Introduction to AI Strategy Development: Strategic Planning + Technology Assistance

Today we're going to be going into AI strategy development. Basically, how can you do your strategic planning and bring your robot friends into the mix? Because in the nonprofit sector, we need to use social technical systems as our approach to strategic planning that recognizes the interaction between people and technology. That's why we're in civil society because we're people centered. Understanding how your team interacts with AI and tech is extremely important. And to do strategic planning correctly, we must now include how AI can assist us in completing our work because let's be honest, there aren't enough resources. There's not enough time. We could use some help, but also we know that there's risk if we get things wrong. And so to dive into this conversation to help us navigate, we are really excited here at TechUt to bring in a guest presenter. Um Tina Krauss works in AI ethics and strategy and is a responsible AI practitioner who has been delivering webinars and in-person presentations on AI since 201. She's also the author of several articles about AI and nonprofits published by Charity Village and facilitates courses and empowering seniors in understanding and using AI. She's also an AI practice coach, which means you've got an expert here with us, someone who is not like a bandwagon jumper on AI, but actually has been exploring this field for quite a while. And with that, Tina, over to you. >> Thanks so much, Eli. That is such a terrific introduction and and it is really nice to get thunderous applause to start uh a presentation. So welcome everyone to introduction to AI strategy development. If you do have any difficulties hearing me seeing slides, please uh use the chatbot let chat box let Eli know and have a lot of slides to get through. I'm going to really boot through this and as Eli was saying, don't be concerned uh because you're going to get the slides, you're going to get the presentation, and you do have the chance to ask questions today and any other time because I'll give you my contact details. So, I am Tina Krauss of Thornberry Communications. As Eli said, I work in AI ethics and strategy. I actually started studying AI in 2020 um and began doing presentations and things to students women under represented groups about responsible AI in uh 2020 2022 actually. If anyone wants to contact me my socials AI practice coach at Facebook and um Instagram and also you can contact me through my send me a text, send me an email, check me out on LinkedIn. I post a great deal there. So, just as we are starting on, I want to tell you about my mission because we're going to get into yours, but I want you to tell mine, tell you mine. I want to help the nonprofit sector feel better and do better through AI adoption. I know that we can. I know the approaches that I'm going to talk about, responsible AI and the social technical approach are things that resonate with the sector with our values. And so that's my mission. I'm here to help. I want to make things better and I know that I can. So let's move on to the agenda. We're going to talk about aligning mission and vision, ethics and governance, proworker AI, and then get to the AI strategy roadmap. So most important thing is that we're going to ensure that AI and your technology supports your strategy, supports your work. That's what it's there for. And when we use it well, we're going to have the outcomes that we want, which are going to be good. So, definitions very quick. Social technical system. There's a lot of academia on social technical system, but I've collapsed the definition, the interaction between humans and technology. So, you get into a self-driving car, hands off the wheel, you don't do anything. Get into your own car, use technology to help you drive. Social technical system. That's what the second one is. responsible AI. So this um definition comes from IBM actually created responsible AI in 2014 more than 12 years ago. Responsible AI aims to embed such ethical principles into AI applications and workflows to mitigate risks and negative outcomes associated with the use of AI while maximizing positive outcomes. Does this not sit well with people? Is this not what we need in the nonprofit sector? We're going to mitigate the risks and the negative outcomes, but we're going to get the benefits. That's why I'm bringing responsible AI to Canada's nonprofit sector. So, these are the six principles of responsible AI. Accountability, inclusiveness, reliability and safety, fairness, transparency, privacy, and security. Is there anybody on here who would argue with these? I actually think these are completely in line with the values that we already hold in the nonprofit sector to hold people accountable including ourselves. Inclusiveness, we're looking for bias. We're trying to make sure that there is much much more representation for people. Uh reliability and safety. Want to do things accurately. Want to protect our clients when need be. Fairness. If we're working on social justice, it's because we want more fairness in the world. And believe it or not, responsible AI can help us do it. Transparency, privacy, security, those are our things. We already live by most of these as our values. So I think responsible AI is pretty good to consume into our work. Okay. So these are some of the responsible AI trailblazers. Yep. These people have been you know doing exceptional work. Kathy O'Neal she wrote weapons of math destruction such a good title in 2016 about algorithms enforcing inequality right more than 10 years ago. Shashana Zubath she's the person who've coined the term surveillance capitalism which we know we're all subjected to now by how people check our not people companies check our social media they watch what we're doing online they increase prices due to who we are where we are where we sit surveillance capitalism yeah we're in 10 years old Joy Bulamini she was the first person to discover that facial recognition did not accurately ely recognize people with black or brown skin. Lita Aferani Hani, she is a professor and she works on technology on society, the effects of technology on society, law and ethics. And Rman Chowry, she was someone who is very very deeply involved in the right to repair movement. if you remember about shouldn't we get to repair our toaster and our and our kettles rather than throwing them into into junk piles and creating environmental problems. She's also uh when she was working at the Turig Foundation, she was also one of the people who worked on a fairness tool where they would determine if u biased data affected the outcome of an algorithm. Lots of developers like to say that algorithms are agnostic. Absolutely not true. She was one of the first people who proved it. Then we have Abida Abiva Bhani and Aji Tang. Abiva is so well known. So she studied how AI disproportionately affected vulnerable groups. She had a PhD in this. And then she went on to establish the AI accountability lab at Triny College in Dublin. And they're able to go into AI models and find the bias and inaccuracies and pull them out and tell developers how to fix them. That's what she and her team do. Audrey Tang. They are Taiwan's cyber minister for a few years now and is the person with the most eloquent speech philosophy philos philosophical commentary on social technical interaction. These people are my my sheer my heroes absolutely of responsible AI. And I put links to all of their Wikipedias at the end of the presentation. So when you receive it, if you want to check them out, any of them out, I really recommend you do, you will have the opportunity to do so. Okay. So let's get into ethics and governance. Your decisions, your choices, your impact. If you're going to use AI, you want to use responsible AI, then you need to create policies. Now, lots of people have been talking about AI policies very difficult, blah blah blah. Um there's actually the BIATI Foundation in Montreal has released open source a set of AI governance policy work. It's a great guide. They released it in French and English. It's open source. It's available to everyone, every nonprofit. So don't reinvent the wheel. Take their work and edit it to fit your organization. You want to make AI adoption desirable, right? Things that are people going to use, things that your staff are going to benefit from, your clients are going to see some positive gains. Make sure your policy work is about that privacy actions. Okay, we know we have to protect the privacy of our clients. We know that. Do we know how our processes can potentially release private information? Do we know how to recognize those problems? It's something that we need to do with our within our ethics and governance. And then data containers. Data containers sounds like a very jargony term. I wrote an article about this for nonprofits and it's also at the end of the presentation. It really is how do we receive people's data and where do we put it? And then finally, how do we protect it? You got to be in the driver's seat. The article lays out the steps for you. Trust me, it's not that difficult, but it is something that you want to do. So, social justice needs good decision- making. This is the world that we're in. Canada's nonprofit sector believes and upholds social justice and so we need good decision making even while we're working with AI to do that. So these quotes by Cassie Kazakov, she's Google's first chief decision scientist and another person to whose work you're going to find really fabulous if you look into responsible AI. Um so Cassie talks about AI doesn't have judgment, values or accountability. None. None whatsoever. It's technology. Do you talk to your word document and ask it to make judgments for you? Absolutely not. Don't do it with AI, right? It can't actually. You're going to expect errors and you can plan from them. That's how it is to use. It's not that difficult to use a two-prong system. The question you want to ask and then you ask AI proof that pro that the answer is accurate. Yep. Two prompts. Keep responsibility where it belongs with humans. If you heard that phrase humans in the loop, this is what we mean. Keeping the responsibility with us for the outcomes, the outputs of the AI. It's something we have to take care of. AI is going to mirror your input, not replace your judgment. Never. Absolutely not. Even if you're using agents, this is not the place to give over your human judgment. It's something that separates you from AI. Um, and so you're going to want to bring that into this work within AI strategy. Okay, this is anthropic's four Ds of ethics and usage. I think these are very good. I don't need to go into them because you're going to receive this check. But when you take um when you look like you're going to or sorry when you're going to go ahead and do your policy work, use this 4 days framework. It will help you decide with what you want to do, how you want to do it, who to include. This is very important work. So usage and responsibility. The first thing about improving your AI is to have diverse representation on your teams. If you don't, perhaps you're a small organization, you don't have a lot of diversity within your team, then include your board, include clients, get an advisory board because diverse perspectives are the absolute best way to ensure better outcomes using AI. the generative AI that we had access to Jack GPT Gemini Claude they weren't developed like this even if anthropic said that claude was it wasn't so when you are going to use the AI you need to have clear guidelines that's what your policies are right you're going to have boundaries about how you're going to use that AI and you're going to hold space to discuss this you're not just going to prohibit people from doing it when you do that, people do it anyway, and then you really have problems. Shadow AI is huge concern for businesses. It's even bigger for nonprofits because of what I said about client data. We as a sector have all the same data as the federal government. We are eight times the size of the federal government and they are spending millions on data privacy. We can't keep up with that. So we have to do what we can and that starts with our guidelines robust testing and validation. If you're going to bring AI inhouse, whether it be genative AI or sector specific AI, you got to research it. Make sure you understand the problem you're trying to solve. Many businesses did not do that in the last two, three years, and they have lost billions of dollars in not understanding what they needed to solve with the AI. You're also going to check about training time. Are you going to bring in new software? You need to train your staff. Do you have time? How you going to do it? When are you going to do it? And test how it integrates with your systems. Some AI systems are going to have to redo all your processes and it will be for the better. Trust me, you will appreciate that happening. But other things, let's say grant writing software, it's not necessary. Very limited time use. Sometimes only by one person or only a few. You don't have to redo all your processes for it. Channels for reported concerns. I'm going to say this repeatedly. People need to be able to talk about their experiences. You have to try to figure out was there a problem caused by AI or something else? What process are we using? Who's involved in it? And you want to set tasks for monitoring looking for bias, errors, any kind of unwanted actions. You want to hear about them as soon as possible. So then the last one, maintaining data privacy. You're going to build your data container. Then you're going to monitor how that data is being held. And if you have any First Nations clients, then you're going to want to use the First Nations, which is ownership, control, access, and protection policies for indigenous data. But honestly, indigenous data sovereignty, they have been first nations people in Canada have been working on this for more than a decade. They're an excellent resource to learn from if you ask them if you could take training or you could learn from them. But definitely have some understanding of OKAP if you are serving any First Nations people. Okay. Dealing with bias. Of course, you all want to talk about this one, right? Because it's a big deal. It's the thing that scares us the most. So we know that bias is pernicious, right? It exists in many forms. Racism, aism, sexism, gender bias, language, culture, economic bias, like in medical scribes. That's crazy. To deal with bias, we have to take time. We have to commit to it. Got to be willing to seek it and change it. Even when we find it, we learn ourselves. And that is such a tough one. I have stories about that trying to address it with the I had a kept forgood social enterprise couple of years ago. It pointed out bias. I was writing an algorithm to have unbiased grant making software. We succeeded but government foundations people did not want to hear that they that their meccas or processes were biased. So if you don't want to face that problem, if that's not your top thing, then you just don't do it right. But now we have AI and it brings out bias more frequently than any of us would like. And so we know we have to look for it, seek it out, and figure out if it's in any of our systems, not just AI. Have policies that deal with it when we find it. Go back to those reporting channels. days have to exist so that the public, donors, your staff, your clients, you can talk about seeing or experiencing bias. Want to deal with it in your organization. So, of course, we want to empower our clients to raise issues from their experience. That's what helps us make better service delivery. And then we want to hold ourselves accountable to change it. The tough part, the very very very tough part, holding ourselves accountable. Hard when we find it, harder to know it exists within ourselves and then committing to change. But we have to do that. This quote from Shri Met Ray, ethics is the map that helps artificial intelligence make decisions that are good for people and the world. It's why we have to have them and start with the basis of them. Okay. How about impacting the environment? This is such a big concern for all of us because AI is very big environmental has a very big footprint on the environment. So there's a couple things to be aware of. This is not polyianish. This is very true. Governments and corporations are working to reduce energy impacts. Governments. Why? Because we have environmental laws. A lot of data centers cross those laws and so the government's being unlawful makes everything very complicated. They don't want to be there. They're working on it. Corporations with all the corporate greed, environmental energy costs are very high and they don't like it. So from that standpoint of corporate greed, they're going to lower those costs, lower those impacts as much as they can. Okay, that's not my approach. I want to help save the planet. But if that's how government and corporations, that's what their motivation is. And mine's going to be reducing my own impacts. And so we're going to come together and we're going to reduce the environmental problems that we have. Okay, great. Work from wherever you are. I want to get to that point where we are reducing our energy consumption and our impacts while we're using AI. Just before we jump over to the personal use though, I really want people to understand that there are ENGOs, environmental nonprofit groups and they are using AI to help save the planet. They are because it can. We just people need people to understand how it works to gain those benefits. So Pano AI, they work on wildfires. God, do we not need help with that? the last couple years, even the last month here in Ottawa, we experienced that that yellow day that was horrible, right? So, over to in your work and your personal use, you want to reduce your energy usage for using AI, think before using a prompt. Really think about it, okay? Because if you're lazy about it and you just put in a general question, you've seen perhaps you've seen Chad TPT or Gemini will spit out an entire page, paragraphs and paragraphs on something you don't want because you were so general it just doesn't hit the mark. Then you have to keep asking it questions. Okay? So don't do that. Think before using the prompt. Be as specific as you can. Right? because then when the question is better defined, you're more likely to get an accurate response. And tell Jai to be brief. Like I was just saying, cla ch just spit stuff out. It's so energy intense and it's completely unnecessary. When you use your prompt, you say, "Make the answer no longer than five sentences. Only give me a paragraph of information. um give me five bullet points and state it and state the information as briefly as possible. All of that reduces the energy consumption that AI uses just those things. Have some familiarity with the subject matter. Businesses over the last couple years have lost billions of dollars because they didn't have people on staff. They didn't have expertise to be able to understand the the outputs, right? They didn't understand how to actually integrate it into their workflows and things because they needed other people working in those businesses. Business owners were actually pushing because AI can do many things, pushing above their own competency levels in hopes of a productivity gain or some more profitability. and they've lost billions of dollars over the last couple of years doing that. It's a good lesson for the nonprofit sector. Use models with less environmental impact. Okay, we can use deepseek. People think no deepseek it has all kinds of privacy violations. Yeah, well the first 18 months so did chat GPT. People found their work conversations on the internet that wasn't deepseek. What's chat GPT? So if you're going to use something like Deep Seek, you're going to anonymize your requests. So it doesn't know who you are, where you are, and I have to ask you really in using General Genai, why does it need to gender? It doesn't need to know I'm a woman. Doesn't need to know my age, doesn't need to know where I live. Absolutely doesn't need to know any of those things. So don't put it in there. Right? That's the thinking before the prompt. If you don't want to use deepseek AI, it's on the resource list. Apparent AI is ethical AI comes out of Switzerland. It's free for everybody. They use hydroelectric power and so they have a smaller energy footprint because they're using renewables. It's ethical AI because it's in line with the EU's GDPR legislation, which is general data protection regulations. There's ethical from that standpoint. If you ask it to do something shady, it will just say no, it won't do it for you. It has no syncopancy. Soy is that word where the NA the the genis are trying to get you to stay on the platform and it says as as my son says, suck up things. Oh, you're so smart. What a great question. And you know how brilliant of you to want to ask me about that. You don't need the synchro fancy and it's very hard to turn it off in the American Genai software because it manipulates human beings to to stay on the platforms to give more information to pay more. But apparistency I think it's great I use it all the time perplexity for research perplexity the AI very small amount of it as well. It's much more comfortable for me to get answers about questions rather than have somebody some something not somebody something chatting me up sometimes feels like being in a bar. It's ridiculous and I don't like it at all. Okay. Don't use voice. It's a nice feature. You know, it's fun. It's easy but it's really energy intensive because it's taking your words. It's understanding your your d your dialogue, your accent, whatever it's going to be. It's converting it into data into the algorithm. It's going and getting an answer and then it translate that answer back into human speech and then talks to you. It's a nice feature, but leave it for people who need it. There are people with disabilities, people find motor problems who can't type. Leave voice for them, right? And you just type your question. Okay? Even when people progress from these kinds of question prompts and they go into workflows and even later into agents, very good idea to be mindful of the environmental impact. Consider alternatives to image creation. Okay, this is a very big one for everybody, right? We know corporate AI, they did web crawling, they stole artists, everybody's images. Okay. Do we want to still be part of that? If not, try to use alternatives. I use Unsplash. You've seen it probably on some of my images here. You can pay a small monthly fee for Getty Images. Getty Ge Ty. You could trade with each other and use images in the Commons that we would all agree to do. Just consider it. It uses a lot of environmental energy to create images and it doesn't do it very well as much as we tend to get it and take some between five and 10 prompts which is literally pages of um pism to go into those images. So if you can avoid it do all right going on this is one of my favorite parts the proworker AI. Okay, proworker AI. This approach is to support your team. It's not about getting rid of jobs. We're in the human services sector. We can't afford people to keep leaving. We need people in this sector. So this is AI is not to eliminate jobs. It's to augment the jobs that we have so that people feel better doing them. So it can expand capabilities, right? we can get to solution thinking rather than always keeping lists of problems and we can figure out new and novel ways to support the mission and some of your staff are going to be really good at this and so let them lean over and help other in other parts of the organization why not AI is going to make it easier for them to do that generate the demand for expertise human beings right connections communication, judgement, responsibility, all those things. That's what we bring to our work in the nonprofit sector every day. We bring our minds, our hearts, and our souls into the sector to help people. And yet, frontline workers spend 75 to 80% of their time on administration work. really the people right there on the front of your mission trying to exercise why your nonprofit exists spends most of their time on administration. Hand that off to AI and let your people get back to experiencing why they came into the sector in the first place. Let people do what they're really good at. That's the best prevention on burnout is to have people have the opportunity to exercise why they came, what they want to do, solve things if possible, really support their clients. AI can help with that. Absolutely. Mastering new fields. My goodness, it's been what, more than 10 years, I think, since small and medium nonprofits have had a budget line for professional development. AI is provides a lowcost opportunity for this, right? Ways for people to stretch and learn things and better themselves, find new ways to help clients. A proworker AI says this is what the AI is for to support your staff and you. And then it's going to be better for your clients. Absolutely. Absolutely. When you use a proworker approach. So this talks about when you're using this proworker AI back to that focus on those core principles, right? Fairness, safety, security, privacy, accountability, and transparency. It's brought it inside your organization to support your staff. You're going to support your clients based on those values. And when we use that, when we choose the proworker approach, it's not for compliance. It's not like a checkbox. It's to bring in inside operations to really live our values of supporting people. It's what proworker AI can do. So here we are got few minutes to discuss really how to develop an AI strategy. So first you need a visualization right this is you need to visualize that road map. How are you going to integrate AI? You're going to use the responsible AI approach so that your values are inside your policies. You're going to communicate with you're going to train your staff and then you're going to measure how AI is helping your organization. You need to measure. You want to know that there are improvements. If you're going to be looking for problems, measure the successes as well. Look for them. They will be there. And then you want to communicate to everyone how you're using AI in your nonprofits. I was referring to this before. You're not going to use it all the time. And so you want to be specific about how and when you do your donors, your board members, general public, definitely clients, your staff. You want to be clear yourself. Communicate how you're using AI in your nonprofit. We talk about transparency and accountability all the time. This is one of the places we must show it. So this is your process. This is how I'd begin training staff in responsible AI. the six principles we discussed. Hopefully, you won't have any staff that disagree with those. Like to see them embedded in your policy. See them operationalized in your organization. And if you have any staff that don't want to use AI, assign them the role of being your quality assurance team. Yes, they're already feeling negative about AI. Give them the chance to exercise it in a positive way because it will help you. Your policy review and your alignment. You've got to have these policies because they're going to dictate how we do things, how you do things. It's not a checklist. AI policy is not just some it's not just a one and done. You're going to have to live it because you're going to be using it most likely every day, the AI. And so, you're going to want to go into a continuous improvement mode. You're going to look at what the AI is doing. Maybe you're going to have a chance to discover new ways to use it. Right? This has to be part of your strategies. Then you do a technology assessment. Well, who could do technology assessments? Well, I can. That's why I put it here. So, you use strategy. It's been my experience over the last 20 years that I've been working with nonprofits that we underutilize the software that we already have. Now, most people don't realize. So, for example, I was working with a client. She's an ED. She thought she would like to start dictating notes and reports and things like that, but she was unsure about using Otter AI or those um other kind of scribes, but she really wanted to do it. So I said, "Wordhouse dictation." Yeah. Right in your toolbar. There's a microphone and you can teach yourself. Only takes a couple minutes. Start dictating. You have to say, you know, you have to tell it commas, period, new paragraph, things like that. Word takes your dictation. I don't know how many of you knew that, but she certainly didn't. And that was a freebie available to her now and forever while she was still in under Microsoft. Also, in that technology assessment, you know, we're going to look at the issues that you think you want solved with AI and we're going to check are there other solutions for it. So, don't use AI if you don't have to. Right? That's going to be a number one rule. Don't use it if you don't have to. You might already have the answers in front of you just might be underutilizing it in your technology. Okay. Then your actions for implementation. Anyone who's done strategic planning, you know that you're going to have to have actions. What are you going to do this year, next year, and the year after? Three years for strategy, right? You're going to need a system for using the AI. Not everywhere, not all the time. You're going to document when and how you used it and measure. Like I said, you need to know the effectiveness. Now, this is something that I want to really point out and demonstrate for you. They've been studying and at at its most minimal usage, AI can help a person gain 30 minutes every day. Free up time for 30 minutes. And you think, well, big deal, 30 minutes. But I'm going to say most people feel better after a 15-minute meditation. Many people feel better after a 15 minute walk. Lots of us feel better after a 20 minute break just talking with co-workers. AI can give you back 30 each day, which over a week is 2 and 1/2 hours. Minimal benefit. 2 and 1/2 hours. Yeah. Now, that's for a person. If you're a staff of four, then that two and a half hours is actually 10 10 hours a week, which is more than a single day back to you as an organization. And what do you do with it? You don't create a vacuum and run ahead and say, "No, we can serve more customer, now we can serve more clients, we can do more this or that, we can work harder." That's not what proworker AI is for, and that's not how responsible AI should be utilized. It's to benefit you. An extra day, every week of breathing space is going to be incredibly impactful for your organization. And let's take this out over a month. That's equal to 40 hours. a week in one month. Your small team is going to gain back a week worth weeks worth of time positive benefit opportunity to reflect, take care of yourselves, address the burnout. That's what AI properly used can do for you. So as I'm winding up here, I wanted to talk about some sector specific AI. So there's some nonprofit professionals who actually built AI for you. They brought all their expertise and they put it into AI software. Transformer is for front AI is for frontline workers. They can measure and produce huge increases in frontline worker satisfaction and huge increases in client intake satisfaction. Imagine that. No longer necessarily heranging clients with with repetitive questions or sending them away because we can't give them service, but making them go through an hour intake process in the first place. Transform can fix that for you. It's fabulous. Grant Orb. Okay. AI for grant writing. I haven't written a grant in I don't know more than two years now. So Grant Orb writes a grant even produces your budget. What you do is you edit it. Does it fit with our organization? Do we have the capacity? Can we do this? Right. takes granter about five minutes to write that grant for you and you can decide right then and there should you pursue this. Yeah. Not after days and weeks of putting time in right then and there in the very first draft that comes from grant orb. Otherwise, you can spend an hour or two saying we're going to go ahead. This is what it needs to look like. It's done. If you don't have a fundraiser, you need AI software that can do this for you in just a couple hours and then you can produce grants all you want. Nagi, I know it's N A D L I uh but you pronounce it Nagi. It's indigenous AI. And what they do is they create modular data centers which use cool climate infrastructure for cooling. Their data centers use 20% less has 25% less environmental impact right now. Yeah. And they distribute the data centers again across different bands lands of course with their approval so that they have economic benefit from it and when they're smaller ones and they're dispersed they have much less environmental impact on the surrounding area. It's a great solution to a data set. Unfortunately, when you use a smaller data center, a modular data center, it can't produce the 100 kilowatt hours that the the government, the federal government is presently requesting and so they didn't get funded, which means they need to seek investors. But hey, if you're an impact investor, this is a place to put your money because it's going to come back and you're going to be doing immensely good things to help Canada in regards to AI console. Okay, we're the human services sector, right? Which means we're the ones that typically end up taking care of people during the day and then in our personal lives too. So console came up with an app that removes the mental load of caregiving, right? And there's so many people, a sound generation. I have a child, my youngest son has a developmental disability. The caregiving never stops. And if we're working in the in the nonprofit sector, we're doing double duty. Is anyone surprised that we're on burnout alert? That's the crisis in our sector and someone made an out to help us. So, I've come to the end of the presentation. Like I said, you could text me, email me, you look for me on LinkedIn, my Facebook, and my Instagram. Please follow. I would love it. So, I want to extend an invitation. I would like to work with organizations big or small who are interested in responsible AI and want to use the proworker approach. That's who I'm seeking right now that I want to connect with because my mission, as I said before, is to help the nonprofit sector feel better and do better through AI adoption. And so I'm really keen to work with the organizations who want to do this too. Let me help. This is so much what I want to do. The last two pages of this deck, I'll just show you very quickly are resources. Responsible AI and social technical theory. These two proworker AI and data containers. I wrote those articles for Charity Village and LinkedIn. You can read those and benefit from them. And here are also the Wikipedia links to my Shirro and responsible AI. Um, also Eli has a couple that I forgot to include such as the Bay Foundation AI guidelines and also the data sovereignty indigenous data sovereignty OKAP reference and Eli will be giving that to you. So Eli, thanks very much for giving me all this time. Thank you so much for expanding our minds and and questions are starting to percolate in. I encourage people to keep adding them into the chat and we'll pick up as many as we can over these next couple minutes. But let's like start up from with a question from Mia who just wants to really like help visualize what this could look like. Um can you just share an example of how you can actually measure how AI is is helping an organization? like how do we actually know that that it's making an impact or rather than just creating more curly burly activity. >> Yeah. Which is such a good one because that's what businesses have been doing, right? So there are several examples and the AI the frontline AI that I was talking about transform they're by they're by a company a social enterprise called flourishing systems. They're owned by Islamic family out of Edmonton which is the charity. They have a a blog that I love and there are a number of articles there and if people want to look about the impact of AI on um frontline workers on data privacy staff morale they have lots of information about that. So they are certainly one source that I would look to for my own climates. I I have to talk sort of in generalities you know because of the privacy concern but particularly that one that I was talking about the ED who decided to use dictation word dictation she didn't use AI she didn't use otter AI but she did go on to use grant writing AI and talking to her is like someone on wings like the excitement of a grant opportunity shows up and she doesn't get it a a sick feeling in the pit of her stomach, right? She just takes a little bit of time, lets uh gets the first draft written, takes a look at it, checks the budget, says, "Holy cow, we can do this thing." Done. That's a tremendous benefit from AI. I'm sure other people have questions and I could give you examples for the next hour, but I don't think have enough time. So speaking of examples, um Mikall has got a question here which is like can you share some examples of AI policies that other nonprofits have developed so we can be inspired by them? >> Yes, the as I was mentioned the Bayatti Foundation, they have their they make their policy guidelines available in French and English. It's mindblowingly good work, right? And there's no reason for a nonprofit in Canada to reinvent the wheel. Take theirs. They want you to. It's such good stuff. Covers, you know, because it covers everything I could think of, right? And I think that is the opportunity to see that when there's good thinking and it comes from a foundation from their perspective. What are they looking for? But what do their clients need? What do the clients of the nonprofits? because the clients are what do they need? How do you have AI work in such a way that a lot of the benefits come out again like responsible AI where you want to deal with the negative outcomes have the positive ones be elevated use theirs >> lovely that's really helpful so uh I've got a commenting question coming in here from Lawrence who's saying so when I'm thinking about policies they're thinking about two scopes one of those is the internal use policy basically guiding the internal teams and then the other part is the external statement on AI usage for clients press the public you know basically to help them understand how are you using AI as a as a public statement. What do you think about these thoughts on this approach? Is there like other contexts or audiences we should consider when we think about how do we communicate and set these guard rails? No, I think they're very good as long as you understand that audience. There's different roles within that external audience, right? Uh your donors for example. So fundraisers have been saying you're having a difficult time with publicizing that they're using AI. Many of the CRM that we use right now have AI embedded. Donors weren't aware and they never were aware that this has been going on for some time. Most of us weren't aware, right? And there's a discomfort with telling them because donors say I don't want to talk to AI, right? I want to talk to your fundraiser. I want to talk to your staff. So that's exactly how you use the AI and you publish this, right? Is that you use AI in emails where you're inviting people to where your fundraiser is inviting the donor to talk to them by providing a phone number and the hours that they're available. So, the donor's not talking to the AI email. They're getting an invitation to come talk to you, your fundraiser. And this is how you present that um to the various roles, your clients who might be concerned that you're using AI and doing something else with their data. Well, you definitely want to be able to tell them, "Oh, no, we're not. We have a data container that is going to ensure that your data doesn't go anywhere else but with us, right? First Nations clients. Okay, we have so much First Nations data. A lot of it has not been used well or wisely for many decades. We have an opportunity by following indigenous status sovereignty to really improve our relationships with the people that we serve from First Nations. Print your policy. Make sure that people are aware. Follow it. It's really important to convey your transparency when using this software because right now the majority of AI that's being used all around us, we don't know about. And it's the majority of AI. We have no access to it. We don't know how how it happens, what they do with the data, who's following us. It has been created as an environment without consent. As a nonprofit, you don't want to emulate that. You want to be different. >> Awesome. Just pulling up the next question. Um, yeah. So, I've got a question here from um Barbar Cook who is wondering just around the order of operations. So they were struck by the the idea that you wouldn't do your technology assessment before doing the AI policy, but rather you create the AI policy, then afterwards do your technology assessment. Can you just talk a little bit about the thinking behind the sequencing? >> Yeah. For me, actions follow policy. Some people do it the other way around and then guess what? Hey, we're all in trouble. We don't know what we're doing or we've done something wrong. We didn't realize it was wrong. we trip over each other. So when you after you t train your staff in responsible AI and then you do your policy work, everything that follows from that comes out from that. So then once you have your policies in place, how are you using your technology? Have you been protecting data? Have you been looking for bias? If you don't have policy, what is going to lead you to do that work in your technology assessment? That's why I set the process like that. Perhaps there's other AI strategy working in the nonprofit sector who bit to do it a bit differently for me, but that is making sure that people understand what they're intending to do. Have the policies that will lead to the actions that helps us evaluate that. Then we know how to make a strategy that we can set for today and the future. That's why I wrote it that way. Awesome. Two very quick questions because we've got two minutes. >> Okay. >> So the first one is when you talk about data container through does you does this mean for example that we'd use the Gemini API with a data container that contains our information and the data will never be used for learning by Google for example what do we containers? >> Yeah please do read the article that I've provided on data containers for the nonprofit sector. No, you don't let external technologies do your data containment because that's going to guarantee you're going to lose your data. It's going to be taken without your knowledge. Your data container is how you receive the data into your organization. Where do you put it? And then how do you make sure that nothing or no one else has access to it except for the express purposes that you've been given, the consent that you've been given by clients to use their data. It also means you might do something as staff where not everyone should be able to peruse client data. It's not necessary. And the more people who have access to data who don't need to have it raises your risk of not being able to contain that data and having problems. Right? So for your own organization, your data container is a set of actions that you will take that everyone is aware of but not that but not everyone participates in. >> Awesome. Okay, last question. This comes from Andrea who says or or wants to basically add a little bit of complexity here which is sometimes your organization may serve some very different client groups. So example, you might have one organization that generally just works with like adults, but then may have one department that's focused really on youth. So that's just a is that just a case where you're going to have to maybe write something a bit more complex to say when we work with these different audiences, we will approach data differently or how would you recommend working with that complexity when you may have groups that have will need to treat the data very differently? >> Yeah. Um that's by necessity something the difference between youth and adult is that we have privacy laws in Canada that say youth of a certain age cannot give consent. So you are going to have to be able to treat that data differently. You are going to need some kind of other parent or person who's taking responsibility to be aware. We have difficulties where some organizations particularly working with homelessness and youth, they need blind data collection, right? They can't or don't want to put uh the youth at risk by maintaining uh data that other people can access, lawyers, law enforcement, things like that. You will need to treat it differently depending on the populations that you serve. But that's a good thing. You have the ability to do that. You are the social workers in the field who understand about mitigating the risk and so it is up to you to create those AI policies that keep the same values as yours protecting clients. >> Awesome. We have run out of time. Thank you everyone for being part of this event today and for your great questions and Tina so grateful for your time and deep expertise Here.