“A p-value is the most widespread statistical concept that is not definable in a single sentence“ CHECKMATE

Darren Dahly@statsepi.bsky.social · yesterday

Stop trying to explain what a p-value is in a single sentence (ICYMI) statsepi.substack.com/p/no-you-can...

Sonu AI music company loses copyright lawsuit in Germany www.reuters.com/world/german...

German court rules AI music firm Suno broke copyright rules

A German ​court ruled on Friday that AI music firm Suno violated ​copyrights and has to disclose illicit revenue, part of a wider ⁠legal battle by artists and publishers against technology firms.

reuters.com

As area chair I’m finding myself too often having to refocus discussion between authors and reviewers away from forced homogeneity, value judgements, gatekeeping, ad hominem attacks and general adversarial issues, and towards clarification and understanding. Not sure how we’ve come to be so weird…

Side point: I still find it fascinating that an entity can have intent(s) but its consciousness is debatable #semantics

Mark Riedl@markriedl.bsky.social · 20h ago

In the details, yes. because corporations are complex. But not in questions like: do corporations have legal standing (they do), and can corporations have intent (they can), and are corporations conscious (debatable).

i will say i thoroughly believe the ACL-ification of cognitive science has fucked with students' expectations -- they now expect (1) reviewers will not read or understand their papers, (2) their work must be highly novel and (3) the existence of prior work diminishes the value of their contribution

New blog: AI for Healthcare: Great benchmarks but minimal impact AI in healthcare benchmarks show soaring perf, but in the real world AI is not actually helping people much. Reasons include messy real world, weak eval, deployment, and lack of utility ehudreiter.com/2026/08/05/a...

AI for Healthcare: Great benchmarks but minimal impact

The paradox of AI in healthcare is that while benchmarks show soaring (and sometimes superhuman) performance, in the real world AI is not actually helping people much. This is partially because rea…

ehudreiter.com

YNLG Is a great opportunity for young researchers to get into NLG research and get meaningful advice from us senior researchers.

Dave Howcroft 🦔 is participating in action short of a strike ✊️@davehowcroft.com · yesterday

Are you a new researcher in #NLG? An MSc or PhD student or a postdoc new to natural language generation? Do you want to find advice and community? Check out the #YNLG workshop happening at #INLG2026! (One week left to submit) ynlg-workshop.github.io

I had thought this Mayo-affiliated surgeon pumping out these slop anatomical diagrams and blog posts was an aberration, but it's entirely consistent with the culture depicted in the lawsuit. This is just the brand now.

Jed Brown@jedbrown.org · 4mo ago

Thought leaders to physicians: "AI isn't going to take your job, but a physician using AI will." Physician using AI: "I’m Dr. Corey Welchlin, a board-certified orthopedic surgeon specializing in shoulder injuries for over three decades [...]" centerforspecialtycare.com/why-does-my-...

This sums up the frantic and coerced deployment of "AI" throughout public and private sectors. Results of even cursory audits are horrifying every time, invariably paired with forced adoption and retaliation for asserting professional norms. Colonialism is a cherry on top.

Minnesota Star Tribune@startribune.com · 4w ago

Mayo Clinic’s former director of research operations alleges she was sidelined, then terminated, for raising objections over lax standards of studies involving AI.

Are you a new researcher in #NLG? An MSc or PhD student or a postdoc new to natural language generation? Do you want to find advice and community? Check out the #YNLG workshop happening at #INLG2026! (One week left to submit) ynlg-workshop.github.io

Information asymmetry ‘lemon problem’ - unable to tell at a glance *which* goods are high quality, consumers assume they are all relatively low quality and thus only pay for cheap ones

Max Dubler 🏳️‍🌈@maxdubler.com · 3d ago

I think my most genuinely unpopular take is that a widely held and genuine consumer preference for cheaper, lower-quality goods, not private equity, is driving the decline in quality of consumer goods like clothing and housewares.

Decades ago when I studied early vision, I felt like the logic of progress in the field was a bit too formulaic (show an exceedingly well-established model is incomplete and extend it). Today I'm envying those for whom scientific progress has a logic that their community generally agrees on.

Why did Bsky show me this again today? What message could it possibly be trying to send me? 🤔🤔🤔

CJ@virmalised.us · 3w ago

Periodic reminder that you should try to finish projects rather than start new ones

This is so frustrating to read from The New Yorker. Reward hacking *long* predates LLMs. The Coast Runners paper on reward hacking (in reinforcement learning, pre-LLMs) is 10 years old! Reward hacking is not new, and not a sign of a breakthrough intelligence, much less sentience. Did AI write this?

this is really heartening to know and I also have it (hcommons.org?get_group_do...) on my bookshelf and front and centre — and hope many others do

Evan Spotte-Smith (they/them)@ewcss.info · 3d ago

I have! It's on my bookshelf in the office, and I have a digital copy as well. I love that book, and it's been a big inspiration for developing my own way of doing research since starting a group.

As of today, the use of AI becomes more transparent in the EU. Here’s how EU law makes sure AI works for you, in full openness ↓

🔹 AI-generated media, ‘deepfakes’ must be clearly labelled as such 🔸Chatbots and virtual agents must explicitly state they aren't human 🔹You must be told when AI is actively analysing you, like biometrics, emotion tracking → link.europa.eu/gTvhRX

Just about 3 days left to submit your #INLG2026 reviews! www.timeanddate.com/countdown/ge...

INLG Reviews are Due!

Countdown to Aug 6, 2026 1:59 pm. Showing days, hours, minutes and seconds ticking down to 0

timeanddate.com

Happy for the @propublicaguild.org that @propublica.org came to terms they could accept! Great job making your employer acknowledge your value through collective action!

Tell me you're not a European academic without telling me you're not a European academic

Annette Yoshiko Reed@annetteyreed.bsky.social · 5d ago

August as an academic

Doom running on nothing but Regex find-and-replace

Human Coders News@news.humancoders.com · 5d ago

DOOM tournant uniquement sur des regex find-and-replace : 544 règles, 96 MB de string

We find 2 distinct frames in IR today: the dominant liberal frame and the emerging critical frame. The former wants good systems, focuses on harm mitigation, and tries to appear apolitical. The latter wants good society (just/emancipated/democratic), leaves room for repair & rebuilding, and... 38/

Decentralization of tech infra is insufficient. We need decentralization of our social infra for moderation and governance, which is incredibly challenging but also critical for us to work on. Relevant listen: www.techpolicy.press/the-whitenes... (interview w/ @shengokai.blacksky.app) 29/

However, we must remember that 85% of global population lives in the so-called global south for whom Eurocentrism is not a meaningful antidote to American Exceptionalism. Building a search engine with EU values is solving a diff problem than building a search engine for justice and democracy. 25/

Last week I presented a keynote at #SIGIR2026 titled "Emancipatory Information Retrieval: Towards Critical IR Theories and Practices". Slides: bhaskar-mitra.github.io/files/emanci... I went in nervously excited but several people appreciated talking openly about this topic. A quick thread. 🧵 1/

Big Tech mirrors Big Oil/Tobacco/Pharma in how it exerts undue influence over public opinion thru academic/regulatory/media capture. There is growing body of work from @meredithmeredith.bsky.social @abeba.blacksky.app and others documenting diff mechanisms and strategies employed by Big Tech. 13/

If we're mutuals or you know me outside of bluesky, hit me up if you want to connect on Letterboxd, Serializd, StoryGraph, etc :)

Next time my computer does a cybercrime I'll be sure to write up a nice blog post about it.

Somebody emailed me to ask about being a reviewer at TACL, and unfortunately this ended up in my spam folder and got deleted. Apologies! If you emailed me about this and did not get a response, please email me again

📢 New blog post describing the COLM 2026 PCs’ analyses of AI use in submitted papers. gregdurrett.github.io/colm2026-blo... (temporary home, new COLM website coming soon!)