Subbarao Kambhampati (కంభంపాటి సుబ్బారావు)
@rao2z
AI researcher & teacher @SCAI_ASU. Former President of @RealAAAI; Chair of @AAAS Sec T. Here to tweach #AI. YouTube Ch: http://bit.ly/38twrAV Bsky: rao2z
For anyone interested, here are the videos of the three ~50min each lectures on the reasoning/planning capabilities of LLMs/LRMs that I gave at #ACDL2025 in Riva Del Sole resort last week 1/

Sorry, @icmlconf, equating "human bribing a human" with "human trapping an LLM which shouldn't be part of reviewing to begin with" is a false equivalence (and bad example of anthropomorphization of LLMs!)..
ICML’s Statement about subversive hidden LLM prompts We live in a weird timeline…
I suggested to the ICLR PCs a couple of years ago that we should print "Outstanding Reviewer" on the conference badge of the top 10% of the reviewers. It is a small token of appreciation but it also boosts their confidence in engaging with others at the conference.
True --- I think also that more positive incentives for reviewers would be good, like a badge for being a continuously solid reviewer or so x.com/nlopitz/status…
#AI Community: Our flagship conferences have peerless reviewing standards! 😎 The reviews: R1 As I delve into this paper.. R2 You missed these recent arXiv papers that we just posted: •R2 & friends 1 •R2 & friends 2 •R2 & friends 3 R3 Who/what is Adam? #AIAphorisms
Things that you didn't know you would need to be thankful for this year: #NeurIPS2025 reviewers who actually wrote their reviews themselves--rather than farming them to LLMs. (Yes, Virginia, it is often darned obvious which reviews are mostly LLM generated! 🙄)
(Visiting Iceland early August and may have some free time in Rekjavik.. wondering if anyone here is involved in #AI research in Iceland?🤔)
Going back after another fun, inspiring if brain-frying couple of days at @MSFTResearch HQ..


Makes sense to scout for places in case you get a mail from Zuck.. λ λ λ

One of the lesser known peaks of greater #Tempe

Do Think Tags Really Help LLMs Plan? A Critical Evaluation of ReAct-Style Prompting Siddhant Bhambri, Mudit Verma, Subbarao Kambhampati. Action editor: Li Li. openreview.net/forum?id=aFAMP… #reasoning #prompting #prompts
About that "we finished within 4.5 hours claim", I am surprised that GDM didn't/couldn't spring for additional scaling/FLOPS/TPU's and complete the darned IMO in half the time! Given that clock time comparison is meaningless anyway, they may as well impress gullible folk by…