 Good evening. I'm Pam Horn, director of cross-platform publishing and strategic partnerships at Cooper Hewitt Smithsonian Design Museum. Thanks for being here. So please you're with us for this evening's design talk inspired by our current exhibition by the people designing a better America. The exhibition is the third in Cooper Hewitt series of groundbreaking exhibitions dedicated to socially responsible design that we first launched in 2007. Our curator of socially responsible design, Cynthia Smith, an expert in the field with an international following, Chris Cross, the United States researching designs responses to the country's continuing crisis of social and economic inequality. The 60 innovative and responsive designs installed throughout the museum are the results of designers engaging and collaborating with community stakeholders to address the complex and systemic issues at the local level with the intention of provoking change and creating opportunity. Collectively, they deliver a powerful message of optimism for a better America, as well as assurance that design has been and always will be committed to social progress. If you have not yet seen by the people, I urge you to return to Cooper Hewitt very soon, because the exhibition closes in just a few short weeks on February 26th. I love what a recent visitor wrote in her Instagram account about her experience of by the people. Quote, it was a beautiful message of optimism for achieving a more equitable and sustainable future for all of America through design. These kinds of stories and powerful messages are just what America needs right now. Several of the designs and by the people emerged from the digital realm where extraordinary advances in information sharing have been made in the last few decades. And the sheer amount of unique data produced each year is expanding almost exponentially. With us this evening are three digital pioneers, designers whose work demonstrates how information design can empower communities and enrich civic life. Sarah Williams is an assistant professor of urban planning and the director of the Civic Data Design Lab at MIT's School of Architecture and Planning. The Civic Data Design Lab works with data maps and mobile technologies to develop interactive design and communication strategies that bring urban policy issues to broader audiences. The Civic Data Design Lab created the Educational City Digits Project, which helps school students learn math by examining urban injustices in their own New York City neighborhood and is now on view in the exhibition. Tiffany Chu is a co-founder of Remix, which builds software to help cities better plan public transit. She was Zipcar's first UX designer and then became a fellow at Code for America where she worked with local governments to improve technology in the public sector. Remix's digital transit sketch tool is used by more than 50 transportation agencies around the world and is on view in by the people. And Adam Cutler is a design studio program director with IBM Design. He is responsible for IBM Design Studio in Austin, Texas while advising the other studio directors globally. He is also responsible for the competency culture and practices of design and designers at IBM, which was a generous supporter of by the people. Thank you, IBM. Our discussion will be moderated by Cynthia Smith and will be followed by a brief Q&A. Please welcome to the stage Sarah, Tiffany and Adam. I had a really exciting day in that I taught a big data visualization in society class right before I got on the plane to come here. And this year we're looking at how we can understand the data of the election. So it's been a really fun conversation. I think many of us are interested in understanding, you know, how the polls went. And so that's that's one of the things that we did. I think everybody woke up with different statistics. And so I would love to talk about that if you're interested during the panel discussion. Some of the interesting conversations we're having at MIT about how data and big data is used in that way. So thank you so much for inviting me here. I am so excited not just to be talking to you all, but also to be part of this wonderful exhibition, which really thinks about design and how we can use design to impact societal changes. And that's what my research lab is really interested in. I run something called the Civic Data Design Lab at MIT. And our goal is to take data and synthesize it into meaningful designs that we hope can affect policy change. So, you know, we always hear, you know, the story about how the rise of big data is going to change the world that we live in. But, you know, in my research lab, I really feel like big data will not change the world unless it's collected and synthesized into tools that have a public benefit. And the reason I mentioned that here is, you know, when we're talking about the idea of cognitive computing, there's still somebody behind this big data training. The data deciding what the models will be and we are learning from it, but we are also training it. And I think it's important to remember that there are still operators within these kind of computational algorithms. And so, because one of the things that Civic Data Design Lab does is tell stories with data, I'm going to tell you a small story. And then I'm going to tell you a little bit about the project that I worked on here in Cooper Hewitt. So, I am really excited. Just last week, we finished a version two of a project that I'm working on in Nairobi, Kenya. And this project started from this picture and image. These are Matatus. They're the small informal transit vehicles that 3.5 million people depend on in Nairobi. Yet, we had no information about their routes and where they went along the way. And just for, I like to give a video, see if it'll play to show you the culture of a Matatus. It's not going to play actually, I think. Oh, maybe it's going to. Well, we'll skip it. The question was, how can we leverage the ubiquitous nature of cell phone use in Nairobi, Kenya to capture data about the informal transit system, which most citizens depend upon, and open that data up for anyone to use and build upon. So, we have these small Matatus. They are the main form of transit. They're very much like a bus system that you would see here in New York or anywhere else. It just operates in a much different way. But people in Nairobi use cell phones for everything. And we actually decided to collect that data in a standardized format. Something called GTFS. Does anybody know what GTFS is? Anyone? One person. It's usually one person in the room. And that's my co-presenter surprise. But actually, you guys all used GTFS probably today. So, GTFS is the underlying data structure in Google Maps, which allows you to route your public transit. And so, one of the things that we did with the Nairobi data is when we collected it. You know, people went around on their cell phones, collected the Matatus going, and they collected it in this format so that we could use it for other tools. And I think this video might also fail. That's okay. But what I was going to show you is what a beautiful text file it creates. But from our data collection, we actually created a map. And this is the Nairobi Transit Map launched in the paper. And one of the things that we thought was really important is not just to create a data set that can be used in Google Maps and allow people to route their transit, but to actually create a paper map that everybody can use. And I think what was actually really important about the paper map was not just that it gave access to everybody, but actually it became kind of an advocacy tool and a symbol for what the system can provide citizens in Nairobi. I also think what's really important about the visual map versus the data is that we had been talking to the government all along about collecting this data set. And they were largely disinterested. And then when they saw this map, they got very interested. They made it the official map of the city. They printed out copies of it. They made billboards of it. And I think one of the things that's really interesting is when we talk about data, I think it's very abstract to people. What does it actually mean? What can I do with it? And visualizations help to synthesize the power of data into something that everyone can see and use. One thing that I want to talk about is that we open this data up for anyone to use and build upon. And that's what the GTFS standard allows. And this is actually a map. Doesn't it look very similar to the map I showed you earlier? This is actually the BRT map that the World Bank created. And so what they're doing is they're leveraging the power. So the Nairobi map now has kind of an iconic association with the people who feel very proud of their city. And so here when the World Bank was trying to propose bus rapid transit lines, these are kind of dedicated lanes for buses. They use the same graphic to do that and kind of copying a graphic visual is also really important to support and garner support. I would say that Google leveraged the power of our map as well. So if you go to Nairobi, you can now, it is the first informal transit system which is navigable on Google Maps because we collect it in the system. So it's really important to think about standards of data and because we have a GTFS standard, we were able to make that data. So really data visualizations take the complexity of data and transform that into something everybody can read. And here I'm using an example from the project in the Cooper Hewitt where I actually made a FOIA request for all the lottery data and I was able to plot that data onto a map and these green areas show how much people spend on lottery and the purples show you how much people win. And so we're really taking something that looks like this and making it clear. And that was really important about working with data for designers. The City Digits project was really about how can we take data and position it in a particular place and teach people math skills, not just people, youth math skills because I think one of the most important things right now for youth is teaching data literacy, right? We are more and more making decisions off data and students need to understand how to read and interpret that. And I think it's going to be a skill just as important as English and math. And this project was really thinking about how can we teach them mathematics using the city. And we picked lottery data because lottery is everywhere. You see it on the streets, this youth to see it in their neighborhood, but also it has a lot of math behind it. And also probability, which I was actually just talking to one of my MIT colleagues and they said probability is one of the hardest things to teach even. People at MIT, it's not something that's intuitive. And so what we did when we decided to pick the lotteries, we thought where can we get data about the lottery system? And we saw that these maps existed online of all the retailers and what we actually did is made a FOIA request, Freedom of Information Act request. And we got all of the data for the last five years. So that's every ticket sold, whether it's a scratch off ticket or a mega bonus, double plus plus. There's so many different lottery tickets and we transformed that data into a digital map which students could explore the data with. So if you actually go to citydidits.mit.edu, you'll be able to see and actually interact with each bodega and how much they win and how much they sell. And this is something we also looked at the percent of your income that's spent on lottery. And here we're looking at the comparison between Bushwick and Park Slope where you can see in Bushwick they send 3% of their income on lottery on average versus a 0.5% in Park Slope. And so the youth can begin to have a conversation about why that is using these maps. We also looked at net gain and net loss between different neighborhoods. So how much overall does the community win and lose in lottery tickets? And one of the things that this map shows is you most always lose. But the odds are stacked against you. And so the youth use these on digital tablets. These were high school students. There were exercises within the tool that asked them questions to look at the ratios that they were developing in the maps themselves. And we did also workshops to kind of ask them to understand the math behind the maps and to explain that through other kinds of design tools here using probability treat to understand the probability of winning the lottery. But one of the things that I think is really important when we talk about using data or representing data is that I think you need to ground truth your results. You have to ask people do you think this is really reality? And so one of the things that we included in the tool is we enabled them to go ask people in the field be journalists using the tool and ask people what they thought of the lottery. And so they actually were able to take interviews using their tablets. Those interviews went directly onto the map and as they were doing it and it allowed us to bring qualitative data into this quantitative analysis. And this is actually the city's digital tool where you can hear everybody's opinion of the lottery. And these are the youth being journalists in the field. And you can see actually that you could use it on your phone as well. And so some of the youth use their phones to take the interviews. And then ultimately one of the things that you need to do in order to be data literate is actually interpret results. So we asked the youth to create what we called opinions today. This group said our tour today is about how we despise the lotto. We feel that the latter is a waste of money. And they use data and the maps to make their arguments. And I think to be data literate you need to make arguments with data. We found that the youth didn't always understand maps. And so we actually created a big map that would allow them to experience the geography of the city as well. So if you're interested in looking at this tool it's at citydigits.mit.edu. We did one about the lottery and we also did one about check cashing. But I want to end here because I know I'm way over my time because I told you two stories instead of just one. But one of the things that I think is really important when we're thinking about data collection and data analytics is it's part of a whole circular model. We collect the data, we quantify it, we ground truth it, we open it up and we visualize it. And we work with teams, policy experts that know about the data and they can tell us whether they think that's accurate. And then the loop goes around again and we create a better model of our data. And one of the things that I think is really important is just as designers kind of work with data iteratively, so do modelers. And I think we'll hear a little bit more about that today. And I just want to leave us with everybody has their own data. Jacobs had her petitions and Robert Moses has his quantitative data. And if we don't actually hear those opinions from people, we could create different kinds of outcomes. So I think adding qualitative data, adding the human voice to data visualizations is really important to ensure that we're not taking kind of black box models. Thank you very much. Okay. All right. Hi, everyone. I'm Tiffany. I'm Tiffany Chu, a designer and co-founder at Remix. And I honestly can't think of a better way to frame my talk and what Sarah just basically put out on the table. A lot of what I think about every day is how to design with data to empower urban planners who are the shapers of our cities. And when I asked Cynthia and Susanna what would be the most impactful things I could share with you all today. I think there's three things. Number one, why transit matters and why we as a company were focusing on it today. Number two, a little bit about trajectory and process, hopefully things that you can draw from what we've learned from our experiences and maybe apply to what you guys do. And then number three, of course, design is great, but it's even better when it actually has real clear concrete impact. So I'm going to share two stories of impact in cities that we're working in. Okay. So first of all, why does transit matter? So for most of you, you probably live here in New York, so you already know inherently. But for a lot of the audiences that I speak with, I need to explain the concept of why transit is the backbone of our cities. It dictates where people can go, what kind of jobs they can get, what kind of lives they lead, and ultimately what type of growth a city can experience. And it's for these reasons why we actually as fellows at Code for America in 2014, we started to explore the rich world of transit and transit data. So these are my co-creators, Lizzie, Danny, Dan, and Sam, and we actually all met the first week of our fellowship by the way Code for America is nonprofit in San Francisco, helping and partnering with cities to help them move into the 21st century with regards to technology. And one of the first things that we did was actually engage in a hackathon and to take all these different types of open data and see what we could create out of it. And actually what we created was a little prototype where you can drag and drop a line on a map and suggest better transit routes as citizens to the cities of San Francisco and Oakland. So this is what it looks like. So you can click a button on the upper right hand side and a couple things will appear. You'll see a map where you can literally with your mouse really simply drag and drop and immediately you'll notice that the numbers are changing on the right hand side. So what's happening is that we're calculating how long this transit route is, how much it might cost a city to operate it, and who would potentially ride it, what the access is. And in return, what you would understand is, hey, is this a good investment for my city? So that was the prototype that we made. And we just thought it was really fun because there were so many open source mapping tools coming out into the forefront by Mapbox and Mapzen and GTFS that Sarah mentioned. All of those were just really fun things that we wanted to play with. But we launched it online just kind of quietly tweeting it to our civic tech friends being like, hey, we made a thing. And in less than 24 hours, we had gotten about 200 tweets from other city enthusiasts who are really excited to draw transit in their own city. And immediately after that, we got about 10 different articles written up. And then after that, we got about 200 emails sent to our inbox from real planners around the world saying, hey, I want to use this for my city, for my transportation plan coming up. And that was kind of the moment where we took a step back, looked at our overflowing inbox and said, why is this happening? What did we just hit upon? And can we really understand what's at the heart of this underlying response? So what we did was we actually combed through all of our emails and called up the people who emailed us. So we called up Stefan Marks who actually used to run planning at the New Orleans Regional Transit Authority. He said things like, this is a dream come true for us because we had never had a tool like this before. We talked to Bradley Tollison who works in Southern California and he said that he used to do whiteboarding before he had remix and now it's better. And we realized there was enough meat here that we wanted to do as much user research as possible so we could figure out what the challenges were that somehow our prototype was attacking that we didn't realize. So we went to the SFMTA which was across the street at the time and we talked to the planning department there and asked, hey, how are you doing your planning? We talked to planners in New York. This is David Moss who is one of the planning managers at the MTA just down the street. And today when he wants to plan out new transit routes or detours for communities in the Bronx, he uses graph paper and ballpoint pen. This is Lewis in Austin and he loves his Sharpie Markers and his paper maps and this is Bradley with his famous whiteboard. And what we realized was that there were all these different pieces of fragmented data that were really necessary to planning good transit but the tools in the data itself were so disparate and in all these different corners of either the internet or their desktop or just across a huge organization that it was really hard to wield. So ultimately what we learned was that it was really hard for cities to explain the rationale behind investing in transit as a public service due to these fragmented tools. And so what we did is we actually did even more research. We asked everyone if you could give us, actually we didn't even ask. They already told us. They gave us 10, probably 10 paragraph emails that said you need to do all these things remix and then my life would be great. So the first thing was definitely add census data. So that's what we did. We also went in and added all the existing transit routes for a bunch of different cities so that you could immediately in a couple seconds upload all of it into a single map which had never been able to be done before. And in seconds be able to visualize holistically what does your transit look like for an internal planning platform. So that was huge. And what we started to learn was that there were so many cities that were hungry for this. And I know Pam in the beginning said that we were working in 50 cities but actually over the last two years we're now working with over 200 cities who are now using remix as their planning platform. So not just the big ones that I mentioned but also the small ones across most of America from Ohio to Athens, Georgia to Bloomington, normal Illinois. All the places where you don't think have transit but the places where transit is probably the lifeline there for the people who need it the most. So in terms of our trajectory one of the biggest questions that we ask and keep asking ourselves to this date is how can we build a platform that would benefit the most number of cities. You know in a lot of settings what you could do is you could say hey I'm going to go for a model where I work with only one city and go really deep and build everything super custom for that city and it will be perfect and great. But in our perspective in order to have the most possible impact we wanted to develop a platform that was flexible enough to work with the most number of cities possible. And the way that we did that was by tracking every single piece of feedback that we have ever gotten. Either over the phone or by email or visiting them at their offices when we were in town. And it's a little sneak peek into our workflow but what we do is we actually log all of them on a board. This is we use Trello and we have all of these different categories and if someone if a planner tells us that this X, Y, Z thing needs to be improved we'll log it under the new column. And if we hear it again then we'll log it again as a comment and by the end you'll be able to step back and see hey how many times has this specific feature been asked for. Which will lead to how impactful will this feature be if we were to build it. So we spend a lot of time trying to figure out how to prioritize features but more specifically when we talk about impact. One of the product requests that we got many, many times was I want to see how far a rider can go on our network. So our question was using the data that we already have how can we visualize access and mobility. And so we built this feature called Jane and Jane is actually named for Jane Jacobs the urbanist. And what she does is you can actually drop Jane on a remix map and you can see how far she can travel on the network. Just using only transit and walking in 15, 30, 45 and 60 minutes. So I'm going to show you exactly how Jane works through two case studies. So the first one I want to talk about is King County Metro up in Seattle and then the second one is Santa Clara in San Jose, California. So for King County Metro they had this long range plan vision for 2040 and they wanted to basically visualize what will transit look like with XYZ level of investment 30 years from now. So I'm actually going to play a video. So this is a really fascinating case study for us because well first of all we had never worked with such a big city before and we never had done a long range plan. And what we realized was that with something new like this it took a third of the time to build consensus amongst 39 different municipalities. So instead of just a planning platform it was really a translation for policymakers who were not planners by trade but they had to make big decisions about planning for their communities. There's a other couple of impacts here but I think what's most important was that the long range plan was actually adopted officially on January 23rd which was just a few weeks ago and we're really excited to see what Seattle is going to bring. So I'm going to quickly go through this one and then I'm going to skedaddle. But I want to talk briefly about the big struggle that San Jose is having with investment and investment in transit specifically. So I think there's always a big tension between investing in high ridership routes which require high frequency and high density. And also coverage routes which are usually lower frequency go to more places but it's basically saying you want to provide some coverage to as many places as possible. So this is a big debate in transit planning right now how much of transit agencies budget should go to ridership routes which are high frequency and coverage routes which are lower frequency but cover more of the service area and maybe serve more people as a result. So VTAs has three different scenarios a network 70 a network 80 and a network 90 vision. And today this is what their network looks like if you were to go to San Jose if Jane were to live right there near I think one of the community colleges this is how far she could get. And I actually had the privilege of attending one of their public board meetings in April of last year and watched how data in action helped to actually sway the board. From a decision making process. So this is Jared Walker a transit consultant advising the board and presenting the data on how they should make a decision about what type of investment to put forward. The way is what she can do right now in network in concept 70 always even concept 70 much better a lot of the effect of the tannins was like yours with one part of it. But also because of some other improvements we've been able to make in that scenario. This is what happens when we go to 80. This would go to 90. Existing. 70. 80. 90. This diagram is a map of liberty of opportunity. It is a map of where you can go which is what you can do and what choices you have. So that was last year and I'm excited to share that after the first round of public engagement two thirds of the folks in Santa Clara County actually want to move in the direction of high ridership routes. So we'll see more towards the end of this year how how this project is going to shake out. So I have three takeaways that I want to leave with you. The first one is not all of us are our planners or can be planners. But I think there are different ways that we can all use data to empower the planner and their day to day jobs and thus improving cities. Number two I think this is also very much highlighted by Sarah as well. But data by itself can be you know convincing to a point but it's literally the visual stories paired with the data that get people to understand why. And then finally when enough people understand the why policy change can happen. So I'm going to say thank you and this is my contact info if you want to reach out. Good evening. Thanks for having me. I'm Adam Cutler. I'm a distinguished designer at IBM. I was appointed as a distinguished designer last April and the mission that was given to me was to define what it meant to design for cognitive computing. And as the previous two speakers have highlighted data is everything. And this what you're about to see is as much a culmination of my research for the past few months as it is an understanding of what we can do with data once it's transformed. So the impact of artificial intelligence and cognitive computing creates a new tension between ourselves and our environment. And the need for human centered design to calm this tension is greater now more than ever to define it simply a cognitive system must be able to understand. It must be able to reason and it must be able to learn for these are the pillars of cognition. We in moving to a system that has the capability to understand reason and learn means that we have shifted from transactional based computing to relationship based computing. And this is a fundamentally different way to think about how we design for these types of experiences changes everything. Whoops. So what we have to start asking ourselves is what does this mean to design for relationships for the context within those relationships and how do you design for conversations around the data that's feeding these systems. So we have three principles that we hold true about cognitive computing across the entire company. First it is meant to augment human intelligence not replace human intelligence. Second is that we provide human confidence through transparency. And third is that we provide and we will provide the skills and knowledge to engage in a relationship with cognitive systems. So I just said a whole bunch and what I found when I first started learning about all of this is that it was really helpful to level set everybody simply on what cognitive is what cognitive computing is what design is and what I think cognitive design will be. So first is cognitive straight out of the dictionary is related to or involving the human thought process. Cognitive computing is a simulation of a human thought process in a computerized model for the benefit of augmenting human capabilities. Design is the purpose planning or intention behind an action factor material object. Therefore cognitive design is the purpose planning or intent behind simulated human thought processes. It also helps to understand a little bit how cognitive computing works. So this is a very basic overview but as we've been talking about this evening nothing happens without your data. So there's three different places that data comes from. One is the data that you possess this comes in the form of customer records transactional systems predictive models etc. There's data that occurs outside of your firewall that's news events geospatial and that sort of thing. And then there's data that's coming some of it's already here but a lot of it's on the way Internet of Things sensory data audio video and imagery. And if you see along the bottom it goes from structured and active to unstructured and dark and 88% of the data that we've that we've created to date is over here in the unstructured and dark area which means we don't even know what we don't know about the data that's out there. The way that a cognitive system works is that you feed your data into this cognitive system and the first thing that the system does is it understands this. And what that means is that it is able to rapidly astonishingly fast volume and quantity and quality of the data that's going in. So it can be structured or unstructured you just dump it in the system sifts through this based on a model. The model once you've under once it's gone through the understanding process goes through a reasoning process and that allows the system to start creating hypotheses and upon which this how the understood data set can be applied to human problems. And then last it is learning from these two processes in order to to place together what it is synthesized. Therefore it interacts with a human being and in that process it learns from those from those processes and feeds it back into the understanding and reasoning system. So how do you design for cognitive. First you have to determine value. Second is establishing human and system context. Third is leveraging cognitive system traits and fourth is adapting for core cognitive experiences. So first starting with value is out there right now is there are tons of cognitive services that are at the atomic level. These are things like natural language processing speech detects things like that. They are easily acquired but the end value to the user is not that high as you start to aggregate these services together where you pair natural language processing and retrieve and rank to get better results. The value increases but what what's really critical here is when you start driving towards insight when you can start pulling something out of the data that was previously unstructured or dark to you. And most valuable is that if you can provide an action that the human can take or that the system can take on the behalf of the human this provides the most value out of a data set that is structured or unstructured in a way that we have not been able to make use of so far. There's also use cases that are really suited well for cognitive. First is engagement in which the system acts as an expert assistant to the human user. Second is exploration which collects the information that you need to be successful so that you can explore your problem area better. Third is discovery which allows us to find the possibilities around the outside of the data sets that we probably wouldn't be able to find even with the most brilliant human beings alone. There is decision which offers evidence options and reduces human bias. And then the last is policy which allows us to establish a lineage between previous successes and failures against written policy conditions. Next is establishing user and system contexts. First is a human context and again if you are a practicing designer human context should be second nature. However, this is not something that people think about consciously when diving into this type of work. So the first human context is the emotional context. There's positive and negative things like excitement, happiness and serenity on the positive end of things and stress, nervousness and depression on the other side. And it's really important for us as designers to be conscious of how the user will be predisposed in their everyday usage of this. So it might be that they're excited to be at work or they may be in a perilous situation which brings me to the physical context and that is are they seated or mobile? Are they comfortable or uncomfortable? Is the scale of their space important to them and how it works? These all must be considered when putting the system, the experience together for a cognitive system. And then there's the social context. The social context is based on how the norms of the group behave in any given situation. So the point of these two images are you wouldn't behave the way that you would at a sporting event as you would in church and vice versa or at an opera in this case. Being aware of how people will use your system, use your experience when with others is critical to be able to center how you deliver this in a way that makes sense for it to be natural. Then there are the system contexts. The first being spotlight context and this is where a cognitive system takes the forefront of an experience. This is Chef Watson that we created with Bon Appetit. If you take Chef Watson or Watson out of the mix, this becomes a reverts back to Bon Appetit's recipes. What Watson brings to this is a recombination of the ingredients and the taste profiles for cooks, both professional and home, to be able to experiment based on the suggestions that the system makes to the user. There's the companion context and the companion context comes in two different varieties. The first is an assistant. Everybody's familiar with something like Siri or OK Google in which the assistant is called forward to do what you ask and then it recedes into the background. Then there's the coach like we have with the H&R Block and Watson application in which the focus is placed on the tax preparer and the client and Watson acts as a coach in the middle, coaching both the preparer and the client. Then there's Ambient. Ambient is critical in that I think many starter cognitive solutions are going to be of this nature. The reason being is that while we have a bit of conversational interface up at the top, there are going to be plenty of systems that we are used to seeing visualized in the same way that we have everyday before it. It's just that these displays will be infinitely more intelligent than they have been in the past. So rather than being fed by a database, they're going to be fed by an intelligent system. OK, so after we've established the context and determine value, leveraging the cognitive system traits. So in order for a system to be considered cognitive, it must be adaptive and adaptive means that it can roll with whatever the user gives it. So it's constantly adjusting regardless of how a user twists and turns through this natural interaction. The system can remember and understand by being stateful and we have to be cognizant and designing to each one of these. It's not the same thing as lining pixels on a grid anymore. This is about making sure that what you have to offer can be adapted by the system that it can be remembered so that it can be recalled. Maybe sometimes a month later, maybe a week later, that it's natural so that it must be easy for the end user to use. And that means that it can't be done using lines of code. It has to leverage speech or typing. That it's transparent that the system has the ability to show its math so that we always have confidence that we can understand how a system arrived at its purpose. And then agency, the degree to which a cognitive system operates without human command and how much do we actually want of that. And then adapting for core cognitive experiences. So the first is training and there's a pre-training or a training that goes on when the data is actually fed in, as Tiffany was saying earlier, and that this is done by experts who train a system in a corpus up to running speed. And this training is the training that we do as end users so that as we're interacting with a cognitive system, if it can ask us questions like, did I help you? Did I get this close? Was I right? And we have the ability to say no, you weren't or yes, but there's actually a slight angle to this that we have the ability to communicate back, not using the system for the sake of using the system, but to train it. Second are pre-can responses. So this is a point and click response where it may be this or that or this option, but these are different in that they are surfaced intelligently so it's about how they're written and how they're exposed. And then there's a full conversational experience to this where with a trained conversation engine like Watson Conversation, a cognitive system can have that intelligent conversation with an end user back and forth, letting the conversation go where it needs to go and surfacing what needs to be surfaced at the right time. And then there are avatars which might be two-dimensional, three-dimensional, but it is a representation of the system itself that allows it to make itself physically understood and known. So in some cases this might be a robot, in other cases it might be a 2D avatar that's on screen. Being cognizant of designing to these traits is critical and the experiences are critical because what's really important here is designing for relationships. And this is the NAP relationship escalation model. And this goes back to the 30s when NAP created this as a part of his social psychology work. I don't know that many of us actually consciously think about how we form relationships and yet this is how they map out and what I really want to focus us on is the left half of this where we are coming together and then that relational maintenance. The initiating, the experimenting, intensifying, integrating and bonding, these are all critical for us as human beings but what does it mean when we are forming these relationships with a system? And that means we have to be very intentional about how we create the conversation, the tone, the way that the information is being delivered back through the cognitive system so that we can intentionally create these moments for a human to have with a system. There are implications that go with this. First is behavioral migration. Very quickly, I want you to cross your arms the way that you always cross your arms. Now I want you to cross them the other way. And for some people they can go right to it and for others it's a little bit awkward. This is similar to what we're going to have to ask people after years and years of transactional computing where we see a field, we see a button that says submit at the end, we've learned to make our Google foo work in our own way so that we can search the way that we need to where we type in a seemingly random set of words and operators and search and we get back what we expect within the first two pages. A cognitive system requires us to have full conversation so that it can derive context out of what it is that we ask. So part of our job as designers is going to have to be how do we migrate people's currently learned behavior towards something new. Second are biases and this goes in both directions. Clearly we don't want any biases going into any system while they're being created. But from a design perspective, we will also see situations where experts are biased and they will have their bias exposed by a system like this. And it is how do we reveal that there is bias with grace in a way that allows the human in the equation to work through this even if they're challenging some deep-seated beliefs about how they do their job or how they approach a problem. And then there's the uncanny valley, especially as it relates to emotion. As designers we are constantly manipulating the elements that we are given to be able to evoke some kind of emotional response whether it is the simple aha or a deeper understanding and it is with the capability of conversational UIs to be able to be written in a way to deliver certain tones to have a certain personality when the system hits a moment where it doesn't know what it's supposed to do. If the response comes back wooden or off-kilter from what the rest of the experience is we hit this moment where it doesn't feel real and the illusion is broken and we have to be very cognizant of this because this is part of how we form a long-lasting relationship and everybody's had a frustrating moment with Siri or OK Google, something like that. The three things I'll leave you with is this is a burgeoning and fledgling field and the important thing is that as designers that we're keeping these three items in sight at all times for a minimum viable interaction and that is one to determine that value for the end user if we're not providing value for the end user then it doesn't really matter what we do after people will walk away from it. Second is that we establish a clear system context so that I know what to expect out of the system as I'm using it and the third is to create that foundation for the relationship because these systems grow and learn over time and the most important thing is that as their capabilities are lower on the totem pole that they might be considered as toddlers, they must be taught, they must learn from us as we learn from them and that as their skill levels are low and we work with them this relationship is going to be what causes us to stick with them over the course of time in order to grow this. Thank you for your time. This is terrific. Thank you all. Very illuminating. As Pam said before, my name is Cynthia Smith. I'm the Curator of Social Responsible Design here at Copay-Ewitt and the curator of By the People, Designing a Better America. It's currently on display upstairs on the third floor. It's going to close at the end of this month so if you didn't get to see it tonight before the program, please come back. We would love for you to see it. It includes designs from around the United States including remix and city digits and addresses complex issues of social and spatial inequalities. I'm looking at my watch. I think we have about 15 minutes that we can have some Q&A here, a conversation here and then we're going to open it up to the audience for about 10 minutes so get your questions ready. Thank you all for sharing your depth of experience and innovative, multifaceted design approaches. I'm a big fan of all of your work so let's jump right in. Each of you use data and technology in different ways, designing and developing alternative approaches. What are the key questions you ask before you begin? Tell us a little bit about your design process, who your collaborators are and how civic engagement has informed your work. You can answer all of those or one of those. That's a lot. I'll start because this is something I think about a lot. First, design process. The last slide in my presentation talks about my design process is that often a project that I'm working on, I want to answer a question whether it's the Matatus or understanding how laddering affects different places. I start with data collection and sometimes that data collection happens on the ground through cell phones or sometimes it's a FOIA request. Then I talk to policy experts in the field that I'm working with and sometimes I'm somewhat knowledgeable but I think there's lots of people that are more knowledgeable than myself and what are the key issues and what are the things that we need to communicate most and then I start thinking about how do I want to communicate that issue. In both projects, and I'm going to focus on the design just because we're all designers here hopefully, with the Nairobi project I really thought about people in Nairobi are really proud of their Matatu system and they talk about their Matatu system much the same way that we talk about the subway. It's like a love-hate relationship and putting the data into a formalized map much the way you would see in New York or Paris was really important component to the design in order to garner support for the system to help people see it as one formalized system but also to create pride about the system. I mean people now wear t-shirts with the map and all kinds of other things so having a connection to these larger transport systems and using a style was really important to the work. I don't want to take it over because I have so much to say about this but the last thing I'll just say is that one of the things that I do with all the projects is have the community help me edit my designs and so that happened with Nairobi. Does this make sense for the way that you guys navigate? There are read maps or within the cultural context and I did that with also with city digits using really colors and different kinds of I guess design tools that made sense to the youth that working with the data. I'll answer the question around design process and collaborators so obviously what I showed today is not something that I did by myself I actually have a team of 35 back in San Francisco and it's a bunch of designers, software engineers, urbanists, planners who really care about this stuff and I think our design philosophy, if it had to be summed up in one word, it would be focus. And the reason why is we are a product company. We build a product, a platform that supports real life users every day around the world and if we were to do every single thing and build every single thing that was asked of us by our cities we would be designing by committee. We would have this complete, bloated software platform that would be just completely difficult to use because everyone had a voice in whatever 5,000 buttons that now appear and so when it comes down to the end of the day what we as designers must do is to choose and to say no. So I would say that's probably our guiding principle. I'll take the process question as well. So we use design thinking an awful lot. As I started I went through just trying to open my eyes to as much as I didn't understand I think the thing that I dealt with the most was I knew enough to be dangerous around cognitive computing but not enough to really go deep with it. So what I did was I started reaching out to people all across the company, IBM fellows, distinguished engineers and others who had a deeper knowledge base than I did and I started cataloging as much as I could and then by diverging and converging around key central themes which I'm still trying to narrow down because there's so much to explain and so much to understand in a new space to start developing these thesis. So for me this is a different type of design process where I'm designing more with words this time than I am with broad pictures and screens. It's more about designing ideas. So I'm still figuring that out. Cognitively. Yeah, cognitively. I'm very cultish right now. Well, if we're talking about cities tonight, Jane Jacobs lives on in your wonderful... Is it called an isochrone? Isochrone. An isochrone. Yes. And also you showed her at the end and you talked about qualitative data. What would Jane say about cognitive computing today? Do you think...do you have any... Oh, you're asking me? I think in some respects it would be like an amazing new tool set where she could take all those great ideas that she has had and apply them in new ways. I think the most interesting thing about cognitive to me is that the rapid assimilation of the structured and unstructured to be able to make sense out of that is going to give people like these two just amazing opportunities to make what they already have and go much further faster with it. So Sarah and Tiffany, do you see ways that this pioneering work that IBM is doing with cognitive computing might inform what you're currently or future work? I mean, already in some of the work that I do, we do some of the other projects in the lab. We use some machine learning tools which are related to cognitive computing. And so we're already doing that. And one of the things that we use it most for is to interpret kind of social media data which is just messy. It's opinionated. It has lots of words used in different kinds of contexts and ways and cognitive computing is really important to kind of make sense of it. And I guess for me, I think where my work is about having the voice of the citizens heard, I'm really trying to see if I can use some of the cognitive computing techniques to bring out this qualitative data. So this is where I would answer both questions. I think Jane would say if we hear the voice of the citizen then I'm all for cognitive computing. And so that's what I'm trying to do with that. Yeah, I think in our work we see actually a ton of attention being placed on public engagement in terms of, hey, asking the community, what do you want to see? What do you think of this plan? Do you have any feedback in that entire outreach process? So recently we were helping the City of Anchorage and their transit system collect feedback on a couple of, I think, two different scenarios actually on their website for the future of their transit system. And what poured in was, I think, close to over a thousand, there were even a thousand people living in Anchorage, but a thousand comments just really passionate about whether or not this route should go by their house or the hospital and whatnot. And the poor planner was sitting there looking at these a thousand comments and not quite knowing what to do. So maybe that would be an opportunity for taking some cognitive computing principles and translating a bunch of qualitative data, a bunch of words, a bunch of spoken data and distilling it into a recommendation, not a law or a rule or a hard and fast policy, but make recommendations that then humans can process and use local knowledge to make a decision on. Great. And so there's been a lot of conversation going on. I know when we were talking before, IBM also used the term augmented intelligence too, rather than just artificial intelligence or AI, right? So augmented intelligence will soon impact every aspect of our lives, we're told. I'd be curious to see if you all agree with that. And currently MIT and Harvard have formed an AI for public interest initiative. And they're exploring questions such as how do we make sure machines don't, we train don't perpetuate the bias that plague our society? Can we find ways to create intelligent socially responsible machines? And so, and can these AI machines reach people in underserved communities and emerging economies? Places not like the US though. We might be moving in that direction. So I'm certain that IBM has been thinking about this. They're kind of at the forefront of all of this. And from each of your vantage points, I know you work internationally and you also are working internationally. How do we address these challenges? It's going to be pervasive from what we hear. Well, we created, well we were part of forming that alliance, the open AI alliance with Apple and Google and Microsoft et cetera. And I think those folks are really tackling those issues head on. I have not had any exposure to that yet. So I'm looking forward to see what they come up with. But I know what I'm thinking about as far as, I don't know if I would go so far as to say ethics, but about what it means to think about it from an augmented perspective is it's not there to replace, it's there to add to what it is that we have as our own expertise not to remove it. And so I think by developing these design principles for us to follow, it is about intentionally keeping the user in the mix the entire time, the expert in the mix the entire time. Yeah, I've been thinking about this a lot because we often think about, how can we try or test something out in these emerging markets that wouldn't be possible here because there aren't regulations and so forth that'll camper. And I think it's something we have to be conscious of because I mean, if we don't pay attention to how we are using these tools in an ethically responsible way, we do have a chance to make mistakes that we have made in the past. And I always like, like, you know, the slide at the end of my presentation was looking at Robert Moses and Jane Jacobs. They both use data to make decisions. Like Jane had her spreadsheets of numbers and people who believed in certain things. I mean, she had a lot of data, a lot of people who signed up and Robert Moses had another data set that was based on an efficiency. Each one of them had their perspective that they were trying to push. And I think that's something that we have to remember with data is that we are also biased in the way that we present and analyze it and how to make sure you're balanced I think is one of the hardest questions because we're always going to be biased and the computer is reflective of us and it will also be biased. So it's a hard question, I think. I say that companies really need to work with people on the ground but they have their own bias so it's not always the answer, right? So I think we're ready to open up to the audience. Do we have a mic? Oh, we're going to take my mic. Sorry. So are there any questions? Okay, I'll be right over. Yeah, sure. Yeah, that was awesome, guys. Thank you so much. I think my question is specifically at least, well maybe it's not. I'm sure everyone has much cooler insights than I do but for the gentleman from IBM when you're thinking about conversational, augmented intelligence or artificial intelligence just to kind of drill into what you were talking about it seems like there's kind of two sides to the building of the relationship, right? So like one is what you said where you don't want it to screw up and people say I don't like this, I don't trust this anymore but that seems to apply from a design standpoint mostly to bots that would occupy a very specific time frame like the H&R block bot where I'm using it because I'm doing my taxes and it works for me and then I don't talk to it again until the next year ostensibly but for bots, I'm using bots as an example but for artificial intelligence it's going to occupy a more prolonged space in my life it seems like there has to be not only design to make me not want to run away from it or hide from it on the playground as you would with a friend you don't like but there has to be kind of like an aha moment where a user is actively impressed much like they would have to be actively impressed with other types of technology so I just wanted to hear your thoughts or anybody's thoughts on how you think about kind of the flip side of building that relationship for a more long term sustainable relationship I think primarily it's about designing in a humility into the interaction itself in that if you can train the conversational system to fail with grace to be able to articulate maybe why it couldn't answer the question that you were asking is a critical moment because we tend to forgive fairly easily and if we understand that there is a limit and that I think in traditional systems and a lot of the stuff we use today we hit that failure point and we know that that failure point is going to exist every single time after it until there's like a system update or we read a news article that tells us that our AI assistant in our pockets been updated I think the difference is as we start to build relationships with these systems that we understand that as we grow and learn with them that they're going to grow and learn with us so it's about again it's the humility it's not you know ha ha you don't know what this is and I do or vice versa but that there's a willingness to say that it's wrong or that it doesn't necessarily know and that the end user has the ability to give that back in the same way Hi, so we're talking about open source data giving the community access to data I'm wondering what your thoughts are on what data it might not be a good idea to share with the community I think typically in local governments if you were to look at the open data portal there's all really amazing stuff in there that you've never seen before for example in San Francisco my favorite data set is the types of trees that are planted on which blocks on a weekly basis by the public works but what you'll notice is that most times crime data and police data are heavily scrutinized and so you'll see that there's redactions happening a lot and I think identity around some of that police data should obviously be a little bit sensitive obviously there's a balance to be struck whether or not you think the police is hiding something maybe you do want access to that data but I think it really depends on whose data is it I think that's that's how one should ethically think about it is it your data or is it the city's data okay trust you hello I'm Robert what can cities and communities do to fulfill the infrastructure for data especially looking at smaller communities throughout the U.S. like small cities, towns, counties, things like that how can they build their data how can they do it in New York City but at least in a city like New York City how can these resources be beneficial to that right I mean I mean what I mean many of those I'll just briefly talk about it I'm sure definitely has more to say about this than I do but you know many of these cities they have their states provide them with the data set so the infrastructure funding comes from so like for example in Massachusetts there's so many small little towns and they don't have the capacity to build the data so Massachusetts has created a data center for the whole state so that which is incredible and has parcel data for the whole entire state and so but I think what I do want to say is that moving forward we need to make sure those kinds of infrastructures are set in place for example this morning the White House just took off its open data all of its open data today and it was a lot of local data so I think one thing we need to really think about is what what is open data is it open always and I think like what we need to work on next is not how maybe not how to generate it because there is a lot but how to save it yeah I love the question about small under resource cities because I can actually point to a lot of examples where many of them are doing a lot like I know Asheville in North Carolina has a really robust GIS and an open data team and what they do is they just default to open so instead of like having it on some you know on-premises server that's like hidden away every single time there's a new geospatial data set they'll just put it on their website and you don't need a fancy shiny open data portal that you know costs many many thousands of dollars like a big state needs but you can literally just put it on your city website and see who clicks on it and see who does something cool with it and as long as you provide it in a computer readable format like a CSV and not like a PDF you'll probably be surprised even in your small under resource city what value someone might derive from it Sarah you mentioned at the beginning that this class or seminar that you had this morning you were looking at data in the election can you talk a little bit more about what you discovered? I'll give you your mic back let's see wow summarize it in a minute maybe or two I think one of the things that I think is really important is we have a lot of questions about the election why did we wake up every morning and see this New York Times statistic that said Hillary was going to win and I think what's interesting is but you know the polling that we have the polling data that we have right now was not refined enough to actually make I mean you can make an estimate to make those estimates but for example like in Pennsylvania they weren't actually calling Pennsylvania they were you know they were using statewide polls and then kind of matching it up with voter trends in that state so I think one of the things that was happening in the election is that there were kind of mixed polls some states were getting pulled some weren't and then we're using that data to kind of make a statement about the overall election so I think that's some of the problems and then there's you know other kinds of issues but one of the things that we're trying to do in the class is kind of understand what happened but also to how to use data to impact kind of policies that we want to see in the future so what policies do we think are important that we want to visualize and communicate to Congress is a kind of important impact but do that vis-a-vis election results right so you're in Pennsylvania you voted this way, you switched this way because you're concerned about something and so how can we look at those switches and then look at particular policy and visualize those to have an impact but I think one of the things that's really interesting is that there were other sites I think there were a lot of sites that were putting them more at an even ranking I think some of the CNN data often there were other data sites in the New York Times that had a much closer race and I think the visuals had a big impact in what we thought was going to happen and so if you're a reader of the New York Times and you woke up every morning and you saw that 95 to 10% versus like if you're a reader of the New York Times or sorry CNN or Fox News I think your impression of what was going to happen in the election was very different based on these kind of different visualizations that weren't inaccurate they were based on polling data that maybe was not not good that's the best way to put it as a follow on to that and in light of what happened at OpenDataInTheWhiteHouse.gov today or two weeks ago whenever it happened one of the things that occurs to me is that you've talked about how analysis can be biased but how do we ensure that the provenance of the data itself is pure you know one of the things that occurs to me is that when the data goes back up in the next administration or if Trump puts it back up how do we know that it's actually the same data that was there to begin with for instance wow that's a hard question I mean one of the we have something called the Federal Geographic Data Committee and this is a committee that creates standards for how geographic data is distributed by the federal government and so I would hope that we would keep those standards and one of the things that we're talking about in my class is that we're starting to see the data come down Clinton was an OpenData President so Obama is the first OpenData President actually Clinton was the first OpenData President he created the Federal Geographic Data Committee he created this policy that made it and I think in administration it's changed the way that they view data and we're seeing that change I don't think it's just a Trump versus Obama it was a Clinton versus Bush change as well I think Republicans don't believe in OpenData as much as Democrats for a number of different reasons so I think sometimes we want to attribute it to Trump but in this case I think it's about a general policy of the government being too big and this data being too intrusive which I think is an interesting thing to think about I guess more practically I know there's a number of folks on the internet who have actually gone and downloaded that by a set of data files and so now it exists on someone's computer somewhere so I guess practically if they were ever to put it up we could run like a massive diff and analyze whether or not it changed hopefully it won't but yes it's a really disheartening to see but I think we'll have to fight for it this is save the data moment thank you all wonderful conversation