Latest / Shiny New Object - a Marketing Podcast / Responsible AI is a process, not a policy - ft Crystalyn Stuart-Loayza
Transcript
- 0:00Tom Ollerton: Hello and welcome to the Shiny New Object Podcast.
- 0:03My name is Tom Ollerton. I'm the founder of Automated Creative.
- 0:06We're the fastest way to create scale and test your performance
- 0:10ads. But this podcast is about the future of data-driven
- 0:13marketing. Every week or so, I have the pleasure and the
- 0:15privilege of meeting one of our industry's leaders, and this
- 0:18week is certainly no different. I am on a call with Crystalyn
- 0:23Stuart-Loayza, who is, wait for it... Head of A.I.
- 0:26Transformation at Plus Company, as well as being Chief Digital
- 0:30Officer at Citizen and Mekanism. I've known Crystalyn for a few
- 0:35years now and I've been begging her to come on the podcast and
- 0:38finally she said yes. So Crystalyn, I know you but lot of
- 0:41these lot don't. So can you give us a bit of background on who
- 0:44you are and what you do?
- 0:45Crystalyn Stuart-Loayza: Yeah, thanks, Tom. I am a New Yorker.
- 0:48I'm based here in New York and have been for most all my
- 0:51career. I've worked in-house inside of the ski industry,
- 0:55working for a resort and have been an agencies for a couple of
- 0:59decades now, helping to build digital transformation,
- 1:03intelligence, strategy solutions for many Global Fortune 500
- 1:09brands and really excited in the role that I play now to continue
- 1:12that sort of transformational work into the A.I. era.
- 1:16Tom Ollerton: Fantastic! Right, so ski industry, agencies,
- 1:20Fortune 500 brands. So in that career, cast your mind back.
- 1:26What has been the best use of your own time, energy, or your
- 1:31own money in making your career better?
- 1:35Crystalyn Stuart-Loayza: Well, first, I'm a mom of five kids,
- 1:37and so I am extremely forward thinking when it comes to
- 1:40organisation. I sort of write out and plan out almost
- 1:44everything in my life so I can tackle more with less time
- 1:47because I seem to always have less time. A few years ago, my
- 1:50husband actually got me this, The reMarkable, which is like a
- 1:53paper like writing implement that he got when it was still on
- 1:57Kickstarter, and I take so many notes, and I still love like the
- 2:00art of handwritten notes. It makes me pay attention more,
- 2:04participate more actively, but also listen more actively. And
- 2:08I've been using this for so many years and it's the perfect
- 2:11implement for me. It takes core notes into digitised, but it
- 2:15still keeps kind of that human layer of drawing and writing
- 2:19Tom Ollerton: Well, I'm a reMarkable person as well.
- 2:19that I love so much.
- 2:22However, I say that, I used it religiously and I paid for the
- 2:26kind of monthly syncing to whatever, but when A.I.
- 2:31transcripts came along, I just don't use it in the same way. I
- 2:35still will, but now you get like 100% accuracy from A.I.
- 2:38transcription, give or take, but far less so in written notes. So
- 2:42how are you- How are you using those two things together?
- 2:44Crystalyn Stuart-Loayza: Yeah, I mean, you're right. I import a
- 2:46lot of the notes from the reMarkable in as PDF documents
- 2:50and use sort of A.I. transcription over top of them
- 2:52because it can and will read like the actual notes as if they
- 2:56were any kind of form of multimodal input. So I use it
- 3:01that way but you're right, I've been more choiceful about where
- 3:03I take written notes versus where I just use A.I.
- 3:06transcription and or notes inside of Google Docs. And I
- 3:09still find the right use cases to kind of keep it written by
- 3:12hand and port it over more manually.
- 3:14Tom Ollerton: Yeah, because I find that like I use the notes,
- 3:17the reMarkable to guide the meeting as it happens.
- 3:21Crystalyn Stuart-Loayza: Yeah.
- 3:21Tom Ollerton: Because like surely we can chat and I can
- 3:23just forget about it. You have all the recall, but if I'm like,
- 3:26don't forget to ask about... It's coaching yourself to have a
- 3:30better meeting or sales call or whatever it is.
- 3:32Crystalyn Stuart-Loayza: Agreed. I will say one thing. I use it
- 3:34often when I'm in conversation with colleagues and I don't want
- 3:37to type something into something that could be an accessible
- 3:40online document or want to make notes for myself that maybe I
- 3:44don't want to have an agent trained on accessing. And I find
- 3:48actually that that's one of my biggest use cases.
- 3:51Tom Ollerton: Yeah, yeah, yeah.
- 3:52Crystalyn Stuart-Loayza: The original notebook, the reason
- 3:54for the original notebook.
- 3:55Tom Ollerton: Yeah, I was at an event last week, someone was
- 3:58handing out notebooks and it said on it, notes for things
- 4:00that I can't say out loud. And I think, that's kind of it, isn't
- 4:03it? Which was really nice.
- 4:04Crystalyn Stuart-Loayza: Exactly.
- 4:04Tom Ollerton: And have you gone colour or are you still black
- 4:06and white?
- 4:07Crystalyn Stuart-Loayza: No, black and white. Black and white
- 4:09v1.
- 4:10Tom Ollerton: And I've also bought a small- Anyway... I
- 4:12gotta stop talking about reMarkable. Right, okay. So
- 4:15apart from getting a nice tight little online-y notebook, what
- 4:21advice would you give to a marketer to become better at
- 4:25data-driven marketing?
- 4:27Crystalyn Stuart-Loayza: I think I've noticed in the last
- 4:29probably a year or so that more people are preoccupied with
- 4:33applying data-driven insights to incrementality. It's like the
- 4:37whole world is still focused on like single channels and wanting
- 4:40to demonstrate incremental improvements to a metric that
- 4:44they've been evaluating. What's good news about that is, you
- 4:47know, A.I. has gotten quite good at actually helping us to more
- 4:50rapidly do reporting and call out, be able to read and
- 4:53evaluate signals on incrementality. But what I think
- 4:57is sort of the most important thing for the future of
- 5:00data-driven marketers is actually that we use the very
- 5:02intensely human context, reasoning, pattern spotting
- 5:07ability to be able to look beyond sort of single channel
- 5:10incremental improvement and to start to look at the
- 5:13interconnectivity of data points, both attaching them up
- 5:17to things like commercial objectives and business impact,
- 5:21and also attaching them to each other to start to find the grey
- 5:23spaces that I think give us more intelligence on our customer, on
- 5:28our brand, on our products, and our business. And I think we've
- 5:32been so preoccupied with like how detailed can we get with
- 5:36those data-driven signals around incremental improvements that
- 5:40we've kind of forgotten about the power of data to unlock a
- 5:42new level of intelligence in that kind of interconnected
- 5:46space, if you will.
- 5:47Tom Ollerton: Well, that sounds dreamy, but how do you do that?
- 5:50Crystalyn Stuart-Loayza: I mean, first you have to just keep
- 5:52asking, but why? About 15 times every time you're looking at a
- 5:56report, you have to suggest, but why? But why, but why? I think
- 5:59this gets you to the point where you stop just looking at media
- 6:02incremental performance and start connecting it to creative
- 6:04effectiveness. It's when you stop looking just at your big
- 6:07brand awareness sort of brand equity metrics and start
- 6:11connecting them a little bit more to consumer, customer
- 6:13segment insights. And so you have to keep asking the "but
- 6:16why?" question and you have to make space on your teams to do
- 6:18that. You have to encourage people to ask that question
- 6:21versus just saying, "Be quiet and do the report we just did
- 6:24last month."
- 6:25Tom Ollerton: I did a talk about the book we published last year,
- 6:28actually, at a footwear retailer somewhere in Europe and they
- 6:33were saying that the most important thing that they do is
- 6:35setting up their team so everyone's got shared KPIs. So
- 6:38it's not, it's not the, you know, it's not the, the retailer
- 6:42team trying to optimise like the banner in the top left hand
- 6:45corner whilst the brand team is over- like everyone else,
- 6:48everyone is responsible for everyone's KPI. Have you seen
- 6:50that work?
- 6:51Crystalyn Stuart-Loayza: Oh, totally. We use OKRs. I'm a big
- 6:54fan of OKRs sort of system, and we set kind of universal
- 6:57objectives at the business or commercial level. And then we
- 7:00try to anchor like a modern integrated marketing team, both
- 7:03in-house teams and agencies together around those, you know,
- 7:07three or four core objectives so that we're all reporting against
- 7:11kind of the same consistent set of very important business
- 7:14challenges. And I think even just setting the tone for having
- 7:18those, making them something everyone is held accountable for
- 7:21and then holding the whole group accountable quarterly, for
- 7:24example, is a good way to action that and to make sure you're,
- 7:28you know, not getting distracted by the immense amount of data
- 7:32signals that each of us could pull up on a daily basis.
- 7:35Tom Ollerton: So we're going to talk about your shiny new object
- 7:39now, which is a first in about 320+ podcast, and this is, it's
- 7:44the first and it's responsible A.I.. Everyone knows both of
- 7:48those words mean. What's responsible A.I. for you? Give
- 7:51us the elevator pitch for that.
- 7:52Crystalyn Stuart-Loayza: Well, if A.I. is humanity's current
- 7:55shiny like maybe not so new object but definitely shiny then
- 8:00responsible A.I. for me is the way we take some of that shine
- 8:04and turn it into more warmth and sort of lasting lustre if you
- 8:09will. It's the way we look at the entire picture of the way
- 8:13A.I. is being used, is being built, is being managed and the
- 8:18way our society is also changing around integration of A.I. as a
- 8:23new part, a new member, if you will, of our society. And in
- 8:27particular, you know, I'm not solving all the world's social
- 8:30impact challenges, but in my role, I'm really focused on
- 8:34trying to understand how to give our actual team members, our
- 8:38marketers, our employees, real world decisioning tools,
- 8:43resources, and frankly, new skills, so that when they are
- 8:48approaching their day to day work, they know how to leverage
- 8:51and deploy A.I. ethically and with responsible decisioning at
- 8:55the core. And it's actually like super hard work, Tom. It's a lot
- 8:59easier to build a bunch of agents and like put a lot of
- 9:02stuff out there than it is to actually build a set of values
- 9:07that you're going to be able to access and leverage in your day
- 9:11to day work to make smarter decisions on behalf of our
- 9:14clients, on behalf of the work, on behalf of our company, and on
- 9:16behalf of yourself. But I love that work, and I've been really
- 9:20interested in learning so much lately in what responsibility
- 9:23looks like in the A.I. space.
- 9:24Tom Ollerton: So you mentioned at the start of the call, those
- 9:27like kind of four key things that you discovered. You've got
- 9:30a deck. You definitely got a deck. What's the- What's the
- 9:34framework? What's the Crystalyn method at this point?
- 9:37Crystalyn Stuart-Loayza: Well, I can't take all the credit for
- 9:39this, although I've been at it for several months. I've been
- 9:41working really closely with a couple of our partners at
- 9:43Gartner, some really wonderful analysts that have been helping
- 9:46me kind of construct and build what is going to be our
- 9:49responsible A.I. framework, and what I can tell you is we've
- 9:53learned right at the beginning, the first key learning is that
- 9:56responsible A.I. is a process, it's not a policy. And I
- 10:00certainly learned that from Gartner quickly, that you aren't
- 10:03done just because you have a policy in place or you've got
- 10:07some sort of governance team that's kind of managing and
- 10:10sending out demerits. In fact, we need to be able to take the
- 10:14process and the humans required to build the process for
- 10:17evaluating, understanding, solving dilemmas and put that at
- 10:23the centre of what you're developing. So that's one of my
- 10:25first learnings is just because the policy is there doesn't mean
- 10:28you're done. This important, overlooked part of collecting
- 10:31cross-functional humans, getting them together to learn how to
- 10:36solve dilemmas and assembling around a core set of values and
- 10:39morals is one of the most important parts. It's decidedly
- 10:42human, isn't it?
- 10:44Tom Ollerton: I like "process, not a policy." Actually sounds
- 10:47very ChatGPT, isn't it? Process... Okay, let's go, let's
- 10:51go to the list and then we can dive in in a bit because I want
- 10:54to, I want to hear everything first.
- 10:55Crystalyn Stuart-Loayza: The second thing that I've learned
- 10:57and have spent quite a bit of time on is the idea that A.I. is
- 11:01a tool mate. It is neither a colleague like someone that you
- 11:05chat with next to you as a team member, and it's not just a tool
- 11:10or a technology because of the way A.I. now can solve problems
- 11:14somewhat autonomously through agenetic A.I. because of its
- 11:17role in generative as kind of a partner that you can kind of
- 11:20workshop with. It's kind of creating this like new entity
- 11:24inside the work force. And in order to work alongside A.I. as
- 11:27a tool mate, humans have to build the skills to work
- 11:31alongside it, and managers need to understand how to manage this
- 11:35kind of new workplace of the future, and we need those
- 11:39responsible A.I. components to help us decision every day how
- 11:43we're partnering with, when, where, when's the right moment
- 11:46to bring A.I. into the mix. How are you managing a team and kind
- 11:50of giving them the right tools to be able to support this new
- 11:53relationship between A.I. tool mate and our talented humans?
- 11:58And so we've spent a lot of time really evaluating how to do
- 12:01that. And I think no one in the market is spending enough time
- 12:05or energy or dollars on the L&D required to do that well.
- 12:09Tom Ollerton: Okay. Yeah, I've got a good question for that but
- 12:11I'll let you crack on with the list. This is great. Thank you.
- 12:15Crystalyn Stuart-Loayza: The third one is I think what's cool
- 12:18about A.I. is that everybody can sort of be a builder now. I
- 12:21mean, most of us who have access to any major LLM have, you know,
- 12:24inside workspace, for example, the ability to kind of create
- 12:27new workflows or engineer sort of mini apps or code, and
- 12:32because of that, we're all starting to take on this like
- 12:34builder's posture but builders, when you build something with a
- 12:38A.I., you have to know how to weigh and understand the impact
- 12:42of what you're building. And when you democratise building to
- 12:44so many people, we're starting to get into a territory where
- 12:47it's important that we have collective ethical governance
- 12:51kind of built into the core of a Builder's Code, if you will. A
- 12:55way of working and a way of building with A.I. responsibly.
- 12:59And I think most organisations haven't built yet that Builder's
- 13:02Code. They kind of rely on individual product teams to kind
- 13:06of self optimise that or they have a governance policy. But as
- 13:10we move from like little things like teaching Gemini how I want
- 13:14it to read my emails to big things like Anthrobotics and
- 13:18like building robots with human thinking characteristics or
- 13:22building autonomous agents that are designed to autonomously
- 13:25solve problems in the complex workforce. We have to start
- 13:28building a better Builder's Code and that's a part of what
- 13:31responsible A.I. needs to do is build that deeply into the
- 13:33process.
- 13:35Tom Ollerton: Okay. And then the, the fourth corner of the
- 13:38square.
- 13:39Crystalyn Stuart-Loayza: Yeah, the fourth corner for me is
- 13:41about raising the standards for disclosure. And I say this out
- 13:46loud to everybody who's listening because I think we all
- 13:49know people are losing trust in each other's work. They're
- 13:52losing trust in their vendors or their partners, they're saying,
- 13:54"Did you just put that into Claude and send it to me?" Or
- 13:58they're starting to lose trust in what they see in their
- 14:00newsfeed and sort of ignoring some of what they see as a A.I.
- 14:03driven content. And what's happening is like an erosion of
- 14:07trust in what we're doing and how we're working with each
- 14:10other. Even employees questioning like, how much work
- 14:12did you do on this versus partner with Gemini and oh, I
- 14:16told you to partner with Gemini, but how much do I actually want
- 14:19you to use Gemini versus your own human insight? And that line
- 14:23has gotten blurrier and blurrier. And so part of what I
- 14:26think is the fourth component of responsible A.I. right now is
- 14:30the idea of disclosure and starting to more transparently
- 14:32talk about when, how and where we're implementing A.I. and our
- 14:36methodology behind how we used it so that we can build back
- 14:40some of that level of trust and also teach one another, make a
- 14:44safe space to show one another the best ways in which A.I. is
- 14:48working alongside us as a partner without trying to kind
- 14:52of hide that or begin to do some of those behaviours that
- 14:56continue to erode trust, if you will.
- 14:59Tom Ollerton: I've got a problem now. I've got a real problem
- 15:01that we're about with six, 5 minutes left to talk about all
- 15:04of this. This is so good.
- 15:06Crystalyn Stuart-Loayza: I know. I'm sorry.
- 15:06Tom Ollerton: Okay, so let's... So let's start with #4. Because
- 15:09that is the pain that I'm feeling, you know, I run a
- 15:13business with my business partners, Alex, Rhoda, you know
- 15:15well, that there's... Initial problem we've got A.I. is either
- 15:19badge of honour or it's an arse coverer, you know. And so it's
- 15:22like, "Oh, I'm sorry that thing I did was wrong because it was
- 15:25A.I." or like, "Hey, here's this thing I did, it must be good
- 15:28because I did it with AI." Help me understand point #4. What are
- 15:30you, you said you're becoming better at disclosing, building
- 15:34trust. Like what are the tools and techniques that you found
- 15:37are working at this point.
- 15:38Crystalyn Stuart-Loayza: Well, there's a big problem still
- 15:39right now because what we're seeing is a lot of companies are
- 15:43still saying no A.I. use. So a lot of our own partners as an
- 15:47agency, many of our clients are saying in their MSAs and legal
- 15:50contracts, like you're not able to use A.I., in which case you
- 15:53have a real conundrum, because we are, as a part of our day to
- 15:57day work using A.I. in our workflows. And we need to find
- 16:01ways to be open and transparent but maybe our current contracts
- 16:04haven't been updated or we haven't had the chance at a
- 16:06macro level to build that sort of sense of knowledge. So point
- 16:09one is you have to go to the contractual level and begin to
- 16:12really talk openly about what's being used and where the risks
- 16:17are, who's assuming those risks. And so that becomes
- 16:19foundational. Second, I think that as we're beginning to talk
- 16:24more about disclosures, setting a tone in place for your
- 16:28organisation and saying, "This is the way in which we're going
- 16:30to disclose." For us, I've started to test using at the
- 16:34bottom of a document like this was developed in consort with
- 16:38Perplexity. Humans did the final distillation and we had both a
- 16:43technical and a brainstorm in person in order to be able to
- 16:46deliver this final version. It kind of documents the
- 16:49methodology but it also gives kind of that clear disclosure.
- 16:53In some cases on responses to RFPs, for example. I'll have a
- 16:57whole section on what we used A.I. for in the response, so
- 17:00I'll be able to disclose to the client up front the sources and
- 17:04where I learned the information we learned, what data was pulled
- 17:07together and how we distilled it. And I'm finding that even
- 17:10just doing those little motions like simple things like that,
- 17:13eliminate some of the questions that people might have around,
- 17:17"How much did you use A.I. for?" But it also demonstrates to the
- 17:20team how smart your team is at finding new solutions at being
- 17:24kind of innovative in the way they are creating things and I
- 17:28think it opens a dialogue and at a minimum, the dialogue is one
- 17:31of the most important things you want to open so that you can
- 17:34find the right middle ground for use between you and your
- 17:37customers or you and your partners.
- 17:39Tom Ollerton: Yeah, we're finding just to share the
- 17:41prompt. You know? This came out of that plus the human thought.
- 17:44Crystalyn Stuart-Loayza: I think that's exactly right. And and
- 17:46initially, as generative came online, people were starting to
- 17:48share more prompting advice and counsel and guidance and in
- 17:52origin. And now I think as we're moving into agentic, you know,
- 17:55the disclosures get a little bit more detailed. But it's the same
- 17:58concept. We're still creating safe space and openness around
- 18:02what we're doing and developing.
- 18:03Tom Ollerton: So tell me more about what is a Builder's Code.
- 18:05That's a new one to me. What does that specifically mean?
- 18:09Crystalyn Stuart-Loayza: Well, it means different things for
- 18:10different teams. And so we're still kind of working to put
- 18:13forward our versions of Builder's Code that support
- 18:17different teams inside of our organisation. In fact, I think
- 18:20largely what a Builder's Code centres around is both a
- 18:23process. So what are some of the key points during the
- 18:26development of, let's call it a new agent, for example, during
- 18:29the development of a new agent where we have some level of
- 18:32human on the loop that is a required individual to support
- 18:37in the development, but also where in the process do you seek
- 18:40counsel from others? Where are you bringing in multi
- 18:42stakeholders to kind of do reviews together not unlike
- 18:46traditional, you know, digital ethics or digital development,
- 18:49but it requires more cross-disciplinary kind of
- 18:52review during throughout the process and then after the
- 18:56agent's been deployed, we still need that kind of human on the
- 18:59loop component to make adjustments to protect
- 19:02ourselves, to continue to build the controls and that is
- 19:06definitely a muscle that teams are trying to work to build in
- 19:10real time.
- 19:11Tom Ollerton: Right. So the code is there to ensure that you've
- 19:13got the right human sign off at the right points to ensure
- 19:16safety, accuracy...
- 19:18Crystalyn Stuart-Loayza: Yeah. And the right escalation paths,
- 19:20the right kind of decisions that can help you solve dilemmas
- 19:23because inevitably inside of this new world of A.I.
- 19:26development, like we're reaching dilemmas more often than maybe
- 19:30we were before when we were doing traditional web, like
- 19:32linear web development, if you will. So we're trying to really
- 19:35empower the teams to have that dilemma, solving capability by
- 19:39enabling a process that lets them do that, not hide it.
- 19:42Tom Ollerton: I'm going to have to, I'm gonna have to call out
- 19:44tool mate, right? 'Cause it's not a teammate, it's not a tool.
- 19:48It's a tool mate. I'm very interested to hear that that
- 19:51you're working out how that it's not a colleague. It's not just
- 19:54tool. It's this new entity, I think was the words you used
- 19:58inside the work force. That's one description of it whereas I
- 20:01have other people who work in the space who say there'll be a
- 20:05certain point where you will have an A.I. boss. You know,
- 20:08you'll come in and you'll be a call centre operative or I don't
- 20:11know, junior this that or the other. And actually, you will
- 20:15report in and be monitored by an A.I. boss, which in some ways
- 20:18might be better because they won't be tired and grumpy and
- 20:21won't need lunch, and they'll be empathetic, and you could, you
- 20:25know, chat to them any point during the day and they might
- 20:28even talk in your language, God forbid. So help me understand
- 20:32where tool mate has come from and, and why that is working
- 20:35across your organisations.
- 20:36Crystalyn Stuart-Loayza: First of all tool mate comes from
- 20:38Gartner. It's actually an approach that they have come
- 20:40forward with to help us kind of describe that in between a
- 20:44colleague, which EY has a really good study out on the four
- 20:48collar workforce which kind of looks at, you know, robots and
- 20:51agents as actual colleagues. I think tool mate sits somewhere
- 20:55in the centre which evaluates the idea that you can both have
- 21:00A.I. in the workforce, both to support you as a partner to you
- 21:04and potentially to your point overseeing teams and or large
- 21:08tasks but the fact is that it's not the same as a human. So when
- 21:12we think about the, the relationship between us and that
- 21:15tool mate, it isn't a friend but you know, Frank from Gartner
- 21:20always tells me this isn't a friend, but it's also not, not a
- 21:22friend. So there's a way in which we have to start thinking
- 21:26about the communication skills and style and ways of working
- 21:29but then there's also very much like competencies and role
- 21:32changes that have to happen as we talk to employees about their
- 21:35own path and how they're going to excel when they have A.I.
- 21:38tool mates inside of their organisation and that's a very
- 21:43complex kind of subject. So A.I. tool mate just helps us simplify
- 21:46it. Make it feel slightly less threatening and a little bit
- 21:49more focused on the role of the human changing, the role of
- 21:53partnering with the A.I. to do work that we couldn't have done
- 21:57before.
- 21:58Tom Ollerton: And so I think your first point was around it's
- 22:01a process, not a policy. So to finish up, my experience is that
- 22:06everyone is in a fairly blind panic about how to incorporate
- 22:11A.I. Is it going to steal jobs? There's a lot, there's a lot of
- 22:14fear, a lot of uncertainty. I was on stage at MadFest a couple
- 22:17of weeks ago and I said to the panellists, I said, "Describe
- 22:20the next 12 months of A.I. in your work." And someone said
- 22:23"Chaos." Someone said, "Opportunity." And someone said,
- 22:27"Distrust." It wasn't like an emphatic, "This is great." You
- 22:31know, there is, there is an anxiety there, right? So but it
- 22:35feels that if you're intellectually and practically
- 22:39at a stage where it's a process, not a policy, you seem very
- 22:43advanced. So if someone's listening to this, what are the
- 22:47initial steps that they can take to start following this fairly
- 22:51extensive, several-month long programme you've taken on with
- 22:54Gartner. What are the things that you could just hand off to
- 22:56someone listening to this podcast?
- 22:58Crystalyn Stuart-Loayza: I mean, if you're starting at the
- 22:59beginning, the most important thing is bringing key
- 23:01stakeholders together from around your organisation to
- 23:04start a list of questions that you are running up against in
- 23:08your day to day. And when I say key stakeholders, I mean legal,
- 23:12H.R., operations, your infosec teams, your product teams, your
- 23:18most creative thinkers, your president, bring your sort of
- 23:21cross-functional, cross-disciplinary group
- 23:24together and get them to share on paper the questions that they
- 23:27have. Often in those questions and when you organise them,
- 23:30you'll start to understand and be able to prioritise where
- 23:33should we look first to begin to implement a more responsible
- 23:37platform? And it doesn't have to be a platform at first. We
- 23:40started with base level governance, like the earliest,
- 23:43most simple approach is governance. And then we've moved
- 23:46up over time to start to get to the point where we better
- 23:48understand values and can start to look ahead but start
- 23:52somewhere and start where the questions are inside your
- 23:55organisation. You'll find a lot by asking.
- 23:57Tom Ollerton: There's an irony to finish, isn't there? Like in
- 23:59a world of A.I., what do you do first? You have a meeting.
- 24:04Crystalyn Stuart-Loayza: And you don't have just one. You have to
- 24:06have many. In fact, you have to have many that are open and
- 24:09really give people a safe space to bring those challenges
- 24:12forward or else you'll never find, kind of the core of where
- 24:14you need to reset unless you have that brave space.
- 24:18Tom Ollerton: Brilliant. What a lovely place to finish it,
- 24:20Crystalyn. Thank you. So if someone wants to talk to you
- 24:23about this work, where do they get in touch with you and what
- 24:26makes a message that you will respond to?
- 24:28Crystalyn Stuart-Loayza: I would love to talk to anybody working
- 24:30on these challenges right now. So please, you can find me on
- 24:33LinkedIn. Super easy to search because I have a very unique
- 24:37name and I'm super glad to chat to anybody who wants to compare
- 24:40notes. Send me a message there and I would love to connect.
- 24:43Tom Ollerton: Brilliant. Crystalyn, thanks so much for
- 24:45your time.
- 24:46Crystalyn Stuart-Loayza: Thank you, Tom.