Transcript
Transcript: Innovate on Demand, Episode 23: Inclusive Design
Jutta Treviranus
I actually did this TVO interview once. I never wear makeup. So they gave me lipstick. And I just showed like this. And it was a live production. So I smeared it right across my face. So they had to keep the camera on this time
Nathalie Crandall
That's great. That's awesome.
Todd Lyons
I'm Todd Lyons.
Nathalie Crandall
I'm Natalie Crandall.
Jutta Treviranus
And I'm Jutta Treviranus.
Todd Lyons
And this is the Innovate on Demand podcast.
Todd Lyons et Nathalie Crandall
Welcome. Welcome, Jutta.
Jutta Treviranus
Thank you.
Nathalie Crandall
Please tell us a little bit about yourself, and what brings you to Innovate on Demand today?
Jutta Treviranus
I'm the Director of something called the Inclusive Design Research Center, which I began back in 1993, with the beginning of the web. Our primary purpose, vision and mission is to ensure that emerging technologies and their associated practices are inclusive of everyone. So especially individuals that are often left at the margins, or that are excluded due to various barriers, or unanticipated bias, or any number of reasons. And that's partly why I'm here today. I'm also a member of the Digital Academy. I'm one of the External Fellows. So and I think I was brought in specifically to look at inclusive design. I've established a framework for inclusive design that, unlike accessibility (that) begins right at the beginning, is not seen as a gatekeeping function. And I've used this framework to help to transform a number of organizations and industries, such as Microsoft and others. And so this is, in part, I think, why I'm here.
Todd Lyons
You've been working on this sort of work for maybe your entire career.
Jutta Treviranus
Yeah, I started back in 79.
Todd Lyons
Because I was reading something to the effect that that even as far back as the Apple 2, the Vic 20, the Commodore 64, you've been trying to make these sorts of devices more accessible. So I'm fascinated, over such a long period of time, how well you think we're doing compared to back in the 70s.
Jutta Treviranus
I can tell you talk a little bit about that. And you're right that when the first personal computers came out, I had been working with a number of people that were facing barriers to writing, speaking, participating in university, and I thought I hear are wonderful translation devices that we can use. So that whatever means you have of controlling something can be interpreted into something that can be used or that that you can use to control things - whatever means you have to send something can be translated. And back then, I mean, it was a very sort of a techno triumphalist time. We were imagining that the these things would continuously progress and make things better and better and better all the time. And it was very much a skunkworks period. You could take off the cover of a CPU and simply interrupt where the keyboard goes in, etc. So, it was much more of, I guess, do it yourself, hacker bottom up - Let's fix it type of culture and opportunity at the time. And I mean, I used to fix my own cars, but now it's so complex, (with) lots of black boxes and lots of things that are no longer 'tamperable'. I can't hack things quite as much. And the same thing has happened with computing systems.
Nathalie Crandall
That's interesting, because when you first started introducing yourself, what you're doing, my brain was firing on so many different workflows. Really you guys started in 93, as you said, when the internet was coming out. And so, how has that progressed? Have we succeeded in actually bringing some of these perspectives into, as you say, that initial design and not coming out at the end to make sure that that the usability is accessible, but that the whole process is accessible?
Jutta Treviranus
In my naive days, I thought things were just going to get better and better and better. But there have been these cycles where things have gotten somewhat worse. And I think there's so many opportunities and so many risks associated with this particular domain of incorporating technology. Especially at this conference where we're talking about data and AI and machine learning. I think we're missing a really great opportunity. Why the web was so successful, why (the web) was adopted, and became much more than we, the individuals that were involved right at the beginning thought would be possible, was in large part because there was involvement of individuals that had very, very compelling uses for it - but also would likely be the ones that would experience the greatest risk with it. And that's the people that Pareto would call the difficult 20%. Those people that usually markets don't actually try to address because they're so different. There's no economies of scale. But when we started the W3C, we made a very conscious effort to actually include those individuals, because they were not able to use the systems before. So they had much more compelling reasons to use the web. That caused the web to stretch and adopt a structure that would be much more adaptable and dynamically resilient. But I think that's, of course, things changed. And things got locked up. And the web was used for things like popularity metrics and things of that nature. And that then skewed it again, towards the average, the typical, the popular - the 80%. Yeah, there's been a progression, and then two steps forward one step back with respect to actually using the potential of people that are diverse.
Nathalie Crandall
So one thing I've noticed, there's a whole conversation that's happening now in the public service around those perspectives. I feel like accessibility conversations in the past, to me, always sounded like how do we make sure we can include these people? Because it's not fair for not everyone to have access? And the articulation I'm hearing today is something that feels much more empowering. It's about everyone got all of these things to offer, and how do we make sure we can put all of those things on the table? As a society, to me, that it feels different. It feels like, if that's our ultimate goal, then it might empower us to design and build things differently.
Jutta Treviranus
Right. Especially in the public sector, where it isn't that we want a very quick win. We don't want to get the low hanging fruit and then sell out. The intention, I mean it's our responsibility and intention to serve the entire populace the entire citizenry. And so yes, the previous way of looking at accessibility certainly has been as a gatekeeping - a legal risk issues - where there is an evaluation done of something before it can be released. Which is the wrong time to do it. You have to do it right at the beginning. Especially if we're talking about digital systems, and not built environment or industrial system, where we can retrofit - but digital systems morph, they change, and once the genie is out of the box, it's impossible to retrofit. So the time to think about it is right at the beginning, even if you're only thinking about a legal risk issue. But that means that we completely miss out on the opportunity. We don't need monocultures, we need diversity. Diversity is an asset. It gives us different perspectives. It allows us to detect the weak signals. It's the impetus for innovation. And the people that can't use or have difficulty using your current designs are the ones that are going to prompt you to innovate the most.
Nathalie Crandall
People have such a richness of value of perspective and work that they can bring to the table. It's how do we get to the point where we can access all of it? Where we are expected, there are these obstacles that mean that this person is unable to contribute in a meaningful way? We have to be able to figure those out. So would you say that there's a hopeful period with the onset of this whole concept of digital and transformation coming back to this being a very human and user centered thing? And it's not supposed to be about the technology, but was supposed to be how the people will use that technology?
Jutta Treviranus
There's some fundamental things we need to unlearn. We're inheriting a whole range of different mind frames that are not conducive to actually taking advantage of the diversity of perspectives and talents and skills that are there. Artificial Intelligence, let's say, is based upon statistical methods, statistical analysis, which of course means, the average, the typical, the majority rules. The data systems are based upon determining in a probabilistic or predictive way - What's going to work for the majority? Well, then what happens to the minority? So we need to rethink this. That goes way back to just research in general and research practice. And what are the chosen, or the respected means of research? That all needs to be deconstructed to some extent. The notion that there is an average or typical person, the notion that it's the top of the bell curve that is the sweet spot that we need to reach, as opposed to the edges or the full spectrum. And then business practices as well. I mean, markets, economy. We still talk about scaling and scaling as a formulaic replication of whatever happens to be with a winning design. We talk about winning, optimal, best. Well, that's not actually conducive to diversity, either, because then we're choosing one above all others, as opposed to what's the range of choices that we need? Which is the range or the spectrum of the design considerations we need to include in whatever structure we create? So a lot of the fundamental underlying things that are largely unconscious and are so baked into our processes need to be rethought. But the benefits are huge. And we were living in so many crises at the moment. There is such great disparity. And if we want to address the disparity, if we want to have a system that can have a more of a 360 degree perspective on detecting where things are going, in terms of prediction and in terms of the innovation, we need all of those and are helped by having diverse perspectives lived experience of complexity. And one of the other things of course, all of our systems tend to try to reduce things to get rid of the complexity to find the answer. It's either a binary choice, or it's a winning choice. And that's not the reality at the moment. And it means that we have all these blind spots. So yes, it's a good time, but boy, do we have a lot of work to do!
Nathalie Crandall
So what would you say to yourself 25 years ago, when you first embarked on this, what would be the biggest thing that was going to change in those 25 years? You talked a lot about one step forward, one step back, one of the forward steps that you think is actually along the right path?
Jutta Treviranus
I think there is a step away probably from some of the silos. There's a greater respect for the individuals that cross all of the disciplines. Because the goal that I have, has an economics perspective, it has a statistics perspective, it's so diffuse. So even that question, choosing one thing, is sort of illustrative of... we're frequently given these limited choices, limited time. And what my field is complex, and it's dispersed, and it's diffuse, and there isn't a sort of a single answer. It's a very complex and interconnected and entangled answer. So my answer is going to be multi perspect title as well.
Nathalie Crandall
Very interesting. It is it's definitely a complex issue. And it's probably getting more complex and not less complex as technology changes and the amount of data that we have explodes. 744-7403
Jutta Treviranus
Right, but I think where it gets less complex is at the deeper commonalities, just from a process perspective, by virtue of making room for listening to supporting diverse perspectives, you then find the deeper commonalities and those tend to be quite fundamental and simple.
Nathalie Crandall
Do you have any thing you would like to talk about maybe as being one of the external fellows for the Digital Academy? And what that role means for you?
Jutta Treviranus
The challenge that I put to myself, and that was one of the things that was asked of the external fellows, what's the project you want to work on? And so one of the projects I've set for myself is to look at what happens to outliers and small minorities in artificial intelligence and automated decision making. So the issue with data at the moment, and we're at a data conference, is that often the risks that are associated with data and those outliers and small minorities are not large enough that people actually pay attention to them, or that the warning bells go off. And because there's not large enough numbers that are facing that risk, they can be quite critical. And on the other end, the impact, especially if we're instituting evidence based governance. But evidence based governance, is that the impact because what we're how we're assessing impact is by saying, What's the single measure that has the largest impact, and that's on the largest group or the large, the largest number. And of course, that assessment is usually requires a homogenous group, and people that are at the edges are not very different from each other. And so the impact is often not large enough for government investment in those particular measures or issues. So what I'm looking at is how do we a) ensure that the individuals that are small minorities and outliers are not excluded. But then also there is such huge potential there to actually create a system that is much more adaptable, extensible. We did an assessment of services that were designed for the 100% versus services that were designed for the 80%, with the 20%, left to later or as an additional measure a segregated separate measure. And the cost if you do the second, where you leave out the 20%, the difficult 20%, as they're called, until later, is it over a five year period tends to be much more expensive, because you have lots of requests for changes for things you didn't anticipate the training, the bug reports, the help change management, change management, yeah. And then it looks like one of those houses with lots of things patched on at the end of life is reached fairly quickly. As we're if right from the beginning, you take a little bit more time, maybe a little bit more cost investment, and you actually create a system or a foundational structure that works for the 100%, then you're going to have much more resilience, adaptability, flexibility, you will have lower costs in terms of help and training and yeah, change management and all of those things. Yes. Wonderful. So I'm trying to think, specific to this audience. So most of my examples are, well, I've been doing a lot of work in the provincial governments (as they) are moving ahead much more quickly on natural, automated decisions. So one is in health, one is in transportation, etc.
Nathalie Crandall
You know, obviously, it's a big topic of conversation these days, and there's some sides of the AI that I I can totally just feel very comfortable with I think about robotic process automation, right? I think that's, you know, in in terms in the AI world, it's pretty low risk. It's, you know, just how to how do we become more efficient?
Jutta Treviranus
Well, actually, efficiency and optimization is a topic that I like to target as well, because certainly work surveillance and those sorts of things. Say if you have a call service center, and you are gathering metrics regarding efficiency, and the metric is how many calls do you process within a period of time, you're going to probably ignore the difficult ones, right? Because you want to optimize your performance. And so that the sort of efficiency optimization is that that happens everywhere, right?
Nathalie Crandall
I read an article about that actually, specifically related to the Phoenix pay system. You have all these backlog cases, and you have quite a few sort of targets around how many cases we need to get through. So some of the more complex cases are often remaining in the queue longer than their natural shelf life. So you get a lot of that happening. And that's everywhere and call centres everywhere. Sometimes we're actually measuring the wrong things. I have a colleague who's working on a research project on predictive analytics. It was trying to a) determine, you know, what is it that defines good performance or superior performance, if you will, and then b) what are some of those data core relations to some of the assessment processes to get down to it. Because we're fairly confident that in our interviewing and recruiting and other processes that we use to recruit people into jobs or to move them around through jobs that we're not always, you know, on point in what it is that we're assessing.
Jutta Treviranus
And it's counter to equity and diversity and changing the culture. Because what happens is, the data that you have is an optimization of people who are already within the job. And so if you want to change who the sort of hiring profile, then if you're unlike, or if no one like you has ever been recruited into that job and performed and given data, then you're not going to get chosen.
Nathalie Crandall
And I'm glad that his scope is around assessing the tools. Because even, just thinking about as you're talking, I see what you're saying about this enormous responsibility, that we all have to actually serve the 100%. I'm apologizing to the listeners, because I'm speaking with my hands, and our executive producers giving me the evil eye.
Jutta Treviranus
There's many other cybersecurity examples. We have a lot of National Defense people. So if you look at the false positives, in terms of the security risks, etc, it tends to be the individuals that are anomalous, and outliers. So you have a lot of false positives. The same with CRA are tax audits. The people that are unlike other individuals, which tends to be people with disabilities or because of course, you have a very different asset portfolio and different costs, they tend to get flagged and are going to get audited much more, etc. So, there's lots and lots of examples. And we take the data, and we don't, we don't think about where the data comes from. So
Todd Lyons
Thank you very much.
Nathalie Crandall
Thanks for coming in today.
Jutta Treviranus
Thank you.
Toddy Lyons
You've been listening to innovate on demand brought to you by the Canada School of Public Service. Our music is by grapes. I'm Todd Lyons, producer of this series. Thank you for listening.
