Is your AI system making decisions that could be quietly costing you customers and talent? Every day, businesses are implementing AI solutions that may be perpetuating gender biases without even realizing it. From recruitment tools that favor certain demographics to customer service chatbots that respond differently based on perceived gender, these hidden biases aren't just an ethical concern – they're a serious business problem that affects your bottom line.
But here's the good news: identifying and eliminating these biases isn't just possible – it's a game-changing opportunity for your business. Today's guest is leading the charge in transforming how organizations approach AI equity, and she's going to share exactly how you can audit and improve your AI systems.
I'm thrilled to welcome Dr. Nici Sweaney, a globally recognized leader in ethical AI and the founder of AI Her Way. As the architect of the EquiAI Framework, she's helped over 60 organizations implement ethical AI solutions that drive real results. She's advised the United Nations, World Bank, and shaped Australia's Responsible AI Use Guidelines. Named one of Microsoft News' "Top 10 Trailblazing Entrepreneurs in AI to Follow in 2024," Dr. Sweaney is exactly who we need to help us understand and tackle the challenge of AI bias.
The AI Hat Podcast host Mike Allton asked Dr Nici Sweaney about:
✨ Audit Your AI Systems: Understanding your current AI tools and their potential biases is the crucial first step.
✨ Framework for Success: The EquiAI Framework provides a structured approach to implementing ethical AI solutions.
✨ Business Impact Matters: Gender-equitable AI leads to better decision-making, broader market reach, and improved innovation.
Learn more about Dr Nici Sweaney
- Connect with Dr Nici Sweaney on LinkedIn
- Follow Dr. Nici Sweaney on Instagram
Resources & Brands mentioned in this episode
- AI Her Way
- H.U.M.A.N. Framework for AI Adoption
- Magai
- The Surprising Truth About AI Adoption in Marketing Teams w/ Lightricks & AMA
- AI Work Buddy
- Grab the free Scalability Audit and find your leaky buckets
- Follow Mike Allton on Instagram and get more AI tips
- Explore past episodes of the The AI Hat Podcast podcast
Full Transcript
(lightly edited)
The Hidden Biases in Your AI (And How to Fix Them) with Dr. Nici Sweaney
[00:00:00] Dr. Nici Sweaney: On the risk side, when we have AI that accidentally reinforces buyers and stereotypes that can be really harmful to your brand and your customers and clients. Now, the opportunity I think also is that it represents your culture in a more kind of authentic way. And what I'm really passionate about is that everybody gets to design and own and mould AI that reflects them as a brand and a company, and even as individuals.
[00:00:28] So I think on the positive side, you can really nail your messaging and your culture fit if you use these things in a way that resonates with you. And then it's a scalability factor. So yes, we want equitable ai. We want AI that we are using in a way that's conscious. We don't wanna hurt our customers, we don't wanna hurt our brand.
[00:00:48] But then once you've got that right, and it's really reflecting your company culture, the scalability is where it hits your bottom line, right? The fact that you can have it deal with customer inquiries, the fact that it can do reporting analytics, it can understand strategy. It could be involved in your operations.
[00:01:05] You can do automated workflows. You can have agentic AI that replicates staff members and they're on 24 7 and they can process so much information at once. And I often tell people that I'm a small business, but I've got four human staff members and 11 AI staff members. It does the same sort of revenue as companies with double, triple the amount of human staff that I have because of what AI offers us.
[00:01:29] Mike Allton: Mike's got The Hat, the AI kind. Unlocking power in business, you'll find. Cut through the hype, explore tech. Every week, new ideas to check.
[00:01:54] Allton leads us on a digital quest. AI secrets, he'll always invest. Tune in tight, don't miss a beat. Knowledge, groove, keep you on your feet. Listen up! To the AI flow. Business growth, here we go. Cuttin' edge insights, that's the plan. With The AI Hat, you'll understand.
[00:02:38] Greetings program. Welcome back to the AI Hat Podcast, where I selfishly use this time to pick the brains of experts at harnessing the power of AI across all areas of business. From streamlining operations and boosting productivity to getting a competitive edge and driving explosive growth. Oh, and you get to learn too.
[00:02:54] Subscribe to be shown up. Prepare yourself and your brand for this AI revolution and come out ahead. Is your AI system making decisions that could be quietly costing you customers and talent? Every day businesses are implementing AI solutions that may be perpetuating gender biases without even realizing it from recruitment tools that favor certain demographics to customer service chat bots that respond a little differently based on perceived gender.
[00:03:20] These hidden biases aren't just an ethical concern. There is serious business problem that affects your bottom line. Here's the good news. Identifying and eliminating these biases isn't just possible. It's a game changing opportunity for your business. And today's guest is leading the charge in transforming how organizations approach AI equity, and she's gonna share exactly how you can audit, improve your AI systems.
[00:03:43] I am thrilled to welcome Dr. Nici Sweaney, a globally recognized leader in ethical AI, and the founder of AI Her Way. As the architect of the EquiAI Framework, she's helped over 60 organizations implement ethical AI solutions that drive real results. She's advised the United Nations, World Bank and shaped Australia's Responsible AI Use Guidelines. Named one of Microsoft News's Top 10 Trail basing entrepreneurs in AI to follow in 2024, Dr. Sweaney is exactly who we need to help us understand and tackle the challenge of AI bias. Hey, Dr. Sweaney, welcome to the show.
[00:04:17] Dr. Nici Sweaney: Hi. Thank you so much for having me, Mike. It's a pleasure to be here.
[00:04:20] Mike Allton: My pleasure to have you. I am so excited for this conversation. This is super important. I'd love if you could start by just helping everybody understand what are some of the most common forms of gender bias that you've encountered in some of these business AI systems we talk about all the time.
[00:04:35] Dr. Nici Sweaney: Yeah. I think one of the big things when it comes to AI is, what we find a lot of the time when we start these conversations with businesses and individuals is there is this perception and this fallacy that AI is great because it's unbiased, right? Or AI is fantastic because it has so much information in there, you can always get the right response.
[00:04:56] And what a lot of businesses and people don't understand is that that is just simply not true. So as an overarching structure, when we talk about ai, and a lot of the time we're talking about that, we're talking really about large language models. 'cause that's the entry point for most businesses and individuals.
[00:05:14] These tools, they're based on lots and lots and lots of data, right? And I used to be a data scientist for 10 years. So when you're looking at data, it's always about understanding patterns and looking for the story in the numbers or the code. And what large language models are doing is exactly the same thing.
[00:05:32] So when we're asking a large language model about who is the best candidate for something, can you write me a story for a boy? Can you deal with this question from a woman? Can you go through these HR complaints and tell me which ones are the most serious? They're using all of their data to look for patterns in how best to respond to that.
[00:05:52] And so what we see is that patterns often reinforce stereotypes, right? So people might have seen this memes flying around of asking large language models to make a picture of a doctor. And again and again, it's making this picture of a male doctor. And in fact, there was this viral me that went around, Gemini to create pictures and it created children, doctors before it even created a female doctor. Hmm. So what we get is these large language models, unless you understand that that's how they work, and unless you are directing them to correct for their bias, we can often fall into this fallacy of having these stereotypes reiterated back to us, and if we don't know what to look for, that can be really harmful to the way we represent our business, our brand, the care that we show our customers and our clients. And that's what we really wanna try and negate.
[00:06:41] Mike Allton: That reminds me of a situation I had run into Q4 last year.
[00:06:45] I was hosting an AI just as a social gathering here in St. Louis, like an after hours we were getting together at a pub and having some drinks and just anybody locally who was interested in Ai. So I was creating the event page so that I could send the link to folks and they could RSVP. And I wanted a, a graphic of people talking about AI at the top.
[00:07:03] And my thing at the time was, a morph, morph, sizing. I can't say that. Rightsizing, anthrop, anthrop, I can't even say it. So I was using bears in human clothing. I'll just skip that word. And that was like my theme for my blog. And so I said, Hey, let's, let's have a group of bears in, in, in human clothing, talking around drinks, have about ai, talking about ai and it still created five male bears.
[00:07:31] Dr. Nici Sweaney: Yeah.
[00:07:32] Mike Allton: I was like, come on. No women whatsoever in this picture. And I guess fortunately for myself, this has been an issue that I've been acutely aware of for many, many years. Even pre ai. Yeah. Because it's obviously, unfortunately, gender issues aren't limited to ai. Exactly.
[00:07:48] I'm in the marketing industry. So I was very focused on elevating female speakers and I run these events for Agorapulse, and I was always bringing in lots of female speakers. This podcast, two thirds of the guests are, are, are women. Yeah, something I'm very proud of and focused on. So I re immediately recognized there's a problem with this image. I know a lot of organizations don't have my experience in training for that. No. How can they identify if their current AI tools have these kinds of hidden biases?
[00:08:17] Dr. Nici Sweaney: Look, I guess the safest bet is assume they do. So all Large Language Models look for patterns and like I just said, patterns often reiterate stereotypes.
[00:08:26] And even with your bear example. These, these tools, they're not trying to be malicious. The, the weirdest thing for us to conceptualize is they have no emotion involved at all. They've got no idea that what they're giving us is an image that has meaning or text that has meaning. It's passing code to us.
[00:08:43] And we as humans are the ones that are interpreting any sort of meaning. So the reason you're getting lots of male bears is if you've asked for a group of people talking about ai, it's looked at all its data about who are the people that talk about ai. Over 70% of machine learning workforce are men. So it's gone well my best guess is that you want a bunch of dude bears, right? So it's not trying to be malicious, but assume that that is always happening. So when we advise businesses and when we teach people about how to start using these tools in a workplace setting, we give them examples about how to construct the prompts to always be.
[00:09:18] Looking for the other side of the story. So what we instruct is to actually ask these tools to remove any bias to remove any identifying demographic information before they make a value judgment. And then once you have information to say, Hey, is there another way of looking at this that we haven't yet thought of?
[00:09:36] Who would argue against this point. What would they saying? What would they say about it? What's a different perspective that we haven't yet included? So when we design even automations and agent AI in the way that we instruct our AI bots, large language models, agents, whatever it is, we're always also including some sort of way to negate against possible bias and gendered responses.
[00:10:00] Now that all being said, that's one construct. The other element to it is just purely the human element, right? I think too many people think of AI as this tool that they can either interact with over there or not. You are severely underestimating what it is. It's changing the way that we behave.
[00:10:18] It's changing what we do at work. It's changing how we'll spend our time, how we consume information, how we even relate to one another. So again, for a business, what I always advise is having some sort of working group community of practice, round table event and making sure that you have diverse voices that are playing a part in what AI means to your organization and are also playing a part in how that's designed and rolled out. If you've got diverse voices simply by the nature of them being diverse, you're going to get tools that better serve you because you've got lots of people having their say about how they want to interact with it.
[00:10:53] And as a woman myself, I don't have to think, oh, how would a woman say this? I am saying it like how a woman would say it, right? So when you've got diversity involved in what AI implementation looks like in your organization, you are naturally going to safeguard against a lot of those pitfalls that we see.
[00:11:11] Mike Allton: First of all, I could not agree more with everything that you just said. At least twice in every single episode that I've ever recorded of this show, I use the phrase AI Revolution, and I use it very deliberately. This is the fourth industrial revolution. It's not, to your point, just a tool that marketers are using or not.
[00:11:28] And you could just, yeah. Keep it at arm's length. No, no, no. It's a fundamental shift in how we act and react as a society, and people need to realize that, and they need to understand, to your point, that requires a change in organizational behavior. Creating an AI council that has diverse views, for instance, is a great example.
[00:11:46] And I loved your point about basically building into the prompt or instructions to the ai. Hey, please help us look out for, for biases and correct for that. I don't know that I'd ever heard that put quite that way before. Yeah. And I really appreciate that. I've heard tips that I now swear by, for instance, telling the ai it's okay to ask me questions, right?
[00:12:10] Yeah. And we prompt it, say, here's, here's all the information I'm looking for. Ask me clarifying questions. And it does, and it makes the whole process so much better. And I also tell it to challenge any assumptions and, and one of the funny things is usually it doesn't usually if think not my ideas are fantastic and I'm like, my ideas can't be that fantastic.
[00:12:31] I was just telling Liza Adams this this morning on a Slack channel. Just last week, my AI did challenge me and I was stunned and it couldn't have been more excited that it challenged me. I was like, oh my gosh, it works. It's fantastic. So I can't wait to implement your tip about telling it to look out for bias.
[00:12:48] Yeah, I think that's gonna be super helpful.
[00:12:50] Dr. Nici Sweaney: The thing is, like I've just said, it's not inherently malicious. It's also not inherently stuck on one mode because there's so much data in there. So again, it's like, and for some people they may not relate at all, but coming from a data background, when we're looking at a graph of the numbers and you often remove those anomalies, right?
[00:13:08] That don't fit into the story. So you can make your numbers look better. That's, that's really ugh. A hot take on what data scientists do. But then every now and then, we're really interested in the outliers, in those anomalies. So what happens when we're using that kind of language of the large language model, for example, and we're asking it to correct for bias, we're asking to look from a different point of view, or we're asking who's opinion have we not yet included?
[00:13:33] We're asking it to look at those anomalies and they're in there 'cause it has, quite frankly, a buttload of data in there. So by default it's looking for the most common thing because all it's doing is predicting what you want it to do. And by that it's saying, what do most people want here? What do most doctors look like?
[00:13:50] Okay, what's the most popular flavor of ice cream? Vanilla, right? But if we want some butter scotch, chocolate chip, fancy stuff, we can ask for it. It's in there. It's just not gonna give us that by default. So this is where I think AI literacy is really key. And when we talk to people. I often say like you, you've gotta, you've gotta teach man to fish.
[00:14:09] You have to teach everyone about how these models work and how to give good instructions to get the output that's useful for you that serves your organization and you as an individual before you can ever go up. I call it the kind of, the pyramid of AI integration before you can go up into automations and agent ai and AI systems and workflows and all that sort of stuff, where you're starting to really build staff members powered by ai.
[00:14:32] There's no point jumping there if you haven't learn the baseline fundamentals of how to get good stuff outta large language models.
[00:14:41] Mike Allton: Fascinating. Fascinating. And, and folks, I apologize for the jargon of a buttload of data. I know it's a technical term. We'll try to keep that outta here. I like,
[00:14:49] Dr. Nici Sweaney: is it, is it more or less than a switch of data?
[00:14:51] Is it more or less than a truckload? Who knows? I'm from Australia after all.
[00:14:57] Mike Allton: That's right. Well, Dr. Sweaney, earlier I mentioned your EquiAI Framework. I'd love if you could help us map out, you know what, what is that, what are the key components of it? How does it help businesses with their AI systems?
[00:15:09] Dr. Nici Sweaney: Yeah, so the framework is really about making sure that you understand what responsible and ethical AI usage looks like. So the EquiAI Framework is really about balancing regulation with innovation. It's this third way. In the past, we've had an either or situation through the highly, highly regulating something 'cause we're concerned with the safety or we're saying, yep, green light gates open, open the flood gates and here's all the innovation.
[00:15:37] When it comes to ai, I believe that standard regulation is too slow and too rigid for what AI offers us and how fast the tech is moving. There's no way that standard regulation can catch up to how AI moves. But I also think that just green lining and saying, Hey, shove it everywhere, everyone use it, is really dangerous if we we're not really careful about it.
[00:15:58] So the EquiAI Framework is a way for governments, but also organizations and even individuals to really step through that implementation. And there's a lot of moving parts to it, but basically it is about building fundamental ethical AI literacy and understanding what ethical decision making looks like.
[00:16:15] It's about scaffolding, use cases, testing at small scale, having some way to actually measure what the impact is, and then having a way to scale that and start to roll it out in a safe. And repeatable sort of way. So at its baseline, it's just a framework that really shows you how you can actually step that through.
[00:16:33] And what that means is that you are building capabilities while you're also reaping the benefits of ai. So right from individuals to governments. It's not about putting it everywhere. It's about starting to grow your understanding, starting to grow your internal capabilities, not relying on outsiders to tell you what this stuff is and how to use it and how to integrate it.
[00:16:55] And then starting to roll out iterative use cases so that you can really track and measure what this means, not only for productivity and efficiency, but also to people and culture.
[00:17:07] Mike Allton: I love frameworks. Like you said, they're a scaffolding to help organizations understand and build and grow. I talk about a H.U.M.A.N. Framework for AI Adoption.
[00:17:16] I name dropped at the outset some of these fantastic organizations and governments you've worked with. I'd love if you could kinda just walk us through an example, could be one of those, or another one that comes to mind where you've helped them understand this and kind of step through this adoption phase.
[00:17:31] Dr. Nici Sweaney: Well look, with the organizations that we work with, we do, again, a scaffolded, a scaffolded implementation plan. But what it normally starts off with is AI literacy for everyone. So all staff in an organization, and we, we work with lots and lots of organizations. So when we work with governments, it's often in an advisory sort of capacity.
[00:17:51] But we work with education institutions, we work with private business, we work with corporate. Every single one of those organizations goes through the same process. So the first thing that we do is that we teach everyone how this stuff works and what it means, and how they can identify use cases in their own day-to-day work.
[00:18:09] From that, once everybody has AI literacy, and we talk a lot about ethical use of ai, what are the problems with bias? Safety, ethics, data privacy, ip. From that, we then start to look at AI strategy. So I believe you really have to understand how it works before you can identify where you can use it. Some people do it the other way around, and I think that's a bit of a fallacy 'cause you just don't know what you dunno.
[00:18:34] So we teach everyone, and then we do an AI strategy. So we meet with exec and key stakeholders. We look at how does your business work, what is your tech stack? Where are your pain points? What are the bottlenecks? What do you want it to look like? What's your culture like? What's the readiness? Who here is interested in this kind of stuff?
[00:18:50] Who's gonna carry it forward? What's your organization structure? And then we identify, okay, how is AI going to change your business both immediately right now? And when we look at what we think is going to happen in the next 12 months and five years, it's a bit hard to predict. After that, we're looking at, a situation where we think we'll lose physical computers by 2030.
[00:19:14] By 2040, there are predictions that we'll have newly embedded devices that have an AI overlay across our entire perception of reality at all times. We really try and help organizations think about what does that mean for you and your type of work and the type of people that you will hire and what your current people do.
[00:19:32] What does that actually look like? So we map out an AI strategy and then we choose particular groups of people that are really keen and interested and represent diverse perspectives to start to be that internal working group that's gonna identify those use cases and start to apply it and test it. The really key thing is with everyone that we advise, it's also about capturing your baseline and then also measuring what difference it makes.
[00:19:58] I think way too many people. Governments included, right? I think at every level they're so excited or they're fearful, and the so excited ones just go use it everywhere. Like we worked with a company I can't name them, but Unicorn Company, and they spent a million dollars on subscriptions for all their staff, for Chat GPT and or to Gemini.
[00:20:21] And had absolutely no plan of what to do with it or how to identify what was useful and no way of measuring whether it made any difference to their bottom line. So for us, when we work with these places, it's really about capturing what are you doing right now and does this stuff make it better? 'cause it doesn't always make it better.
[00:20:41] It just makes you feel like you're doing something better. So it's really important to measure. And then with that, you get that growth of internal capability. And so our idea is always, by the time you finish working with us, you've got internal skill sets that are gonna carry this forward and are going to enable you to adapt to the next iteration or what happens next in the AI space.
[00:21:01] You are not relying on external contractors and consultants like us to tell you about that. You've finally grown your internal skillset to be able to manage that.
[00:21:10] Mike Allton: It is so gratifying to hear you spell it out that way. That's the exact approach that I have with businesses here in St. Louis. Folks, we're talking with Dr. Nici Sweaney about how critical it is for businesses to understand and build in equity when adopting ai. And I've got several more questions you aren't gonna miss, but first, a quick word from our sponsor.
[00:21:28] This episode of AI for Business Leaders, the AI Hat Podcast is brought to you by Magai. Tired of juggling multiple AI subscriptions and trying to figure out which tool to use and when? Magai simplifies AI for businesses by offering access to all the leading language models and image and video generators for one low price.
[00:21:49] With Magai, you can create custom instructions to personalize AI for every business need. And use them across all LLMs, including Chat GPT, Gemini, Claude, and even Deep Seek. No more building custom GPTs or gems that are limited to just one platform. Magai also lets you create customized workspaces for your entire team, giving everyone in your organization access to the same AI assistance and tools and prompts. Ready to streamline your AI workflow and empower your team?
[00:22:18] Tap the link in the show notes to visit Magai today to learn more and now.
[00:22:26] So as you're working with these, these larger businesses, organizations, governments, and I know they've got different departments, are there different roles that departments play in ensuring AI equity or is it just an IT issue? Who takes the lead there?
[00:22:39] Dr. Nici Sweaney: God, it's an everyone issue.
[00:22:42] I think this might be contentious. I I actually think it people are not necessarily the strongest carriers of, of this kind of stuff, and, and that is. All right, so now I'm being like ai, and that is a bit of a stereotype, but I see AI as more of a human issue. So what we often find with organizations is the people that naturally carry this stuff and identify with the holistic view of it are usually your people in overall operations and hr. So the people that understand how the business fits together and what all the people do in that business. And that's really about culture and approach to work and how people are spending their time and what your overall goal is as a business. It is not an IT issue. That being said, we always.
[00:23:31] We always encourage that there's a diverse group of people at the table, and yes, it should involve it, but I often don't find that they're the ones that are necessarily carrying this forward. So different departments play different roles. That being said, everyone needs foundational understanding.
[00:23:45] And then what we usually do is we'll break up the departments and have a discussion about what are the most useful use cases for that department and what does their work look like, and getting them to map out their workflows. Their customer journey whatever, that kind of, whatever it is that they're responsible for.
[00:24:01] And then having a look at how it applies to them. But as a baseline, the foundations of it apply the same to everyone. It's about understanding how these things work, how you can use them, how to keep you and your business safe, and how to use 'em in a way that. I know this sounds a bit meta, but I think contributes to global show social good.
[00:24:23] It's not always about the bottom line. It's about understanding that we're all playing a role in what AI looks like right now. And I think that that is a role that cannot be undervalued and we all need to take that responsibility seriously.
[00:24:37] Mike Allton: That really echoes a, a conversation I had on the show earlier with Lightricks.
[00:24:41] They had partnered with the American Marketing Association to survey organizations about AI adoption regardless of department. And, they were finding, predominantly it was leadership, it was operations who were, understanding, to your point that this is a much broader technology Absolutely.
[00:24:58] A much broader issue or a much broader opportunity that's affecting everybody. And I, I agree with what you just said about how AI is more important than the bottom line. That there, there's so much more at stake and so much more opportunity here. But I also know, the folks listening, they're gonna be wondering, everything that we're talking about today in terms of AI equity, how does that translate to a better business outcome?
[00:25:22] Have, have you seen examples of that? Yeah,
[00:25:25] Dr. Nici Sweaney: yeah. Well, I think so again, two elements to that. There is the negating risk factor. So you alluded to it in the introduction, we've had examples where big companies have really failed their AI implementation. There's notorious kind of tales of, Amazon having a bot that was screening CVS coming in.
[00:25:44] They discovered that for any technical job it was dumping women's cvs in bin. You had Tay, the check bot on X, only known as Twitter, turned into a racist misogynist homophobe in about five seconds. They had to take it off. So you've got lots of examples where even people that should know better that do this for a living get it wrong.
[00:26:04] So on the risk side, when we have AI that accidentally reinforces bias and stereotypes, that can be really harmful to your brand and your customers and client. Now the opportunity I think also is that it represents your culture in a more kind of authentic way. The other part about ai, I guess I particularly think about this 'cause I'm in Australia, these systems are not made where I'm from.
[00:26:30] And so by default these systems don't really reflect who I am. Just spelling things with Zeds instead of S'S is a great example. It's a very, very minor example. But when it's so apparent that it's not you, I think it's sometimes even easier to think about, okay, how can I make this tool serve me?
[00:26:48] And what I'm really passionate about is that everybody gets to design and own. and mould AI that reflects them as a brand and a company and even as individuals. So I think on the positive side, you can really nail your messaging and your culture fit if you use these things in a way that resonates with you.
[00:27:07] And then it's the scalability factor. So yes, we want equitable ai. We want AI that we are using in a way that's conscious. We don't wanna hurt our customers, we don't wanna hurt our brand. But then once you've got that right, and it's really reflecting your company culture, the scalability is where it hits your bottom line, right?
[00:27:26] The fact that you can have it deal with customer inquiries, the fact that it can do reporting analytics, it can understand strategy, it can be involved in your operations. You can do automated workflows. You can have agentic AI that replicates staff members and they're on 24 7 and they can process so much information at once.
[00:27:43] And I often tell people that I'm a small business, but I've got four human staff members and 11 AI staff members. And we're able to do, we all, all the humans work part-time. But my business, right? It does the same sort of revenue as companies with double, triple the amount of human staff that I have because of what AI offers us.
[00:28:06] So I think it's wanting to represent your brand correctly and then beyond that, it's a scalability factor once you get that right.
[00:28:14] Mike Allton: Oh, that is exactly right. I, I work full time for Agorapulse, so this is a side hustle for me. Yeah. So I have an AI chief of staff. That helps me with everything 'cause it's just me.
[00:28:25] There are no other humans that work with my business anywhere near me. I can't get over to anybody and ask them for ideas. So I've got a custom instruction. I happen to use Gemini. So it's a gem and it's trained on who I am, how I talk, who I'm talking to. What are my challenges, what are my goals for the year.
[00:28:41] I tell the story. Often it, it knows that one of my big goals is to work more with local businesses. And so he's always trying to make me create my content specifically for St. Louis businesses. Even though my content, my newsletters in particular, they're going to a much wider audience. So I keep having to roll 'em back and say, no, no, no.
[00:28:57] That's, this is not just for St. Louis consumption. Shut it down. Totally. Yeah. Right. So I, I had, I've had so much success with this, I actually developed it as a, as a free download for people to figure out this on their own. I call it the AI Work Buddy. And so I wanted to create a prompt and I. On how I built my AI chief of staff, and there are so many pieces of information that I plugged into it.
[00:29:21] The prompt was getting too long, so I ended up building it out as an instruction file in A PDF with like 30 questions. Yeah. And the prompt now is just a paragraph saying, Hey, we're gonna create an AI Work Buddy. Ask me the questions that are in this, this instruction set. And then when you get through all of them, turn all the answers into a custom instruction that can go into a gem.
[00:29:40] So folks, I'll link to that below if you want to get that for yourselves. But yeah, I'm wondering that. Yeah, it's, oh gosh, it is so helpful. So helpful. Yeah. Yeah.
[00:29:48] Dr. Nici Sweaney: I have a business strategist bot. And yeah, I talk to her all the time and, we do like quarterly reviews and we review revenue. We have a look at wasted costs.
[00:29:59] We have a look at what's working the best. But also the big piece for me, like you, is she's really trained on what I ideally want. 'cause when you're doing your own thing, like it's easy. You've got shiny object syndrome, there's a billion tasks being thrown at you, and everyone's like that. I don't even think it's just business owners, I think even in your work. So she's really trained on, Hey Nici, you wanna build this legacy, you want this sort of impact. You want this to look like your lifestyle. You wanna spend this much time with your kids. You know you wanna do this with your partner on Fridays. You want this sort of revenue, you wanna donate to these causes.
[00:30:30] Let's refocus. This is how you get there. Stop doing that and do more of this. And I find that so helpful because otherwise, I'm just, I'm a yes girl. I just say yes to everything and then I'm like, oh, that actually was not aligned and my business just pull me back a little bit. And she was like, you can't.
[00:30:47] Stop trying to do everything.
[00:30:51] Mike Allton: That's also good advice. So, so what is your advice for companies who are just starting their AI journey and they're, they're really aligned and in tuned to everything that you just said. Mm-hmm. They wanna make sure that they're building in equity from the ground up.
[00:31:05] Dr. Nici Sweaney: Yeah, well, I think the first part for companies I would advise is to ask your people how they're already using it.
[00:31:13] So even if you don't have a company stance, even if you don't have company subscriptions to these tools, data shows that roughly 30% of employees are using specifically large language models without telling their employers. So there is a really good chance that there is a whole bunch of AI already happening in your organization and you may or may not know about it.
[00:31:34] So I think the first thing is to don't give everyone in trouble. You need to create a landscape of permission and exploration and to get ideas about how it's already being used, because that's going to show you where those sort of opportunities lie. But more importantly, who your AI champions might be.
[00:31:52] The next step after really understanding what's currently happening is for large scale AI literacy. So this is the foundations of understanding what AI is, what is the difference between AI and generative ai, right? Because they're very, they're very different things and we, we've all started talking about AI and most of us actually mean generative ai, but what is generative ai and how do these things work?
[00:32:17] Why do they give us the responses they give us? I taking off the kind of blinders or the wow, it's a magic black box. And actually understanding a little bit about how it works and what I say is it's often like driving a car, right? Not all of us are mechanics. We can't all take a car apart and put it back together.
[00:32:33] Not all of you need to be machine learning, data scientists, but what you do all need to know is if it's diesel or petrol, and you need to know that the steering wheel actually makes it go in a certain direction. So you've gotta understand that sort of stuff. And then after everybody understands that, then you start with very small use cases in certain groups so you can actually measure what has happened, what works well, and you can actually see how that change is delivered and if there is a change in how you work. So that'd be the way to start and and I always also say, just start with a large language model, whether that's Co-pilot, Chat GPT, Gemini, or Claude. They're really the four big players on the market.
[00:33:13] Just choose one and learn how to use it well before you start thinking about all the other stuff. cause everyone gets very excited about automations and workflows and there's AI agents on the market and maybe I can just talk into my phone about the 17 things I need to do and I'll just have these little bots go off and do it.
[00:33:29] That is all really possible. And if you dunno, that's possible, that's possible, but it's all really possible. But you can't actually do that until you get good at that base of the pyramid that I talked about. And the base of the pyramid is prompting a large language model really well.
[00:33:45] Mike Allton: That's absolutely right, but I did a whole series just on Gemini.
[00:33:48] 'cause I found a lot of people are already paying for Google Workspace, which means they've got Gemini. They don't need to go out and pay for Chat GPT, and Claude and download Meta and do all these other things. Just use the tool that you've already got. If you don't then exactly. Maybe look at another one, but it complicated.
[00:34:04] Yeah.
[00:34:04] Dr. Nici Sweaney: And I, And people always get overwhelmed with, which one's going to be the best? And it's like, well, you can try all of them for free firstly. And then whatever one you feel like gives you the nicest sort of answer for you. They're all a large language model that fundamentally do a similar sort of thing.
[00:34:20] They've all just been trained in slightly different ways. They might have slightly different data sets. So you don't have to be across 'em all. We're an AI consultancy, so yes, we do use them all because we have to advise when it's the best fit for the right thing. But for an individual and even a whole organization starting out, there is really no need to try and be across everything.
[00:34:39] Choose one thing, learn how it actually will help you, and then start to scale it from there.
[00:34:44] Mike Allton: Yeah, and we're at a moment in time where the businesses behind these models are in a race to be the best, which means every other week there's new features coming out, new versions of the models coming out. Yeah.
[00:34:58] Like, I just opened as an individual user, who cares? We just, just
[00:35:01] Dr. Nici Sweaney: Exactly like I just opened Claude yesterday and I don't know how long it's been there, but it, it prompted me that now it can link to my Google calendar and to my Gmail. So within Claude, I can pull, I can pull my calendar and say like, let me help me reprioritize all my task and what's on tomorrow and what should I actually be delegating to my assistant.
[00:35:21] And I can work with my calendar inside of Claude. I've done it in past, through API and Web hooks of sending my calendar to various places to talk about it. But now it's integrated into Claude. But then, they'll bring out that functionality, which means Gemini and ChatGPT will bring it out next week.
[00:35:37] Mike Allton: Exactly right. That's exactly right. Now, earlier you were talking about. The, how you talk with your clients about the future of ai, within the next few years, not needing physical computers potentially, and, and all that kind of thing, which I, I agree with. What do you see as the future of ethical AI implementation in business, in the few minutes we have left.
[00:36:00] Dr. Nici Sweaney: That's a small question, in a few minutes we have left, Mike.
[00:36:05] Look, I, I guess I have Utopian hope. I have Utopian hope that by teaching as many people as possible about what ethical AI use looks like, that we will not only make our businesses better and our personal lives better, but we will also make the world a better place. So, at a big scale, I think that AI is.
[00:36:25] Is actually probably our only real shot at solving really big challenges that we have, like climate change, like world hunger, like water salination, like, food security. And I think that without ai, we've been trying for the last 30 years to solve these things we haven't yet. And so the power of being able to look at lots and lots of data at once I think might actually help us solve some of those challenges.
[00:36:50] But on the social side of it, AI is shaping the way that we relate to each other and the way that we think. There are already studies that show that girls' ideas about what careers are applicable to them are being molded by the responses that AI give them. So when we think about what equitable AI looks like.
[00:37:10] For me, it is about, let's add to your bottom line. Let's let you launch and scale those new fabulous ideas and let's let you bring things to market that would've never, ever been possible for. And that's amazing. 'cause I think the more people that are able to put their ideas out there into the world, the better place that the world is going to be.
[00:37:30] But having an understanding of what equitable AI looks like. Means that when you do that, you're not just launching some business revenue model and you're not just taking a new idea to the world. You're taking a new idea to the world in a way that's gonna show the world and reflect back to them the good type of society that I hope that everybody hopes that we live in.
[00:37:48] One that's fair. One that's equitable. One that has representation and diversity and shows everyone that there is space for them. I think if we don't have equitable ai. We will only give people that already have a lot of power, more power, and we'll leave people that are notoriously left behind, even further behind.
[00:38:06] I don't think we're gonna close the gap unless we are very serious about the responsible use of ai. I think we will only widen the gap. That's my kind of view.
[00:38:19] Mike Allton: I love that message and I'm actually seeing that every single day. I love how my girls use Pi and it's incredibly empowering and supportive to them.
[00:38:27] It's terrific. This has been such a powerful conversation and yet also you've just been an absolute delight to talk to. So thank you so much. Thank you for coming on. Show. So the show, I know folks are going to want to follow you and connect with you. They're gonna have tons more questions. They absolutely sh need to watch and pay attention to the content that you're putting out 'cause it's terrific.
[00:38:47] Thank you. Where, where, where should they go?
[00:38:49] Dr. Nici Sweaney: So I am on LinkedIn a lot as Dr. Nici Sweaney and on Instagram as Ai her way. And you can find my website at www.ai her way.com au. But reach out to me on any of those platforms. We have lots of free resources, online learning hubs, and then of course we do workshops and PD and then company consultancy agreements.
[00:39:12] So whatever you need, there's something, but I love the questions. I love having. Really big discussions about what this means to us in our future. So please reach out.
[00:39:23] Mike Allton: Fantastic. Thank you and thank all of you for listening. That is all the time we've got for today, friends. But remember, if this is your first foray into the world of AI in business, I've got the perfect resource for you in the show notes below an AI primer.
[00:39:34] Decided to just quickly get you up to speed on all the terms, the tactics, and the underlying technology that's powering this AI revolution. Check it out and I'll see you in our next episode. Welcome to the Grid.
[00:39:45] That's a wrap on another episode of The AI Hat Podcast. I hope you've gained some valuable insights and actionable strategies to boost your business with ai. If you enjoyed this show, please subscribe, leave a review, and share it with your network. And don't forget to visit the ai hat.com for more resources and to connect with me.
[00:40:03] Until next time, keep innovating and remember, the future of business is human and it's powered by AI.
