Harry Harrold

We don't know how the AI trainers will work out how to identify authoritative information - maybe we will one day? Do they even care? (Maybe it's just about volume...)

We all know it's a questionable industry, but we rely on them. Be sceptical.

#charitydigitalaisummit

And maybe we need to participate on Reddit, if that's where the ai bots go to get training data.

How can we provide good quality information despite AI overviews inserting themselves between us and them? Good quality document structure, being authentic and authoritative.

#charitydigitalaisummit

Telling AI to go away (but politely) Nick shares the ways we can steer crawlers for AI services into doing what we want them to when it comes to consuming content on sites we run. Here are the most interesting and useful bits of what we ...

If you'd like to know more about tell AI bots to go away (but politely), then you might like this:

www.dxw.com/2025/04/tell...

(I'm now fascinated by him semantic html interacts with ai training bots, which I should go read about...)

#charitydigitalaisummit

We're looking at delivering answers, not pointing at places to find information.

A conversational model, Vs a search model.

And that might lead to disintermediation. (That word again...)

We've lost control of our message.

#charitydigitalaisummit

AI overviews on search sites are allegedly causing a 60% drop in site traffics, with subsequent impact on authoritative information distribution, potential fundraising visits, potential volunteering visits, etc.

Silence in the room.

...and we don't *know* the real impact.

#charitydigitalaisummit

An interesting case for Agents; you can wrap prompts in best practise, so you use fewer iterations...

Peer support and learning is critical - so go to ukcharitycamp to share and learn.

www.ukcharitycamp.com

#charitydigitalaisummit

Make the change *with* people not *to* them, because they will be worried about losing their job.

Give them an outline of AI, and specific training into prompt engineering so they can do more better and faster.

As an organisation, get your data in order first.

#charitydigitalaisummit

"it's so easy to get things very wrong.." Governance should be 'yes, if...' not 'no.'

Transparency statements ftw. "We're conscious of how AI contributes..." (to the social media landscape)

#charitydigitalaisummit

AI in funding: an experiment in time-saving. This summer, we explored how AI could save charities precious time and resources. Partnering with the Martingale Foundation, we tested practical tools that balance innovation with responsibility.

Look for the problem. What could save you the most time?

Check.

More at www.dxw.com/2025/09/ai-i...

What's *interesting* is that one of the problems we found wasn't best solved using AI...

#charitydigitalaisummary

Be conscious of your "surface area" for threat. Start with real problems. (Testify!) Learn from the fails. Share internally.

#charitydigitalaisummit

AI Experimentation canvas We have had a number of requests to share Generative AI (GenAI) policies that charitable organisations can adapt and reuse. The challenge…

AI implementation... Remember user-centred design... Stay nimble and open... Foster curiousity. (Interesting to think folk thought the Digitals of a decade ago *weren't.*)

Be part of a network and share your stuff: www.wearecast.org.uk/our-work/fre...

#charitydigitalaisummit

AI Experimentation canvas We have had a number of requests to share Generative AI (GenAI) policies that charitable organisations can adapt and reuse. The challenge…

Strategising... Understand your internal users circumstances to turn them into advocates. Use a working group to share knowledge...

Maybe with the CAST experimental canvas at
medium.com/we-are-cast/...
#charitydigitalaisunmit

Library of AI experiments - CAST

AI guardrails... Offer a safe environment to experiment in... Create principles not detailed policies... Allocate space for team learning, and mandate (a small amount of) training... Culture of openness, and progress not perfection.

#charitydigitalalsummit

"Who's heard of vibecoding?" Maybe six hands go up.

Golly, I think might be a one-eyed man.

#charitydigitalaisummit

Someone has just spelled 'potato', 'potatoe' on a slide and I have *never* felt an audience start to concentrate harder on a slide full of text *ever*.

That's the cleverest presentation strategy I have learned in *years*.

More from #charitydigitalaisummit - "Are aliens real?" Today's answer...

Understanding how young people engage with generative AI Hannah shares her work with CAST to help 15 - 17 year olds understand, examine & use generative AI in positive ways.

"Ground yourself in your users." Testify. We did that with young people for CAST - www.dxw.com/2025/01/unde... #charitydigitalaisummit

"Disintermediation" is a word I've not heard in decades; this time I'm seeing Zoe Amar using it here www.thirdsector.co.uk/zoe-amar-dis... - the last time the warning was to bricks-and-mortar shops, and the threat was Amazon. So that article will be worth a read.

"If you think about it too closely it doesn't work." Radical honesty from Mo Spargo; I applaud. The only 100% accurate model of reality is reality and that's too big to be of any use. I think I should go read this: abookapart.com/products/jus... (Took a while to find a link that wasn't Amazon.)

...I'm currently trying to see if "We've seen what happens when you let suppliers own your data, so can we not do that next time we buy a bit of software?" is a thing we can say we've learned.

So we'll see how that pans out...

Cracking deck from Charity Change Collective. Strong recommendation for this material.

Yesyesyes... We missed ethics when we did digital transformation, and look where we are now.

This is the slide. Golden stuff. #charitydigitalaisummit

"Test-and-learn" - preaching the gospel. There's something fascinating about using the "small steps" approach with Enterprise software purchase - I don't know how that works.

Library of AI experiments - CAST

Share your experiments. (Qv www.wearecast.org.uk/our-work/fre... ) And redesign processes, rather than automating them. (Apparently, we're not using our enterprise tools effectively, and the same will be true if AI unless we re-engineer our processes.)

"It's impossible to keep up"; devolve effectively. Shadow AI; personal tools used in organisational context; over-restrictive policies. Nick Scott again - principles to enable not detailed policies to restrict. Keep the humans in the loop, to stop the workslop. (I love that term.)

Charity Change Collective nearly comparing Digital transformation with the rise of AI; "skills and mindset" to get folk from 'digital' to 'curious' and test and learning into an uncertain future.... Which is what I thought Digital people were - so I'm absolutely on board here.

"AI is full of extremism" says the wise Nick Scott at Charity Digital AI Summit.

Who do I know at the @charitydigital.bsky.social Charity Digital AI summit?

UKCharityCamp is back Harry shares the return of UKCharityCamp on 13 November 2025, an unconference for the third sector.

@ukcharitycamp.bsky.social is back this Autumn.

More here: www.dxw.com/2025/07/ukch...

Tickets here: UKCharityCamp 2025

Date: Nov 13 • 9:30 am

www.eventbrite.com/e/ukcharityc...