IGHS58 - Exploring the ISO Standards: A Conversation with Stephen Clarke

Jim Merrifield (00:00.707)
Well, hello and welcome to the InfoGov Hot Seat. I'm your host Jim Merrifield and with me today is Stephen Clarke. Welcome Stephen.

Stephen (00:08.206)
Good morning, good afternoon, good night, wherever you are.

Jim Merrifield (00:11.897)
Yeah, awesome. Stephen's coming all the way from New Zealand and I'm in the US. So this is a truly international podcast, podcast, interview, whatever you want to call it. So let's get to know you a little better, Stephen. Can you provide a brief introduction of yourself, your current role, and one fun fact about yourself?

Stephen (00:33.968)
Sure, I'll start off, I'm in Aotearoa, New Zealand and out here the indigenous people are called Māori and there's a traditional way of introducing yourself so I'm gonna do that.

Go steffan mcconchion o clirich, o hol, no reira, ten a couta, ten a coorl, ten a tata o catoa. Good morning, good afternoon, Ciorra. What I just said there was that my name is Stephen Clarke. I'm originally from Scotland. It's a little bit my clan lineage, so you kind of know who I am. What my mountain was, was the Five Sisters of Kintail. And in Scottish Gaelic, that is...

Well, I said all the Scottish stuff in Scottish Gaelic, which is kind of similar to Irish if you kind of know your languages. And I just have a nice way of introducing yourself and rounding yourself out to the audience. So my name is Stephen Clarke. I'm currently an independent consultant in the information management space. Last couple of roles in government, whereas the chief archivist of Archives New Zealand, which is kind of like NARA in the US. And I was chief data officer for

New Zealand Transport Authority which is I guess our Ministry of Transport and how we get we run people around the motor around the islands. and my fun fact is that I might be the only person in the world or certainly one of the few that's eaten a piece of the moon.

Jim Merrifield (02:15.493)
That's unreal. Thanks for introducing yourself, but can you elaborate on that? What do mean you ate a piece of the moon? You know I was gonna ask a follow-up question, right?

Stephen (02:24.964)
Yeah, well, used to, so when I graduated and left my Archives and Records Masters course, well, often I worked for the National Museum of Wales in the UK. I was the records manager there. And Neil Armstrong, after he came back from the moon, did a kind of world tour in 1969. And his family is originally from Wales. So when Neil Armstrong was in Cardiff, where the National Museum is, he said, well, I'm going to send you back a piece of moon rock as a

thank you and a way of giving back to kind of my ancestral homeland. And when I was working up in the roof space, you as you dig around, you're an information manager looking at where stuff is, doing cataloging. I found this kind of aluminium or aluminum flight box and I thought, what's this? Opened it up and there's a letter from in there from NASA saying, here's your moon rock. Here's what you can and can't do with it. It remains our property always, blah, blah, blah, blah, because the moon rocks are in display in the museum.

So was this huge big big big box and a tiny little bit of cut out in the foam. So there's coat foam, they'd cut out the shape of the moon rock and in the bottom of that little piece of foam there was a bit of moon rock dust. So I kind of licked my finger, of picked up and sort of smelled it. I wonder what the moon smells like. So I picked it up and the bit of moon dust and I just I don't know what came over me but I just licked my finger to see what it tastes like and it does not taste of cheese.

Jim Merrifield (03:55.731)
My goodness, that is too funny. my goodness. I've never heard a fun fact like that, but that was probably one of the most interesting ones that I've heard. So thanks for elaborating on that. So let's get into your background. I see you're a member of the ISO working groups for records management and AI. Can you please tell us what you've been working on? What should we expect in the future from this group? Maybe take that piece by piece one at a time.

Stephen (04:25.712)
Yeah, well, I I guess how I started off with ISO, so as I came out to New Zealand 2005 to roll out our new Public Records Act, which set up a whole bunch of new expectations for government agencies. And one of the things that we did to roll that new piece of legislation out was write a bunch of new standards so organizations knew what they had to do. And at that time,

15489, the records management standard was reasonably new and it came out of Australia. And I started working with some of those folks and we developed some standards and what happened was essentially the standards we developed for New Zealand filled a bit of a gap in the international market. And so we took them to ISO and said, hey, look, we've got these and we think there's international demand for them. And we took them through the ISO process and they became standards such as 16175,

13028 and the new one for information management and artificial intelligence. And that sort of rose out of, from my perspective anyway, doing a of, pitched the idea about five years ago to Amazon and Microsoft. I'm actually blagging, went into the office and white boarded, just blagged it, white boarded to the kind of innovation people. Here, I've got this great idea, this concept for.

using machine learning to do information management. And I just white boarded up, white boarded the concept. I said, what do you think about that? And they went, wow, love it, great idea. And I said, well, I'm working for government, so we've got no money. You've got money, let's do it together. And you've got the tooling. So we worked with Amazon to do feature extraction and auto classification and retro meta-tagging and labeling of shared drive material for three use cases. One was retention and disposal.

One was personally identifiable information. And the third one was for indigenous data sovereignty. So an indigenous data sovereignty rights, you shouldn't offshore Maori data. So how do you segment that data so that you're only putting stuff up in the cloud that you can? So that was the concept. That's what I pitched to Amazon. And so essentially that's what we did. And then we did the same thing with Microsoft for SharePoint. So this was 2021. So it was kind of...

Stephen (06:49.744)
It was really sort of before Purview and even Cortex and stuff. So we were kind of really at the bleeding edge back then. Published it in 2022 to like zero fanfare. It went nowhere because no one was interested. And it was like a year later, ChatGPT came out and some everybody's interested in AI, but it just shows how things have changed. But it showed to me that it could be done. And I've been pushing within the ISO community for a few years. And finally, they got bored of

hearing me moaning about it. And the Chinese, who have got some pretty cool approaches in artificial intelligence and probably the first country in the world to do their own standards and do their own legislation around AI. And they're really pushing it, kind of stepped up and said, we're going to lead this work. And they've got loads of resources. So it's like, I'll work with you guys. Yeah, so the idea really is, if you're looking at doing transparent, open,

ethical AI? Well, the only way you can do that is if you've got records, otherwise it's all right memory, trust, opinion. So we're going to set out what those records management requirements would be for meeting the threshold for things like transparent and ethical AI that the 42,000 series sets out, sets out these expectations, but doesn't say how. So that's the bit that we're going to do.

Jim Merrifield (08:13.58)
That's awesome. That's great. Good for you. And you're one of, so you represent New Zealand. And how many other countries are kind of involved in this project?

Stephen (08:21.87)
Yeah, I'm in a slightly strange space because New Zealand and Australia have a joint working committee for standards. So I'm kind of straddling Australia and New Zealand. on that particular committee, there are 38 countries, including the US. The secretariat is the people leading it, or China, as I said. And yeah, some really cool, smart people. And I find really exciting is hearing what

different perspectives around the globe. So the African perspective is very different from the US perspective, which is very different from the Australasian perspective. And then the EU is kind of out there on its own. China is out there on its own a bit. But for the first time in human history, we genuinely have a global standard. Because normally what happens is ISO tends to serve a big chunk of the world. NIST in the US tends to do its own thing.

The Chinese tend to do their own thing, the EU sets its own standards. But at the moment anyway, and it might fragment, but at the moment all of those bodies are together. If it becomes an ISO standard, NIST publishes it, the EU publishes it, and China publishes it, and we do here in Australia and New Zealand. So fingers crossed if we do land this and we are working in liaison with the 42,000 series people that this will become the default global standard.

Jim Merrifield (09:48.389)
That's amazing. Well, I can't wait for an update. We'll probably have to have a part two of this conversation in the near future. So here's another question about how do you see information governance professionals playing a role in the advancement of GenAI technology?

Stephen (10:06.586)
Gosh, it's a big question and I would argue that...

This is our moment to shine. If you, I if I think back to when I did my archives and records masters course back 2003, and you know, it probably hadn't changed massively, the core stuff. I mean, I still had to do medieval Latin, believe it or not, and diplomatic and paleography and all that stuff. But also I was doing it back then, let's say 15489 just been published. There was a move in the global world to do standards. And I think that the skills that we've got, deep understanding of information,

and records and stuff like ontology and conceptual understanding around semantics and classification uniquely position us in the world to really understand the AI at quite a deep level. I if you look at LLMs, for example, they're basically an ontology. And as you start to kind of model LLMs and a lot of the kind of artificial intelligence processing, you really get into knowledge graphs, which is totally within our wheelhouse, you know, you think. know, ontology,

Well, that was like Socrates back in the day, 500 BC. And then you get Ptolemy, 50 AD, doing his world map and writing data tables. And archivists have still kept that so we can still produce those maps. And then Al-Hwarizmi, the Iranian academic from the ninth century who invented algorithms, because that's his name in Latin, and brought those concepts of algorithmic assessment into information management. And then

By the late 1700s, you have the respect for the idea of knowing stuff in context from the French Revolutionary reimagining of the nation as an information and semantic object. So we've got this amazing thousands of years of history of really deep understanding of information management, how it fits into societal contexts and how it fits into cultural contexts. Ultimately, all data and information is a cultural artifact, an expression of humanity.

Stephen (12:07.888)
and a way of encoding meaning between one human to another. And essentially that's what AI does is it encodes information and data and transmits it through time and space, the internet and through cultures. And that's one that you'll know yourself, that's one of the tricky things about using AI is that you get a different answer depending on what culture you're from and what language you use. So it's really culturally embedded. And I think that's something that

archivists uniquely and records managers uniquely understand in a way that our techie cohort don't because they're just not brought up, I don't think, in their academic learning to think about ethics, to think about ontological modeling, to think about semantic drift, to think about how do we do metadata modeling, how do we express the digital twin of the real world within ontological relational multi-entity.

of models and I think that's something that we can bring to the party that no one else really has.

Jim Merrifield (13:10.211)
Yeah, 100 % agree. I mean, every conversation I've had around AI is everybody saying the information and data is foundational to using any GenAI AI product, right? No matter what it is, whether it's a freebie product or something, you know, private LLM or a public LLM, it's critical to have, I guess, some data hygiene and know what data you're pointing the technology at.

So I see a lot of companies, lot of organizations, you all over the globe focusing on data hygiene. You know, that's like top of mind these days. It seems like it's like almost back to basics, if you will. Right. I mean, we've been talking about data hygiene for quite a while. And I do agree with you that, you know, the time is now for InfoGov records professionals to be in center stage, right, around AI. So.

Very good. I know, Stephen, we've talked about a lot here, especially around the ISO standards, AI, your background, InfoGov. Is there anything else you'd like to share with our audience before we let you go?

Stephen (14:20.918)
I think you hit the nail on the head when you said this is foundational. For my money, there's never been a better time for us to pitch our USP to the C-suite and the exec board. And the model that we should be using when it comes to AI within our organizational context is pushing the use case model. And once you start to bring people together to understand what's a good use case, and you think about what's the risk management, what's the risk profile,

what's the benefits return on investment. We should be playing a critical part in that role and sort of waving our hand up at our tech and business partners within the business and catapulting ourselves into this conversation and being at the top table. So I don't think there's been a better opportunity since probably, I mean, I remember coming over to Alameda in the US to talk to IBM about big data back in

back in 2011 when things like Watson and Hadoop were coming out, because that was kind of the first AI hype wave. And we didn't grasp it then, but we've got an opportunity now for ourselves as a profession to get into the limelight, get in front of our executive boards, and show the value in the USP that we can bring both to that hygiene level within making data and information AI ready.

and intelligence ready and actionable, turning it from being a digital landfill liability into being a massive business foundational data as an infrastructural asset for our organizations. And we should be driving that conversation and in the top table. And as you say, getting our moment in the sun. And I don't think we'll ever get a better opportunity in our lifetimes to shout about what we do and get ourselves, let's be blunt about this, get.

get headspace at the top and get funding to do some cool stuff.

Jim Merrifield (16:18.201)
Yeah, it's all about the money. It's all about the money. and value, of course. So, and information data have values. So, I just got to extract it with some Gen. AI. anyway, Steven, thanks so much for sharing those insights. I think you wrapped that up nicely. You know, thank you so much for taking some time to talk to me, to talk to our audience about AI InfoGov and what the cool stuff you're doing around the ISO standards of working groups. certainly...

appreciate all the hard work you've put into those standards so we as organizations can benefit from them. If you'd like to be a guest on the hot seat, like Steven here, all you have to do is submit your information through our website, infogovhotseat.com And thank you so much and enjoy the rest your day.

Stephen (17:05.808)
Thanks.

Jim Merrifield (17:07.215)
Got it.