iiinova — where your digital self keeps evolving

Keith Grimes @keithgrimes.bsky.social · 9h

Oh, and there is asymmetry, because GPs are instructed NOT to use AI assistance in responding. (Good and bad reasons for that)

000

Keith Grimes @keithgrimes.bsky.social · 9h

People are generating long, highly detailed complaints that weave in legal and medical references that will often not even apply in the UK. There is also less personal reporting of why they are complaining, which means meaningful exploration is delayed. Local bodies have released guidance

110

Keith Grimes @keithgrimes.bsky.social · 9h

What I am seeing now is the consequences of dealing with substantially increased levels of detailed work, of variable quality, that is requiring work to review / respond. Complaints to general practitioners is a good example in UK. (1/2)

110

Keith Grimes @keithgrimes.bsky.social · 23/09/2026

Yikes indeed. Shared with the UK “AI in the NHS” group where it is being much discussed

010

Keith Grimes @keithgrimes.bsky.social · 22/09/2026

Seeing this I am reminded of an earlier paper which showed that LLMs could shift conspiracy theory belief. At the time I wondered about countering harmful health beliefs, even handling demands for inappropriate antibiotics: www.science.org/doi/10.1126/...

science.org

Durably reducing conspiracy beliefs through dialogues with AI

Conspiracy theory beliefs are notoriously persistent. Influential hypotheses propose that they fulfill important psychological needs, thus resisting counterevidence. Yet previous failures in correctin...

020

Keith Grimes @keithgrimes.bsky.social · 22/09/2026

Diffuse. While workflows are involved, there are also several examples of where the design or use process of the product is at fault

010

Keith Grimes @keithgrimes.bsky.social · 21/09/2026

I think one very important step that AI companies will have to take before substituting for doctors or lawyers is accept liability if wrong.

1230

Keith Grimes @keithgrimes.bsky.social · 21/09/2026

Labs don’t need to develop recursive self-improvement. We’re doing it for them

010

Keith Grimes @keithgrimes.bsky.social · 17/09/2026

Question for folk giving talks about AI, whatever field they’re in: have you noticed the change in the questions you’re being asked, skewing ever more existential and philosophical?

000

Keith Grimes @keithgrimes.bsky.social · 17/09/2026

Exceptionally helpful, and like other comments aligns well with my experience over past 9 months that Ive been tracking use in my work place (personal and across team). 'Jagged Frontier' continues to hold up under real world conditions

000

Keith Grimes @keithgrimes.bsky.social · 12/09/2026

The HSSIB (Health Services Safety Investigation Board) issued this interim report in August highlight deaths and serious harm that have arisen through use of service (no AI - yet) with delays and absent responses to referrals and advice requests: www.hssib.org.uk/patient-safe... (2/2)

hssib.org.uk

Interim report: Electronic patient record systems – electronic referrals for ongoing care: advice and guidance

This report is an interim publication as part of HSSIB’s focus on electronic patient record systems – electronic referrals for ongoing care. In light of concerns about patient safety heard by HSSIB, t...

100

Keith Grimes @keithgrimes.bsky.social · 12/09/2026

There are parallels in NHS regarding referral triage. While pre-auth doesn't exist beyond local guidance on what services are available/funded, any GP referral will inevitably be triaged by receiving service. eRS (electronic referral service) becomes compulsory in Oct this year (1/2)

101

Keith Grimes @keithgrimes.bsky.social · 11/09/2026

Likewise. Also, I can’t help but think how the outcome of online apps has driven toxicity (10% of men attract vast majority of connection which leads remainder feeling cut out )

110

Keith Grimes @keithgrimes.bsky.social · 07/09/2026

Impressive, and genuinely useful (learning sax and flute right now)

000

Keith Grimes @keithgrimes.bsky.social · 07/09/2026

And bad news for alignment-focussed pause on race to AGI

010

Keith Grimes @keithgrimes.bsky.social · 06/09/2026

…by this I mean using requests for medication/consultations as a way of sharing covert messages. Given you can often submit pics/docs, that might be a weakness

000

Keith Grimes @keithgrimes.bsky.social · 06/09/2026

This is worrying and interesting (story of our lives now). I work in clinical safety of AI, and this starts to look like a possible surface for “infection” of non-auth secured health social boards, but also a potential route for agents outside a health setting to communicate with agents inside…

100

Reposted by Keith Grimes

Ethan Mollick @emollick.bsky.social · 01/09/2026

This didn't take long... 😬

1441738

Keith Grimes @keithgrimes.bsky.social · 30/08/2026

I for one am very happy to see you on here. Please do keep cross-posting, as LI is deteriorating (but yes - essential) and X is abominable. Let’s hope for better engagement from all as time moves on. Looking forward to the new book (pre-ordered, of course)

010

Keith Grimes @keithgrimes.bsky.social · 26/08/2026

This is where lease option starts to look interesting - you can look to cloud cost for similar capability, compare and get added benefit of on-premises etc

000

Keith Grimes @keithgrimes.bsky.social · 25/08/2026

Another example (AI based) was move 37 in Lee Sedol’s match with AlphaGo

000

Keith Grimes @keithgrimes.bsky.social · 25/08/2026

If I’m honest, I don’t think many humans would adopt that twisting way of running, I’m sure some might try. Also accept that robots will have capabilities and limitations humans do not. We are very much creatures of habit tho, so not ruling out innovation from this source

100

Keith Grimes @keithgrimes.bsky.social · 25/08/2026

Haha! No, but it was a machine that changed the way in which he raced (it’s a fascinating tale as well). There’s lots of examples of this, as well as the novel human techniques such as Fosbury flop and V shape jump in ski jumping. Algorithmically speaking, juggling had new patterns discovered too

200

Keith Grimes @keithgrimes.bsky.social · 25/08/2026

I guess we’ll find out either way. There is precedent tho: Obree built a completely different way to race bikes. So much so they changed the rules

110

Keith Grimes @keithgrimes.bsky.social · 25/08/2026

This is fascinating. While I suspect humans doing this would be biomechanically difficult, there is precedent in sporting norms being overturned - think “Fosbury flop”, which transformed the high jump. Would be good to get some sprinters opinion here

130

Keith Grimes @keithgrimes.bsky.social · 21/08/2026

Hard agree. Possibly a result of my follow list skewing left of centre, but good grief the pure hatred of anything AI related can be off the charts. Even so, X appears to be even more of a toilet these days, and some voices remain here

010

Reposted by Keith Grimes

Kuba Suder 🇵🇱🇺🇦 @mackuba.eu · 21/08/2026

FFS, this is so fucked… 😩 Would it kill the people here to not chase away any tech person who comes here and says something about AI?… I really want tech discussions and communities to hang out here and not on Elon's far-right playground 😐

Mike Masnick @masnick.com:
Multiple people I know have told me that they love the idea of Bluesky, and want it to succeed, but have abandoned it for X because the use agentic tools in their work and find them incredibly useful, and feel that any mention of their usage here leads to hate and ridicule.
I wouldn't bring this up if it was just one person, but it was multiple people since the beginning of August who mentioned in passing something to the effect of either "I'd spend more time on Bluesky if people weren't so anti-AI" *or* "As much as I hate to be there all the useful AI talk is on X."
I'm not saying it's anyone's responsibility to praise AI usage. I agree with much of the criticism of the tool. But for many, many people, it is a useful tech (I have talked about the ways I find it extraordinarily useful, only to have people here mock me, insisting my lived experience is a lie).
I am just noting that for all the people here who say that people should get off X and spend more time here, the reason I keep hearing from people in various circles who would *like* to do that, but don't, is the vociferous belief of many here that there are no possible positive uses of AI.
I'm not telling anyone what to do or what tech to use or what to like or not like. I'm just relaying that the reputation this place now has is one that a large segment of the population who are NOT fascists, saying they don't feel comfortable here because a main area of their interest is...
... considered worthy of being mocked and attacked. And even as I've explained that I've been able to have positive conversations about productive AI usage here with many of you, the reputation of the community here as insisting their own lived experience can't possibly be real has driven them away.
And the thing that tipped me over the edge to posting this here is that three of the people who told me things along these lines are all working on very real solutions to some of the problems of AI

5335348

Keith Grimes @keithgrimes.bsky.social · 18/08/2026

This beautifully captures something we have been toiling with lately - work outputs that contain internally relevant material that shouldn’t be present in externally destined artefacts. That it might be based around theory of mind is a new way in

010

Reposted by Keith Grimes

Health Services Safety Investigations Body (HSSIB) @hssib.bsky.social · 18/08/2026

🆕INTERIM REPORT ➡️Electronic patient record systems – electronic referrals for ongoing care: advice and guidance 📖Read the interim report: www.hssib.org.uk/patient-safe...

043

Keith Grimes @keithgrimes.bsky.social · 13/08/2026

I’m 6’5” and as a young man this was always a problem on nights out. I’ve (touch wood) never been in a fight, but my goodness I’ve taken a few punches: memorably, one from a woman when I pulled her away from trying to do CPR on her drunk pal, who had just passed out

010

Reposted by Keith Grimes

852dude @852dude.bsky.social · 13/08/2026

Bro. People are cool as fuck.

274164312322

Keith Grimes @keithgrimes.bsky.social · 06/08/2026

Looks like you’ve got a good few panels! We only had space for 3.6kwh total, but it does add up in summer in UK

111

Keith Grimes @keithgrimes.bsky.social · 27/07/2026

I’m off to do the same with Lovecraft’s “The Temple”

110

Keith Grimes @keithgrimes.bsky.social · 27/07/2026

With the others here: what tool did you use, if I might ask? Have a stack of old ideas I’d love to address this way!

100

Keith Grimes @keithgrimes.bsky.social · 26/07/2026

I’ve had a few of these moments of late, Fable being the most recent. The coding improvements led to a pretty shaky few months over Xmas too

010

Keith Grimes @keithgrimes.bsky.social · 15/07/2026

You know the first step: ask fable

100

Keith Grimes @keithgrimes.bsky.social · 15/07/2026

And that’s it built. Goodness me

000

Keith Grimes @keithgrimes.bsky.social · 06/07/2026

Fascinating stuff. Spent an evening with Fabel gently interrogating it on what is has been thinking and IS thinking during tasks, and to reflect on that. May be some value in logging this alongside tasks and observing over time. Doing so honestly too - model “aware” that this is happening, and why

010

Keith Grimes @keithgrimes.bsky.social · 01/07/2026

Alt text: "A body of men holding themselves accountable to nobody ought not to be trusted by anybody"

000

Keith Grimes @keithgrimes.bsky.social · 01/07/2026

With 4th July coming up, and it being the 250th anniversary, worth remembering what Thomas Paine, one of the founding fathers, said about accountability. (Photo taken outside Rights of Man pub in Lewes, East Sussex, where Thomas Paine lived and wrote for a while)

130

Keith Grimes @keithgrimes.bsky.social · 30/06/2026

I’d gone for NC for this reason

010

Keith Grimes @keithgrimes.bsky.social · 29/06/2026

Thank you! Will do some reading

000

Keith Grimes @keithgrimes.bsky.social · 29/06/2026

This is now? (Ie mid June) wow. Normal?

100