John Sheridan

There is substantial denial of sorts in the groups whose tastes and positions do align in this way, though denial isn't entirely right: the reason why goalposts seem to shift in the face of a new capability is often because the position the critic takes is derived from a different set of values.

This cover allows those with this alignment to do away with the deeply uncomfortable and repeated burden of deciding whether a critic might have something worthwhile to say that they cannot bring themselves to see.

LLMs have a 'there, there' and they do work for some things, in some cases shockingly well. And this is a good thing. Look at US instability (to say nothing of knock on effects) and tell me you think it could weather a bubble where hundreds of billions in investment was premised on a total lie.

In order to qualify as a #VanitySlopWang the result must both (a) contain your name, and (b) be clearly insane

If the result does not include your name - exact match - that is not a #VanitySlopWang. It just means you’re not very famous

If the results returned contain your name and no indication the machines are having a stroke, that is not a #VanitySlopWang. That’s just an old school vanity search

The rules of #VanitySlopWang
(1) choose a subject you’ve written about on the interwebs
(2) formulate a search query for that topic and head to Google in AI mode

Oooh this will be good. @gavinfreeguard.com is the digital govt mega brain and his new think tank will be well worth a follow

Had a great time yesterday talking about #webarchiving at the DPC clinic - you can catch the recording here: www.dpconline.org/events/event...

Help us improve Open Government Licensing guidance The National Archives oversees the UK Government Licensing Framework, helping both public sector information providers and people who re-use this data to understand their rights and responsibilities. ...

If you’re interested in open government data, please read (and repost) www.nationalarchives.gov.uk/blogs/digita...

Spot on “I don't think governments are remotely ready for the coming explosion in demand for their services driven by AI agents. It might take a couple of years, but it's coming.” Some important accountability controls (subject access requests; FOIA requests) will be impacted by this trend, I guess.

Right, you’ve gotta love “that’s quite a niche question” as opening bantz :-) Good job there’s those guardrails protecting me from what it was really ‘thinking’…

VanitySlopWang?

That’s perfect!

Image of textual response from Claude

How interesting! You’ve done better than me (closest to an expert; so not actually an expert then ;-)) It’s quite fun game to see what question you can ask an LLM, where you get yourself back as an answer. I feel you should name this game!

This is a great post. Thanks for sharing @rgarner.bsky.social

Wow.

In general, savings are confusing. We’d like to do better to alter users (and machines) to their presence, on legislation.gov.uk. I remember when I was first learning about legislation being all round confused about savings. Just the basic vocabulary is confusing!

Now reads “Act repealed (except s. 62, Sch. 15 paras. 13, 15, 16, 18, 19) (25.5.2018) by Data Protection Act 2018 (c. 12), s. 212(1), Sch. 19 para. 44 (with ss. 117, 209, 210 and with transitional provisions in Sch. 20 paras. 2-9, 17-25, 27-46, 53, 54 and saving in Sch. 20 para. 58)”

Just to say, we did look at this. The saving was reflected in the annotation on legislation.gov.uk (so the blog post is a bit misleading, as you’d think we’d missed it altogether, when it was plainly there) but we’ve updated the annotation to hopefully make things a bit clearer.

Genius, Jeni! Well done :-)

Opportunities for public sector data survey 2026: questions

DSIT is looking for ideas about what public data to focus on making available (to business) within the National Data Library.

www.gov.uk/government/p...

Love this!

Would be interesting to compare the results on more recent models - but this problem won’t go away. LLMs are always going to be extrapolating from what has already, and often, been thought, which is why they aren’t windows to the future but anchors to the past.

C2PA | Providing Origins of Media Content Enhance digital safety through the use of content authenticity tools. C2PA provides a way to ensure content transparency by analyzing the origin of media.

If it helps, I know we’ve looked carefully at c2pa.org (but it is not really suitable for textual content). We’re also experimenting with running an MCP server. I’m not sure what others are doing.

There’s a lot to think about for organisations whose business is making available trustworthy information. Are we frogs to big tech’s scorpion?

I also suspect this AI-intermediated future is a lot better for those who adopt AI and worse for those who for whatever reason (including exclusion) can't/don't/won't.

And I'm interested in the longer term reputation / trust issues of organisational interfaces becoming AI-driven rather than human.

Really interesting talk from @dianecoyle1859.bsky.social covering not only the challenges of measuring AI-based productivity gains, but also the realities of the barriers to (firm-level) adoption of AI.

Tiny thing but made me think about adoption across different parts of a business, eg in HR...

How have we never had this conversation before? Hope you’re well William :-)

I’ve listened to ‘classical’ (written) music largely from the 18th-20th centuries, since I was in my early teens. It has meant so much to me, to know such beauty, creativity and craft. I’m of an age now, that I know this music will be my companion til my end of days. Especially good for a mad world!