---
id: "workshop:dont-think-of-a-robot"
type: "InnovateUS workshop"
title: "Don't Think of a Robot: Low-Tech Questions for AI Planning"
url: "https://innovate-us.org/dont-think-of-a-robot"
date: "2026-03-31"
duration: "60 minutes"
series: "Ideas in Action"
instructors: ["Jed Miller"]
recording: "https://www.youtube.com/watch?v=2GRiPqYm7DM"
transcript: "automatic, from the recording"
source: "InnovateUS (innovate-us.org)"
licence: "CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/), as stated on innovate-us.org"
---

# Don't Think of a Robot: Low-Tech Questions for AI Planning

InnovateUS workshop · 31 March 2026 · 60 minutes

Artificial intelligence is often framed as a technical challenge: algorithms, models, data pipelines, and procurement decisions. But before governments think about tools, they need to ask better questions. This workshop invites participants to “not think of a robot”, and instead focus on the low-tech, human-centered questions that should shape any AI initiative from the start.

Participants will explore how to clarify the public problem they are trying to solve, identify who benefits (and who may be harmed), and assess whether AI is even the right solution. The session emphasizes governance, accountability, ethics, and institutional readiness over technical complexity. By grounding AI planning in mission, values, and public trust, this workshop equips public leaders with a structured approach to decision-making that works regardless of the specific technology involved.

## Learning goals

- Distinguish between technology-driven and problem-driven approaches to AI adoption.

- Identify key governance, accountability, and ethical questions that should precede AI procurement or deployment.

- Apply a structured “low-tech questions” framework to guide responsible AI planning in their own context.

## Speakers

- **Jed Miller**, Senior Non-Resident Fellow, Accountability Lab

Part of the series [Ideas in Action](https://innovate-us.org/workshop-series/ideas-in-action).

Recording: https://www.youtube.com/watch?v=2GRiPqYm7DM

## Transcript

Automatic transcript of the recording: names and terms can be misheard, and the times can be a few seconds off. Each time links to that moment of the video. The transcript alone: https://innovate-us.org/dont-think-of-a-robot.txt (plain text) and https://innovate-us.org/dont-think-of-a-robot.json (every segment with its start and end second).

**Jed Miller**

[0:00](https://youtu.be/2GRiPqYm7DM?t=0) I'm very pleased to see you, and I know that there are a lot of conversations that you have about change management and a lot of conversations you have about digital transformation. Having been through a lot of processes like that myself over several years in several fields, what I try to bring to this conversation is, as I've said, a tech-agnostic perspective about some of the dynamics that pertain any time you're adopting a new technology. My hope in doing that is to make it easy for anyone to walk away with a few basic principles in mind rather than getting too immersed in the details. Along the way, I have some links that I'll share and that Agueda will share about other resources you can use.

[1:03](https://youtu.be/2GRiPqYm7DM?t=63) My background includes working initially in journalism, and then for some large NGOs, and more recently as a consultant working with philanthropies like the Open Society Foundation and NGOs as well, and more recently with Accountability Lab, always with a focus on digital transformation, always with a focus on the realism that I think we often forget as we get into hype cycles. This map that I've tried to give is distilled down to six steps, expectations, inclusion, acquisition, implementation, assessment, and adaptation. My premise is that whatever your technology, whatever you're confronting with AI, these basic steps pertain, and my hope is that you will come away almost like with a rubber band on the finger to remember, what are our expectations?

[2:21](https://youtu.be/2GRiPqYm7DM?t=141) And to ask that at the outset, AI gets seen as a savior, AI gets seen as a big danger, and those challenges can often influence how your own conversations are going internally and even how you come to the question. When we look at the expectations we have for digital transformation and for AI in particular, we need to ask ourselves, even if not all our colleagues are asking, and even if the debate we see in the media is very large scale, AI doom, AI gloom, AI excitement, AI hype, what are we actually looking for an AI to do? We can say, oh, we're going to build a bot to deliver our data or our agency's data. That's better than saying, oh, it's just going to help streamline our process or reduce staff or deliver more information to more constituents quickly.

[3:24](https://youtu.be/2GRiPqYm7DM?t=204) But to ask at the outset, how will we train the AI to use that data? What data will we use? What vendor will we use? To ask ourselves, what will it really look like when I bring my constituents, my web visitors, my citizens, my residents into these encounters? To ask that early, to ask that fast, and to check yourself about how you feel about AI. To check and say, oh, I'm baffled, like we asked in the quiz, or I get it because I use it at home, or I trust it, but my colleague doesn't, or I trust it, but I don't know how to translate it into seven languages, so it becomes a more practical challenge. To ask yourself about your posture early. Ah, I've deleted a slide.

[4:25](https://youtu.be/2GRiPqYm7DM?t=265) All right, so the next thing I want to ask about is who is in the room, and again, I don't know why not all my icons are displaying. We will provide a tidier chart, a tidier presentation for you when we're done. The ideal way to create a digital plan is to have different stakeholders in the room together, whether literally or figuratively, and different stakeholders can mean different things along a vertical access, for instance, the people with more and less power, your boss, an elected official, your head of budget, a local resident, a local resident with less or more economic or social power, and different stakeholders can mean different things along horizontal, along a horizontal access as well. If you're planning a healthcare app or a healthcare bot, are there frontline healthcare providers in the room with you when you're designing the app? Is the hospital's head of privacy or your health agency's head of privacy in the room when you're designing the bot?

[5:26](https://youtu.be/2GRiPqYm7DM?t=326) Do you have someone fluent in Spanish or Vietnamese? Do you have an attorney in the room? Sometimes it's easy to think of these steps as sequential. I urge you to think of them as holistic and collegial. Let's do whatever we can to bring people early and often into the conversation as you start. No matter how many bureaucratic settings I work in, universities or nonprofits, agencies, it's still always a surprise to me how often people don't do this, how often we plan in a small group and then circulate things later. Some people do that because it's safer. Some people do that because it's more, it seems too technical. But we'll share some links with you. I'll share some links with you now, actually. Articles about what it means like to include the community sooner.

[6:28](https://youtu.be/2GRiPqYm7DM?t=388) I have to... And these are just a couple of examples and we have many more from Agueda and many more in the packet we'll provide about effective process. I direct you in particular to this very, very recent blog from the Open Government Partnership about their work in local municipalities, Austin, Texas. I saw this today on Instagram. Austin, Texas is engaging its citizens in AI policy planning. And look, you all have probably experienced questions of focus group and different forms of citizen consultation and the challenge of what some people call participation washing, which is we'll hold the meeting, but we're not going to change anything. One good thing about AI being so mainstream and so popular now in the public imagination is there's a little more demand and a little more consideration of what it looks like to plan with the community.

[7:40](https://youtu.be/2GRiPqYm7DM?t=460) And I urge you to investigate some of those things and some of those processes. And if you have colleagues in your own office or across the street or across the square who are working in public engagement to chat with them early and often. Either way, research does show that policies and programs developed inclusively are more resilient, they're more politically durable, and they're more effective at delivering your intended results, especially results for people who are the least often in the proverbial room where it happens. And as you'll probably hear me say a couple more times, we say we're going to do this. Actually, maybe I will leave this slide blank without the image, because that sends a, that's a compelling message too, is are there enough people in the room?

[8:42](https://youtu.be/2GRiPqYm7DM?t=522) Sorry, I'm working iteratively. You will hear it's hard, you will hear it's politically challenging, you will hear it's costly. Now more than ever, the different methods of bringing people into conversation early and in policy decisions and in design decisions, groups like Microsoft have actually a pretty robust inclusive design suite that they offer to their own engineers and developers. And let me just say another thing about inclusive design before I go on to my next thought. Sometimes inclusive design or co-creation, sometimes it works at the speed of the most resistant member of a project, the most resistant team. Often in the tech world, there are a lot of people at big tech companies who understand the importance of license to operate, the importance of getting out in front of new rollouts, whether it's to get more people using it or whether it's to get fewer people skeptical of it, to get out ahead of what some call headline jeopardy.

[9:53](https://youtu.be/2GRiPqYm7DM?t=593) Engage with your most resistant stakeholders early. The culture, whatever culture of change you face in your office or institution, that same AI lands into that same culture. AI lands in medias res in the story of how you do your work and how your colleagues do your work and how your CIO or CTO does their work and how well you or your team or your agency or your mayor or your governor or your city council does or doesn't engage with the community. And we think of AI as other. It has alterity. I highly encourage you to think of it as same. Procurement is another area where there's a rote process that's observed in the breach or observed effectively or captured by a few vendors or broad and highly transparent, depending on what the product is and who's buying it.

[11:00](https://youtu.be/2GRiPqYm7DM?t=660) But I, I encourage you to look at whatever the challenges you face in procurement of technology and a procurement of anything and bring that same perspective to procurement of technology. Procurement is a governance issue. And I thought about framing this whole conversation around governance, but I think that governance is too wonky a word. And I think there's a lot that you can read about technology and governance and technology policy and governance. Furthermore, about governance, it's a shifting landscape. We don't know a month from now in the U.S. what the AI governance framework will be at the federal level. We don't know how in contention it will or won't be with initiatives that are already launching at the state or municipal level. There are some cities that are very far ahead like New York or Austin, but there are some states like California. So it's like any place you put the finger on the governance and technology question, it won't be there long.

[12:04](https://youtu.be/2GRiPqYm7DM?t=724) So that's one of the many reasons I didn't want to make governance a theme, but governance applies very deeply, particularly at the procurement issue, because the public purse is involved and there are rules around procurement that pre-exist AI. We often think we have to change all those rules or rethink them. I urge you to start with, oh, all those rules and how they're observed or observed in the breach, they apply here. Let's get out early in front of that and not let technology be a mystifier. Here are some further links I want to share about this and I want to direct you in particular to the, sorry about this longer one. I want to direct you in particular to the middle one because this merges what it takes to bring the public into policy conversations and what it takes to expose the public to your AI decisions, which can feel scary and risky, but I highly recommend you to look into that and to find ways.

[13:15](https://youtu.be/2GRiPqYm7DM?t=795) Agueda, I think can share this last link about public AI buying. If you're not already bought in on the outsized influence of procurement in government technology decisions, think about a new tool, like a new house you buy only to later discover a weak sewage pipe running under it or a sacred burial ground, like in the movie Poltergeist, or think of your tech vendor, like a money manager who you later discover is under federal investigation for something. The decisions we make about who manages our data, which jurisdictions they are beholden to in their own work, in their contracts with us, about what those decisions, they change our tools, they change what the tools can do, they change our nimbleness and they change what risks we pass along to our colleagues and our communities in the contracts we're signing. Procurement is a governance issue as well as a consequences issue and sits in a value chain along this sequence where, just to reemphasize the sequence again, what we expect, who we include, how we purchase what we do, and then next, how we do what we do.

[14:38](https://youtu.be/2GRiPqYm7DM?t=878) What does it look like to be ready to implement? Who stays in the room and who leaves the room only to come back later and say, why did you do it this way? Who do we need to keep in the room when we go from the plan and the contract to the building of a tool and the launch of a tool? When I worked at the New York Times, we were trying, this was back When online conversation and online community existed, some of us will remember this, some of us won't, in a button up in the upper right corner of your website that said community or discuss. And it took a lot of arguing and advocacy with my New York Times colleagues to get them to put a button that says comment or reply in the article. This is the Stone Age, of course, when it comes to online interactivity, but it was before Facebook, it was before blogging. And it was when if you wanted to get someone to do something different than they already did, they're reading, you had to really show them why.

[15:38](https://youtu.be/2GRiPqYm7DM?t=938) And I mentioned this question of progressive disclosure and user experience, because what we tell people is going to happen is what will happen. And what do I mean by this? I mean, if we're launching a bot to help people find a pharmacy or to help people see what their housing benefits might be or different ways to get a card or an app about to purchase food stamps. If we say, it's finally here, our new app, and you have a picture of a robot with a grocery cart, that's a very different expectation to set than maybe a picture of someone at the grocery store, maybe a video of a mayor or a city council person walking in with their phone to the grocery store. Expectations determine how people will use the thing. So this is an argument against shiny. So we're not ready to implement isn't the same as implementing and implementing has within it all these sub-steps that again, I encourage people to continue to think about what expectations are we working with?

[16:51](https://youtu.be/2GRiPqYm7DM?t=1011) Who is or isn't in the room? Did we make the right choices with our purchases? Are we governing our purchase as well? How are we going from ready to action and then to launch? It's a logistical issue and it's also a cultural issue. And I don't just mean the culture of whether your grocery app is in Khmer and Spanish and Vietnamese I mean, the culture of your developers and your visual designers versus the culture of your risk management people or your attorneys or your marketing people. Here's a series from Innovate US that is working on this and the digital government hub. Oh, I can click it right in the screen. The digital government hub is out of I think Brown University and you'll find you can find a few different really cool repositories of all the different policies and policies and AI laws, policies, regulations coming online in different principalities.

[18:04](https://youtu.be/2GRiPqYm7DM?t=1084) Very hard to keep up. Like I said, you put your finger on it and it's already moved. But this is one of the ones that I really like and I will share the other links from Innovate US. In my experience as a digital director and more recently as a digital strategist with multiple clients, what we plan to do is rarely what we end up doing. My friend, Rich Harwood, who works a lot with cities and state innovators, Rich likes to say our plan is something that we keep in a drawer. So it's okay to take any finalized technology plan with a few grains of salt. With a few grains of salt, things change. But hopefully the principles we're talking about here can work like those reflective signs on the highway that make unexpected turns safer or like a locker room that your teammates and you can go back into when you need time to plan a next move.

[19:05](https://youtu.be/2GRiPqYm7DM?t=1145) Just because we've launched, just because something is facing the public doesn't mean that's the great thing about technology. You can always iterate. You can always iterate. I know that bureaucracy doesn't like quick change. Technology moves fast. The human mind moves pretty fast. We're pretty adaptable as a species but institutions move slowest of all. But that doesn't mean, the more you set the expectation that something may need to change, the more it might be easy to unlock some money or some developer time to make a change that comes up later. And this nimbleness, this readiness for the unexpected is fundamental in technology in particular, particularly with such high beam scrutiny to continue my automotive metaphor on how people are launching technology, what the risks are and even how the rules change from month to month, given where we're at. And if that wasn't uncomfortable enough to think about, I also wanna talk about how we assess the tools that we've launched.

[20:09](https://youtu.be/2GRiPqYm7DM?t=1209) Did you have an assessment plan from the beginning in those lists of rules and policies being launched at state and city level around the country? You'll see many have a mandated technology plan that include mandated audits, mandated procurement rules. And those can be confining and constrictive and points of political contention but they're also just fundamentally important because if we don't know, it's just like measuring the water downstream from a power plant or the job data when you have an employment program, if you don't measure, you don't know what the impact is. So be mindful of what you're assessing, plan to measure and in your measurement plan for change. Assessment like procurement is a governance issue too. It's a place where you can build your infrastructure rules in early and thoughtfully and inclusively to act as a guardrail or a road sign to yourselves.

[21:15](https://youtu.be/2GRiPqYm7DM?t=1275) And it often gets neglected. It often chases after the rest of the project, as you know. Here are some examples. And I now know that I can click into my slide, which is great. Oh, some of them I can, some of them I can't. I'm not sure. Sorry. Okay, I'm gonna click this. I'll share this one in the chat. This is an example I really like from a couple of years ago where New York City had a very advanced rule about if employers are using AI, they have to report it and then they have to audit it and they have to file the audit with the regulator. As this story tells, the number of people in this example of local law 144 in New York who complied, completed the audit and delivered the results of their AI implementation was woefully low.

[22:24](https://youtu.be/2GRiPqYm7DM?t=1344) I think out of 300 possible employers who were under the law at that moment and two or three years ago, fewer than 15 actually filed the required report. And that's not simply scoff laws or people who are trying to get away with something or who don't wanna be transparent. Often what it takes to do an AI audit is as challenges of what it takes to implement AI. Who are you gonna hire? What contract will govern how you're working with your auditor? Is there an internal auditor and is the internal auditor well enough trained to have the tools they need to audit or do they need an AI expert? Who's in the room? So again, lather, rinse, repeat. We bake in the same processes almost in a spiral fashion at each stage of the AI development and implementation cycle. We'll make sure to share these other links if Agueda has not already.

[23:26](https://youtu.be/2GRiPqYm7DM?t=1406) And you'll also see that the rules and examples, the rules and the standards for AI assessment are shifting and in Europe and in some of the announcements the White House has made here in the US, there are some ways in which assessment is becoming more permissive and less rigorous. And if speaking not from my NGO side, but as an agnostically, there can be pluses and minuses to freeing up innovation versus adding regulation. I would say though, that speaking from civil society, a minimum amount of evaluation is necessary not just for compliance purposes, but for human rights purposes. So assessment and evaluation, they can be like the vegetables no one wants to eat and too often they can become an exercise in mere compliance, which is a race to the bottom. The example of New York's local law 144 is here.

[24:28](https://youtu.be/2GRiPqYm7DM?t=1468) The World Bank has another even lighter weight example. The World Bank measured once a few years, about 10 years ago, how many people were using the reports that the World Bank issues. And out of thousands and thousands of report about seven from over a five-year period, the World Bank found that 75% of their own reports had been downloaded under a hundred times and half of those had never been downloaded once. So measurement helps you know where to spend your resources and time. And if you're looking at what has happened after you measure and you find yourself saying, well, at least we did that, at least we launched on time or at least the thing doesn't throw that error that it was throwing during the beta phase, or at least we didn't give anyone botulism. That means you're leaving a lot of insight on use.

[25:28](https://youtu.be/2GRiPqYm7DM?t=1528) So be comfortable making an assessment and be comfortable trying to weave the results into the work. Whether it's AI or a website or an app, we can't design for launch. We have to design for use. We have to design for use and outcomes. So remember, if you're planning to make change, that's when you've closed the loop on measuring what you do. A study I really like by the academic Mona Sloan and her colleagues talk about how, even after we audit our AI work, the socioeconomic dynamics, what data and society and some others call socio-technical factors are absolutely fundamental to why a policy works or doesn't work. AI, like any policy, like any change, like any new hire, lands into the environment that pre-existed it.

[26:29](https://youtu.be/2GRiPqYm7DM?t=1589) So as Mona and her colleagues wrote, it's important to consider political economy and how an assessment or an audit and how an assessment or an audit can, or I would say can't, cue in more holistic policy mandates that address, in this case, inequality or ecological impacts, but also address efficiency or address cost savings. And if you conduct an assessment and find yourself lacking the budget or the permission to use findings, use those findings to fix the problem. Say you're like, oh, I heard what Jed said and I was already there and he was preaching to the choir. I love assessment and we love doing retros and looking at what happened, but then we just don't have the money always to implement it or there are some people who don't feel good about, don't feel good about what happened, don't feel good about making change, then you're gonna need to learn to manage up. And that's another key part of this work, working in the team you're already in with the people who are already with you day to day.

[27:33](https://youtu.be/2GRiPqYm7DM?t=1653) I mentioned about governance. Things that underlie all of these steps are the culture that you work in, the design process and how that process of how you did something affects how it's used, if it's used at all. Like in my example, where you promote something as shiny and new, I was part of a civic team once where, I heard a demonstration once from a civic team who had built a housing app for, a municipality in California, maybe Alameda County. And they were demonstrating their new website to understand your housing benefits. And they were showing the residents different designs. They were bringing the community into the design process and the design that they all liked the best from the web shop was lovely and beautiful. And the feedback they got, the community, the residents that they had with them chose a different design. And they said, oh yeah, we saw that one. We saw the shinier one. It just, it looked like it wasn't for us.

[28:36](https://youtu.be/2GRiPqYm7DM?t=1716) So are you designing culturally toward use? Are you not designing for launch? Not designing for demo? Not designing for accolades, but designing for use? And then finally governance. And if governance can mean, what are the different rules that pertain? Governance can mean what are the different attitudes that pertain? And governance honestly can mean, are we doing what we want to? Are we doing what we ought to? Are we doing what we must? Are we doing what would be nice? And my pitch is always to think about this, not to necessarily land in one place or another in terms of strictness or compliance. That's up to you and your compliance team, but to be aware, to be conscious. And I normally, as you're maybe hearing from some of my stumbles as I talk, I normally talk much more about culture and design and governance, but in this framework of expectations, inclusion, acquisition, implementation, assessment, and adaptation, I'm trying to provide something that is more practical and easier to take away with you.

[29:52](https://youtu.be/2GRiPqYm7DM?t=1792) And as Agueda and I have said a few times, we will make sure that it's even easier to take away when we share these slides around. Here are some caveats and reminders that I won't read aloud. But my hope is that by distilling this adoption process down to simple considerations, we can demystify the new tools. We can take it out of analogies like the Terminator or the magic robot that will clean up my email box for me. The Roomba that goes around the room. We're past that point now. If we think of AI and innovation as landing in an environment that already exists, an environment of expectations, in an environment of easy or hard acquisition, easy or hard procurement, easy or hard implementation, you will be better equipped and maybe even in a better state of mind to tackle some of these questions and move quickly and nimbly.

[30:58](https://youtu.be/2GRiPqYm7DM?t=1858) That is my final slide, and I know we're going to have time for some questions. And while we're getting ready for that, I'm going to stop sharing my screen for a moment and look back at something I wanted to say that I deleted. So if you just give me a moment here.

**Agueda Quiroga**

[31:24](https://youtu.be/2GRiPqYm7DM?t=1884) Sure, we can address the questions that we have already in the Q&A, but we can give you one moment.

**Jed Miller**

[31:31](https://youtu.be/2GRiPqYm7DM?t=1891) If I could ask you to queue up a question.

**Agueda Quiroga**

[31:33](https://youtu.be/2GRiPqYm7DM?t=1893) I will, I will do, I will do.

**Jed Miller**

[31:35](https://youtu.be/2GRiPqYm7DM?t=1895) Then I will go back. Thank you.

**Agueda Quiroga**

[31:38](https://youtu.be/2GRiPqYm7DM?t=1898) Sure, no worries. I will take care of the Q&A.

**Jed Miller**

[31:42](https://youtu.be/2GRiPqYm7DM?t=1902) Okay, so what are some of our first questions?

**Agueda Quiroga**

[31:49](https://youtu.be/2GRiPqYm7DM?t=1909) Okay, I have a question from Melissa that says, how can we add the state level drive change? We are a tadpole in a big pond.

**Jed Miller**

[32:00](https://youtu.be/2GRiPqYm7DM?t=1920) Thank you, Julie. That's the question.

**Agueda Quiroga**

[32:03](https://youtu.be/2GRiPqYm7DM?t=1923) Yeah. How to drive change at the state level?

**Jed Miller**

[32:10](https://youtu.be/2GRiPqYm7DM?t=1930) By we, by we, Melissa, I assume you mean the team that you're working in. And if I have that right, I would strongly recommend that we add the state level drive change. And if I have that right, I would strongly recommend self-promotion. The early, easy, sometimes the earliest, easiest win, you know, it didn't take a lot to do. But if it gets pass around, this is just sort of bureaucratic hack that I've used in the past. Do something shiny and early and simple. You know it's not really what you're aiming for, not really the goal that you're in. Yes, the team I am with. Thanks. But if you're doing something that some people not in your team or not in your agency might like, or that the media might like, or that you can score a win with that's measurable by certain people's perspective, economically or usage wise, or in an outcome that you know is a priority, do that even if what you're trying to do, even if what you're really trying to accomplish is step 10, and you've only done step one or step two.

[33:28](https://youtu.be/2GRiPqYm7DM?t=2008) I had an example of a big institution that was very policy heavy and legal heavy that didn't know how to start blogging in a more timely fashion as they were tracking debates in Congress. And I found one person who the head of the organization trusted not to be reviewed, because she was herself a lawyer and I was a very good editor and we were working very fast. So we started blogging about a legislative process in a very timely way. And it wasn't the same as acquiring a blog platform or launching a blog strategy, but the minute those blogs went up and it was on a very timely topic that was hotly debated in the news, everyone said, oh, we want to blog too. So sort of, you know, attention hacking, I guess I would call that.

**Agueda Quiroga**

[34:17](https://youtu.be/2GRiPqYm7DM?t=2057) Okay. Thank you so much. I have more questions. So I will read another one and say, how can agencies ensure they are hearing from the right voices before making decisions about AI?

**Jed Miller**

[34:30](https://youtu.be/2GRiPqYm7DM?t=2070) I think that we've talked about that and there are going to be some good other resources we can share about that. I'm going to share my favorite, one of my favorites here regarding the processes that can be used to consult with the team. But I would say, if I can flip the question, who needs to hear from whom? You probably know who is missing. And then if you know a little bit about why they're missing, it's almost like matchmaking. So we have some voices we know are absent and we have some people who we know need to hear it. Start on the demand side. How can I persuade? What can I put in front of that constituency in order to make this constituency more in demand by my listeners, by my decision makers?

[35:37](https://youtu.be/2GRiPqYm7DM?t=2137) Usually it's the decision makers as distinct from the users or the people who are traditionally underserved in a community. So one thing is to create demand by engaging with the people who you know need to be heard. Another is, this is an even simpler example, if you're holding a community meeting about a school issue, but the parents of the children who are bused into the school live a car ride away and they're using their car for their second job at night, you have not opened your window wide enough to get the right people into the room. So that is, again, a place where the habits of what consultation look like might not be suited to the people you most need to hear from.

**Agueda Quiroga**

[36:27](https://youtu.be/2GRiPqYm7DM?t=2187) Thank you so much. Great example. A question and comment by Greg. This has been really helpful. Building on the adaptation slide and the culture theme, could you please elaborate on implementation feedback loop practices, rather than or in addition to audits, that helps move from early adopters to iteration and ideally favour durability of the planet change?

**Jed Miller**

[36:58](https://youtu.be/2GRiPqYm7DM?t=2218) What a great question. Well, did you design for your early adopters, or did you design for the lowest rung of the ladder or the least familiar user? I want to try to answer in two ways. The first about going for the least likely adopters sooner. Oh, I know this questioner, and I know him to often ask very good questions. Thank you, Greg. So one answer.

[37:58](https://youtu.be/2GRiPqYm7DM?t=2278) Okay, so you're giving an example where the process or the tool predates you. Well, one example I would give is, just because it's technology doesn't mean that you can get your adopters, your non-early adopters in via technology. You might need to be posting pamphlets. You might need to be posting something on a bulletin board at a community centre they go into. And maybe it has a picture of the app, not just a QR code, but maybe a picture of a phone with the app on it. Always lowering the threshold of expectation. I know what I will see. I know what's coming before I download it. I know what's coming before I click it. You know, UX doesn't begin when you turn on your screen. UX begins, this is like my example from the New York Times. We were trying to get people to have conversation about news, and there was a very famous conversation website called The Fray. Well, if you call a conversation site The Fray, what do you think is going to happen?

[38:58](https://youtu.be/2GRiPqYm7DM?t=2338) It's different than calling it Davos or calling it Porch. All of these are design decisions. As to feedback loops, I would say one thing to think about is sort of the, I forget the term for this. This is the social hack where you get your early adopters to refer other people. Ask your early adopters who they know who doesn't know how to use it. Ask your early adopters to propose other people and to email them with a template email that you give them. Leverage the social capital of your early adopters. I would say the same thing about those people over here, the demand side, the skeptics. The people are like, why do we need to change it? We already did it. Leverage their social capital too. If you're sure it doesn't need to change, can you recommend some people who are unfamiliar? Can you recommend someone less tech savvy or a list of non-tech savvy people we have from somewhere nearby in the project who we can bring into it?

[39:58](https://youtu.be/2GRiPqYm7DM?t=2398) And then if the people that they themselves suggest we look at have problems, then you've got a little more leverage. So what have I said? I've said referral, social capital, and starting at the lowest tech place possible in every instance. Great.

**Agueda Quiroga**

[40:16](https://youtu.be/2GRiPqYm7DM?t=2416) I think we have some time for a few more questions. So I'm going to read one from Melissa. Let's see. How can we access the foundations for these programs if the money isn't there but there's still push for change? What can we do?

**Jed Miller**

[40:35](https://youtu.be/2GRiPqYm7DM?t=2435) I think and tell me if I'm wrong. I think we're talking again about the demand side. Yeah. How do we? Oh, I have found the Q&A, I think. How do we access the foundations if the money isn't there but still push for change? Always start somewhere where you think about the lowest threshold for minimum change for maximum proof, right? What's something you can do that puts you on the path toward greater adoption or greater interest or greater demand? Another thing I would recommend, everything moves at the speed of the most resistant person.

[41:37](https://youtu.be/2GRiPqYm7DM?t=2497) So if the most resistant person said there's not budget, see what kind of low-cost experiments you can do. And another thing I should – we've talked about social capital. There's also competition. Is there an equivalent in – thank you, Bobby. Is there an equivalent agents? Do you have a peer who's had more luck moving past the foundations into action and can you leverage their successes? Like I was talking about, about my blogging example where the other teammates, the other departments saw this blog from our – what at that time was our lobbying department. And they said, oh, well, if the lobbyists can do it, then we should be able to do it when we're lobbying about something else. Or then we should be able to do it in our field team who's working down in the communities. We could blog too. Thanks, Mauricio. I hope that helps. I think it's telling and interesting. I hope to do a future module on implementation specifically because all of these same principles apply down at the implementation stage, within the implementation stage too.

[42:45](https://youtu.be/2GRiPqYm7DM?t=2565) Thanks, Hannah. So I would say you can leverage competition in that way. Find your counterparts who have had a little more success.

**Agueda Quiroga**

[43:00](https://youtu.be/2GRiPqYm7DM?t=2580) Okay. So one more question, Ted. Do you think we have time? Okay. So this is for you, yourself, as a citizen of this society. What wide rails do you wish to have put up as the race to adopt AI-enabled tech in all sectors of our lives advances quickly? Thank you. Great.

**Jed Miller**

[43:33](https://youtu.be/2GRiPqYm7DM?t=2613) I'm going to give an answer that I think is – I'm going to give two answers. I'm going to speak as an AI skeptic, which I am not, but I'm a skeptic of tech speed. I'm a skeptic of moving too fast and breaking too many things. And I think the biggest guardrail – the two biggest guardrails we can put are time and the word why. Silicon Valley is right now too trapped in a because we can. And government, I would say, might be too traditionally caught in because someone else is. We will be left behind if we don't. And that certainly can play out, I think, at the state level too, among the different states and across agencies. So one guardrail is why. Another is time.

[44:34](https://youtu.be/2GRiPqYm7DM?t=2674) Time is a good guardrail. It's up to all of us as implementers and as policy people to allow for sufficient friction that by the time we are ready to implement something, the user base has caught up a little more. The user base doesn't have that moment of awe and shock and confusion. Not only because it will reduce adoption, reduce use, reduce the innovation we're actually trying to deliver, but because when we implement or use out of awe, it's like buying a printer and not knowing how to fix it. We need to make time our friend rather than our enemy when it comes to these things. The culture of Silicon Valley and some cultures of governance favor speed too much. If you Google the India AI Summit and look at some of the things that just happened in India earlier this month, you'll see a lot of great discourse on the tension between government speed to innovate and government capture of innovation and government desire to serve and civil society's desire to partner with government for best service with least harm.

**Agueda Quiroga**

[45:57](https://youtu.be/2GRiPqYm7DM?t=2757) Thank you so much, Jed. I think we now need to close the session. I just want to do just a very brief housekeeping because there were some questions into the chat about this. So when you close the session, the post-workshop survey will pop up. So just give 10 seconds to appear so you can answer that there. And we are sending tomorrow Jed's deck, slides and links in the post-workshop email. So I think that's it. Jed, any other final wrap-up comment, sentence, something you want people to take with after the session?

**Jed Miller**

[46:38](https://youtu.be/2GRiPqYm7DM?t=2798) I'm very grateful to Christine and to you and to Agueda. I really appreciate the questions. I really appreciate everyone's patience with my unfamiliarity with certain of the dynamics here. And in the deck, you will have my contact information and I welcome further questions and we will organize the deck such that it has a bit more depth and a bit more usability as well.

**Agueda Quiroga**

[47:01](https://youtu.be/2GRiPqYm7DM?t=2821) We definitely need to get you back for another session. Definitely. Thank you so much. No, thank you. No, that's, I mean, that we need to schedule that soon. Thank you so much. And we hope to see you soon. And thank you so much to our participants. I am going to close the session now. Thank you.

**Jed Miller**

[47:16](https://youtu.be/2GRiPqYm7DM?t=2836) Thank you everyone.

* * *

Original page: https://innovate-us.org/dont-think-of-a-robot · Licence: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/), as stated on innovate-us.org
