Sarah Downey: AI Ethics When There Are No Rules Yet | Vancouver AI April 2026

0:00Incoming transmission [music] from the future. Vancouver AI community meetup. All right, let's get into the good stuff. Um Sarah Downey, come join me, please. Sarah Downey came over from Victoria. She's been doing change management for nonprofits for like 20 years. She's developed a unique AI methodology for working with nonprofits and onboarding them into this whole new world. And so, I was at the Um VIATEC, come on up. I was at a VIATEC event, Vancouver Island Tech event, earlier this year and we met and connected and um I've known Sarah for a long time, too. She's sort of the Power of Hope Canada, which is a really, really cool nonprofit um that has uh made some of the most unique individuals

0:42that I've ever met in my life. Like uh creative leaders like Roop and like Nadia and stuff like that. So, I really respect your work. She's here tonight to give us a an intro to ethical AI. And then, as I said, Sev and Martin are going to come up next. So, anyway, thanks so much for being here, for coming over and um we'll tell you later about a a course that we're leading together.

1:01Awesome. Thanks for the intro. >> [applause] >> So, uh I am going to tell you a story. Eight months ago, a stranger reached out on LinkedIn. And he was building he was building tools to help people use AI with more integrity. And I was uh building AI governance processes for nonprofits. That's part of what I do. And so, we got into an interesting conversation about a tool that he had built that was a like a bias checker GPT. And it was a great uh solution for me to share with some of the nonprofits that I worked with that were starting to experiment with AI.

1:48And it got my mind thinking about um just the way that people step up and solve problems in this space. And it was really inspiring. All right, I'm going to read you something. The problem that confronts us today, in which the nearest future is to solve, is how to be oneself and yet in oneness with others.

2:15To feel deeply with all human beings and still retain one's own characteristic qualities. Oh, thanks. So, does anybody know? So, that was Emma Goldman in 1910 not talking about AI. But, what she was talking about is relevant to every time in history when something comes along and reorganizes everything around us. And we have to figure out who we are within that change, within that time.

2:55And right now, we got a whole lot of problems. So, I'm going to give you a bit of a drive-by uh look at the ethical AI landscape today. So, let's get going. Um and it really is a drive-by. There's I'm going to leave out a lot. And I'm going to simplify it cuz I'm just talking for 10 minutes.

3:19So, number one, power. We have less than 10 companies that control the AI models that most of the world uses now. And the decision-makers generally aren't the ones most affected. And innovation and speed in those companies is rewarded and care is not. So, let's talk about how people are affected. So, what it does to people, bias.

3:51We know that it's trained on human data and human data is filled with prejudice of many kinds. And when we say bias, we're talking about Oh, you found one. Great.

4:13One I'm I'm going to just I'm going to keep going. >> [laughter] >> I feel bad for people watching the recording. Okay. So, when we're talking about bias, I I like to not just gloss over it and you know, this is going to be quick, but we're really talking about sexism. We're talking about racism. We're talking about all the things that aren't great that we've put into the world and now AI is trained on that data and feeding it back to us and amplifying it. All right?

4:52And then we've got the labor problem. Jobs disappearing. We've got manipulation. So, you know, AI studies your psychology and targets your purchasing decisions and this has been happening for a long time, but you know, throw in amplified corporate greed and this is getting scary. Take four, take two. >> [applause] >> Thank you.

5:20I just needed two hands to hold this pair of glasses. Go for it, please. >> [laughter] >> Got it. Uh misinformation. Deep fakes. Knowing what is real is becoming a skill that we have to develop. Yes. All right. I have somebody close to me in my family who just unfortunately fell for a scam that was based on a a deep fake video and lost $10,000.

5:49Yeah. And this is we're seeing this happen a lot with elderly right now. Something I teach about. And then let's look at what it does to the planet. So we have AI, you know, training models um and generating content, especially images and video, that consumes enormous amounts of energy and water. So there's real environmental costs.

6:15And let's talk about the media. There's so much noise all around us in the media. It's so disorienting. It's hard to feel the ground that we walk on. And and the media is amplifying both the fear and the hype, right? So AI will cure cancer and AI will end humanity. We've got it all coming at us.

6:37Uh governance. The European Union has an AI Act. The US had an AI Safety Office, but just dismantled most of it. Um Canada has neither. And BC has no binding frameworks for its companies. So nobody is drawing the line for us. And so for now, we draw that line for ourselves. So what do what do we do when the rules haven't arrived yet?

7:09And this is where we talk about, you know, having um inner compass. Like what do we personally feel is the right way to go? And these are the conversations that I'm around. I'm certainly not an expert in this, but I do a lot of facilitating of conversations to allow people to think deeply and speak up and start to connect the dots to their own values around how they want to use these tools.

7:38So I'm just going to share with you a a things that I've noticed from my little corner working with nonprofits. Um a lot of them have chosen to use AI to uh share images that they couldn't otherwise share to tell the story of what's going on in their nonprofit, right? Because a lot of times images um you know, you can't take photos of your your vulnerable populations. Just wouldn't be appropriate, right? So, they're able to actually use AI to make excellent images that tell the story, that get the funders and the donors to give money to the charities so those people can be helped. And this is actually working quite well.

8:21So, that's a use case. Other people are absolutely not using images because of what that means for artists. Um when I thought about it for myself, I'm an artist and I trained um uh image maker tool on my style of artwork. I like to work in watercolor and pen and ink, and I trained it on my style. I gave it 20 different paintings and I got it to analyze it so I could make a prompt that was related to my style and and upload images. And so now, as a small business owner, when I need images, I can just call on it. And I'm happy with that. That really works well for me. But we're all figuring out where

9:02we land with all of these. Um writing attribution. You know, who Who has used AI for writing something that has been published? LinkedIn? Yeah. Yeah, me. I've done it. Um many of us are trying to figure out, well, how do we attribute that? Right? And we're starting to see a lot of people attributing it more.

9:32And also, at the same time, we've got journalism organizations that are saying like absolutely no, we want to preserve human journalism and creative integrity. So, like locally the Tyee, um New York Times, Economist, CBC News. I didn't mean those last three were local, just the first one, the Tyee is. That's why you were laughing.

9:55>> [laughter] >> You know, they're saying, you know, absolutely no to um AI-written journalism. Okay. I want to go back to this. I'm just going to read this quote again. So, the problem that confronts us today and which the nearest future is to solve is how to be oneself in oneness with others, to feel deeply with all human beings, and still retain one's own characteristic qualities.

10:29So, this is about finding our own in in inner compass. All right? And sitting with what um what you value. So, like I value this, so I make this choice. And for me, one of the things with writing is I really value transparency and critical thinking. That whole piece about knowing what is real is becoming a skill that we have to develop. It's something that I I think a lot about.

10:58So, now when I'm writing, if I am using AI, I make an effort to always list my sources at the end, so that my reader can they don't just have to believe me, they can follow where I got the good information from. Follow my follow the thinking, right? So, there's just little ways like this. And so, yeah, I invite you um tonight and in your time to just be thinking about what tracks for you and to start having these conversations with with other people.

11:32Um Okay. Eight months later, I reached out to that same stranger on LinkedIn, who wasn't a stranger anymore, but we hadn't been in touch for all of that time. But, I had an intuition. He was just on my mind, and I didn't know why I wanted to reach out, but I did. And, of course, he said, "Oh, Sarah, I I've been thinking about you, and I've been sharing your AI governance framework with some of the nonprofits that I work with." Great. And so, we got chatting about um just what's going on in life, and I was mentioning I was really excited about this responsible AI certification course that we're launching. And I I asked him about that

12:15critical uh thinker GPT that he had created. And I said, "You know, I'm I'm not really using GPT anymore. I'm working in Claude. Is there Is there any way that we can like that I can bring that over somehow? Um I do not have a technical background. And he was like, "Huh. [sighs] Give me a couple weeks."

12:41So, we reconnected, and he basically built a uh connector on Claude that's an MCP server. I had to write this part down. This is really why I have these notes. Cuz this was impressive. It has He worked on This is from uh information that he gathered for the last 2 years he'd been working on this, but he finally put it all together so that we could use it in the course. So amazing.

13:091,300 indexed entries, 200 biases, it's trained on 163 reasoning frameworks, 188 thinking lenses, paradoxes, fallacies, 300-plus methodologies, indigenous reasoning models, and across multiple cultural perspectives and it really will pull out it will de-westernize the thinking it will name when something is westernized in thinking. So this is an incredible critical thinking toolkit that we can basically show any piece of writing to report whatever and it's going to be able to critically analyze what's in there. I've tried it out it's absolutely fantastic and I'm so excited to share it and so the the values we're talking about values that I'm seeing so much right now is the ones of generosity.

14:08Right? And so like I I just want to say I'm so grateful for places like this and proud to be part of this ecosystem where we have people like all of you coming together and all those problems that we talk about I'm just seeing so many people come up with solutions creative solutions very generously working on all of these together. So that's is what I will end you with but not yet because are we going to go right into the the talks? Yeah yeah so I'm super excited to learn from Sev and from Martin tonight as I know you all are and you know I gave you the drive-by ethical picture and and they're going to

14:51go into a much [clears throat] deeper dive. And more from the inside so Sev's going to talk about you know how do we measure true impact on society Martin's looking at responsibility on AI builders and the design choices that they are making. And that's that. >> [applause] >> Thank you. [cheering] >> [applause] >> I'm so happy you could make it over me a part of this and I hope you will come. I don't think I told you but I'd like for you to come back on stage at the end for the debate part.