Written/Assisted Papers I outline some scenarios of different ways in which LLMs can be used to write papers. Discussing acceptability of these specific scenarios is more useful (and less frustrating) than generic discussions. ehudreiter.com/2026/11/03/fo…

Reminder that Aberdeen CS is looking for a new full Professor, preferably in NLP
Aberdeen CS is hiring a new (full) Professor! Basically a replacement for me, so especially interested in NLP people, but open to other CS fields as well. Closing date is 19 Nov. Feel free to ping me if you have any questions abdnjobs.co.uk/vacancy/profe…

I sometimes see faculty members take credit for successful papers from their students, but say that problems (eg hallucinated citations) are the students fault. My view is that authorship means accepting responsibility for problems as well as credit for success.

Congrats to @BarkaviS for submitting her PhD thesis! A year ago I had 6 PhD students who needed to submit their theses. All have now done so, which is great.

Please do not submit LLM-written papers to TACL! I appreciate many people (even high-school students) have success doing this with ARR, Neurips, and other mega-conf. But TACL reviewing is still almost all done by people, and LLM/AI slop is very unlikely to be accepted.

Aberdeen CS is hiring a new (full) Professor! Basically a replacement for me, so especially interested in NLP people, but open to other CS fields as well. Closing date is 19 Nov. Feel free to ping me if you have any questions abdnjobs.co.uk/vacancy/profe…

Congrats to my students Iniakpokeikiye Thompson and Jawwad Baig for winning the "Best Evaluation" award at INLG. This is for their paper "One Feedback System Does Not Fit All:Localising Data-to-Text Driver Coaching for the UK and Nigeria" ehudreiter.com/2026/09/17/lo…

New blog: Advice to Humanities Scholars on Using AI and NLP I recently gave a general talk on using AI and LLMs to a humanities audience (rabbinic scholars), and tried to explain in a simple way things that should be careful about. ehudreiter.com/2026/10/22/ad…

What is TMLR doing under a 300% increase in submissions? - Suspensions of >150 authors - Publicly naming major violations - Submission quotas distinguishing slop patterns - Desk rejections 6% → 53%; focus on clarity - AI reviews evaluating soundness blog.tmlr.org/2026/adapting-…