😶 stuart reeves 😶

The thing I really don't get about academics using LLMs to generate ideas/write papers is: that's the part of my work I 💯 love: creative challenging & fulfilling. I pay the piper by sitting in committees, doing loads of paperwork & abiding by lots of procedures. Why give up the best part of the job?

a truly remarkable series of graphs

from now on will be treating all funding proposal and paper rejections as "definite wins"

UK govs current and recent past presiding over irreversible damage to the research infrastructure of the country - capacity building is infinitely harder than capacity culling

just spending their time pointing fingers at each other, as if any of that matters any more

there is a perfectly good summary of all papers I have authored or co-authored

this is commonly known as the 'abstract'

the abstract is guaranteed to convey authorial voice *by definition* because it was written by the paper's authors

/6

second, why is 'factual error' the only relevant criteria for authors objecting to AI summaries?

all AI summaries by definition take agency and authorial voice away from authors

I am not sure why the ACM seems to be intent on diluting this authorial voice

/5

taking the new policy as read, I'm not sure it should be my job to locate factual errors in LLM generated summaries

ACM is asking authors now to do the job of correcting LLM output, which is a fool's errand given that LLMs are stochastic and therefore cannot be guaranteed free of error

/4

however, it's now *no longer possible to opt-out or remove AI summarisation* from any of your ACM published papers, because apparently the ACM has changed the policy

per a recent response from dl-team@hq.acm.org, I was told "The policy has been revised to allow only corrections to the summary"

/3

"I’m an author. How do I opt out of my content being summarized?

To have a summary removed from an article, please email your request to dl-team@hq.acm.org using your institutional email address."

/2

🧵
in Dec 2025 the ACM Digital Library released its so-called AI tools, which includes "AI summarisation" (with ~100 and ~400 word versions) and "AI podcasts"

in Dec, their webpage (dl.acm.org/generative-a...) on these new "features" stated a basic opt-out method

/1

I'd put money on the FAQ itself being generated text with content free phrases like that surfacing

"emphasizes information that researchers are likely to value, including an article’s results and limitations"

astounding insights - researchers are desperate to get their hands on limitations and results because that information is absolutely never contained in the papers - praise be AI summaries

this is a change from what was on the FAQ in Dec 2025, and I quote (copied from dl.acm.org/generative-a...):

"To have a summary removed from an article, please email your request to dl-team@hq.acm.org using your institutional email address."

they emailed me back, and are now stating that their policy is as follows:

"ACM does not provide the option to opt-out of the AI summaries"

well that's interesting because I got nothing back via email (sent request to them end of Dec)

I still appear to have garbage AI slop summaries associated with my papers

are you suggesting we have to specify a list of DOIs for them to act on?

cc @acm-sigchi.bsky.social

how long before "AI tool use" is just another e-learning task, where you have to click thru pages of videos featuring "Dave" and "Sarah", complete a questionnaire at min 80% correct, while your line manager constantly emails everyone begging to please get the department compliance totals up?

Acting ethically in an imperfect world Life is complicated. Regardless of what your beliefs or politics or ethics are, the way that we set up our society and economy will often force you to act against them: You might not want to fly somew...

two interesting, related pieces by @tante.cc about how we treat technology (in this case LLMs specifically but is more general than that) and also (IMO) how social media is a terrible venue for the doing of that discourse / critique

tante.cc/2026/02/20/a...

tante.cc/2026/02/20/o...

Anthropic AI safety researcher quits with 'world in peril' warning It comes in the same week an OpenAI researcher resigned amid concerns about its decision to start testing ChatGPT ads.

the only constant is how utterly self-important and self-absorbed these people are

www.bbc.co.uk/news/article...

YouTube video by Adam Neely Suno, AI Music, and the Bad Future

I made a long video about technocapitalism and AI music

youtu.be/U8dcFhF0Dlk?...

"You have to use it. You have to trust it.": Forced Adoption of AI is The Subtext of Davos Authoritarianism is the AI Bailout; But If AI Doesn't Need a Bailout, Does it Need Authoritarianism?

I watched a bunch of AI sessions at the WEF so you wouldn’t have to.

But you should read this.

it would be a real shame if the people involved in the original decisions around purchase were held to account by the media @resprofnews.bsky.social ...

Screenshot of a THE story with the headline "Nottingham posts £85 million deficit as value of campus plummets"

Screenshot of a NottinghamshireLive news story with the headline "University of Nottingham's disastrous 'vanity project' campus could be sold for just £14."

Now I'm not an economist, but I feel like this is not great economic news.

the idea that AI criticism is / has had much concrete effect beyond perhaps some minor chin stroking from those in power about 'regulation' is absurd when you consider the enormous power differentials between AI's promoters and everyone else

what exactly is going on at UKRI?

none of this says anything about how previous cross-disciplinary calls worked out in practice - the pros, and the unintended side effects and inequities

if UKRI is to make an argument about more UK research being "bucketised" like this it had better be clear about those practical features

bye

exactly the same process is happening with uni managers

I think I might take one of these "under 20 minutes" AI skills courses the UK Govt. seems very keen on everyone doing and live-post it here... Maybe we can all learn something together! The press release sends me to aiskillshub.org.uk/aiskillsboost/ - let's go and see!

Phillipson is framing the shoveling of money into AI edutech binfire companies as 'progressive'

this is just the present iteration of "everyone should learn to code"