Inside Outside Innovation

On this week's episode of Inside Outside Innovation, Robyn and I talk about the rise of the high-impact individual contributor, more startups and fewer jobs, and how AI is going to destroy our lives or not. Let's get started. Inside Outside Innovation is the podcast to help innovation leaders navigate what's next. Each week, we'll give you a front row seat into what it takes to grow and thrive in a world of hyper uncertainty and accelerating change. Join me, Brian Ardinger, and Mile Zero's Robyn Bolton as we discuss the latest tools, tactics, and trends for creating innovations with impact. Let's get started. [00:00:00] Brian Ardinger: Welcome to another episode of Inside Outside Innovation. I'm your host, Brian Ardinger, and with me is Robyn Bolton. Hello, Robyn. How are you? [00:00:45] Robyn Bolton: I'm good. How are you, Brian? [00:00:47] Brian Ardinger: We're doing well this week. We're excited to talk about all the new things that are happening in the world of innovation. I'm going to just jump right into the articles. The Rise of the High-Impact Individual ContributorThe first article we want to talk about is from Alaina Verna. She has a blog post called IC Work Is the New Career Flex, with the subtitle of The Rise of High-Impact Individual Contributor.We're going to talk about the individual contributor and the fact that it is changing the way that careers are set. She talks a little bit about the fact that, you used to build a career on, how can I become a VP and move my way up the ladder?And now with the rise of AI and tools, et cetera, an individual contributor, maybe not managing a group of people can actually have a major impact in the world by, working with agents or building tools and having the tools to impact and build things themselves. So, it's a interesting talk about this new world and how maybe the rise of the individual contributor is going to change the way we do work.[00:01:44] Robyn Bolton: Yeah, so I love this idea. As soon as I saw this article from you, Brian, I saw Hi-C, like I flashed back to lunch in during my entire K through 12 and every day I had a Hi-C juice box, and I was like, "Yes, tell me more about Hi-C." But this article broke my brain in a way that was ... should've been expected because one of the early points she makes is that as a high-impact individual contributor, you probably are essentially running a team, but you've got a team of agents that are helping you.How AI Agents Are Changing Career PathsAnd so, you should actually be compensated and rewarded as someone who has a team, as a team leader, because you're doing the ... You're producing the value that a team is producing, and but you're still an individual contributor. And for some reason my brain went into HR mode and was like, "Gah, how is... that doesn't work."But I think that's going to be the reality. There are today billion-dollar companies that have 10 human employees and, armies of agents getting the work done. So I think this is yet another disruption that we're gonna be facing of now with agents, now with AI and LLMs and all those things that we can have one person producing the value and doing the work of a whole team, and how do we adjust companies, org charts, incentive systems- all of that to reflect that? [00:03:14] Brian Ardinger: And I think the other important part about this is the fact that, individual contributors, you have to start thinking about what is my actual contribution? Where, what am I good at, and what do I have to... What can I lean on the AI to be good at?And truly understanding what, where you can add impact is gonna be more important. It used to be where you're told, "Here's where your job is and this is where you're going to impact it." Now you have the opportunity to spread those wings, but you also have to be cognizant of where, what value you c- you can add and where you can leverage the tools and technologies that are out there to make you even better, and to maybe- Yeah add value that you hadn't been able to add before. [00:03:48] Robyn Bolton: Yeah. And I will say there was one little kind of red flag in the article about midway through, where she talks about when she had a team, she's part of a, growth team essentially at her company. And she wrote I've always loved what I do, but a big part of my job was draining me, building decks for cross-functional alignment.Why Cross-Functional Alignment Still MattersAnd I get it, building decks is draining. But that muscle around cross-functional alignment and the need to have conversations with your peers and build cross-functional alignment and get support from other humans is a really important skill. And there is a there is a risk that if you are an individual contributor and you love being an individual contributor and you love doing the work, that you go all in on just doing the work, and you don't seek that alignment. You don't seek that input.So there, there is that still little bit of you can't go all in on being the individual contributor because humans are still part of the organization, and you still need support, and you still need to be able to work with other people. So little bit of a red flag, but, still an interesting thought.[00:05:02] Brian Ardinger: And it'll be interesting to see how this, again, how this plays out as more and more people start, learning and playing and with AI and agents and the future has not yet been written. [00:05:10] Robyn Bolton: That is the truth. [00:05:12] Brian Ardinger: All right, the next article is More Startups, Fewer Jobs by Donna Harris. She has a LinkedIn article talking about for 15 years, the startup economy has operated with a relatively stable assumption that more startups equals more jobs.But she goes on to talk about the newest Kauffman data suggests that relationship is changing. Yes, we have more founders, but we're producing fewer jobs per founder. And so maybe it goes back to the previous article, these individual contributors. Can you do more? Can you build a startup with fewer people?More Startups, Fewer Jobs in the Innovation EconomyAnd that idea of startup job creation apparently it peaked in 1997 at 7.9 jobs per 1,000 people, and in 2025 that figure is now 5.3, so it's a decline of, roughly a third. And what is that impact for startups, for communities if you can build companies with fewer people? [00:06:02] Robyn Bolton: This was an interesting article because it raised for me more questions than answers. And she seems to use startup and entrepreneurship synonymously, which I totally understand why you would, but I think they can, in the context of this article, have two different meanings. [00:06:18] Brian Ardinger: Right. [00:06:18] Robyn Bolton: And it even begs the question of how do you define a startup? And she talks about how the definition of a startup has changed, and has become very much around VC funding, et cetera.So it's one let's get clear of what we're talking about on what is a startup then versus now, what is entrepreneurship. But, ultimately it, it does come down to the fact that, yeah, with ...

On this week's episode of Inside Outside Innovation, we talk about how change is changing, the IKEA effect on MVPs, and how AI is making companies more ambidextrous. Let's get started.Inside Outside Innovation is the podcast to help innovation leaders navigate what's next. Each week we'll give you a front row seat into what it takes to grow and thrive in a world of hyper uncertainty and accelerating change. Join me, Brian Ardinger, and Miles Zero’s, Robyn Bolton as we discuss the latest tools, tactics, and trends for creating innovations with impact.Podcast Transcript with Brian Ardinger and Robyn Bolton[00:00:30] Brian Ardinger: Let's get started. Welcome to another episode of Inside Outside Innovation. I'm your host, Brian Ardinger, and my co-host Robyn Bolton is with me. Hello Robin. How are you?[00:00:45] Robyn Bolton: Hello, Brian. I'm great. How are you?[00:00:50] Brian Ardinger: I am doing well. We are in the middle of December.[00:00:54] Robyn Bolton: And hard to believe that the year is, it's almost over.[00:00:58] Brian Ardinger: Well, I'm ramping up for 2026. We've got a lot of stuff to talk about for the conference that we're going to be planning. Yes. We'll talk to you a little bit more about that, but let's just jump in. We've got some articles to discuss, some things that we've been seeing out there in the ether when it comes to innovation.Innovation and AI in 2026. Setting the Stage for ChangeThe first article is Change is Changing How to Meet the Challenge of Radical Reinvention, published by McKinsey. There's a lot of things I want to dive into this. Obviously, McKinsey's probably in the wheelhouse of disruption, as their game is going to be changed. So, I read this article with two lenses.One, the lens of what they are telling their clients, and then two, are they eating their own dog food when it comes to this? So, for reference, the McKinsey article talks about, you know, when change becomes everywhere, every, everything everywhere, all at once. How are companies dealing with this? And it's no wonder that the average employee experiences all these particular changes and is worn out.And I think one of the things they quoted was the number of new experiences that the average employee faces is fivefold increase than a decade ago. The fact that organizations and leaders have all new types of tools and skills and methods to navigate this changing, complex state, and those old tools don't necessarily apply today. So I love your first insight, and we'll go from there. Radical Reinvention and Innovation Strategy. Rethinking Change at Scale[00:02:13] Robyn Bolton: McKinsey turns out great stuff and great frameworks and all that stuff. And also, as someone who's been a consultant for entirely too long, they also make me laugh. Like in a super nerdy consultant way. Because one of the things that's laid out this article is the four Cs of change.You know, it starts with C1 execute, C2 mobilize, and then we get to C3 transform, which was the buzzword of, I feel like the last five years. You know, we're transforming everything. But now we have a new one. It is level four change, and it is reinvention. And I just loved when we rebrand things that are the same thing we've always talked about, but we've rebranded it, and now there's a different diagram, so it's completely new, and you need to buy from us.But there is good content in here. And you know this idea of creating value with the new identity. Talk about way easier said than done. Like every organization has an identity. When you ask someone like, what do we do here? The answer is the organization's identity and to change that is about as easy as changing an individual's identity, which is to say not at all.Organizational Identity, Leadership, and Innovation Fatigue in the Age of AI[00:03:30] Brian Ardinger: Well, and I found that part of the article actually the most intriguing because I think when you think of McKinsey, again, a lot of stuff they've focused on is how do you optimize and execute on your model? They're not very focused on reinvention. And kind of blowing it up and starting over from that perspective.So, I thought that was an interesting take, that either they have to recognize the fact that the companies that they're working with and the companies out there in general are going to have to reinvent themselves. It's no longer table stakes just to, you know, do what you've been doing and make it better, faster, stronger, cheaper.It's how do we navigate and potentially reinvent what we've done in the past. And then the other key aspect of it that really resonated with me was the fact that how do you create a culture such that change is not a drain to the organization? But as a source of energy, again, coming from a consulting background, that's often not the things you talk about.Like how do you actually create change that energizes and excites people, and provides a source of energy around what you're doing? So those are the two things that stuck out in my mind when it comes to this, and I'm seeing it in the companies that I talk to. I think there is this conversation going on, like, we don't know how to do this. We know we have to do this, but the guardrails and the tools that we've used, we just are starting from scratch in a lot of ways. Culture, Participation, and Human-Centered Innovation Beyond Top-Down Change[00:04:44] Robyn Bolton: Yeah, and what made me kind of laugh is that a lot of the advice in this article, you know, you read it and McKinsey's talking about you have to put identity shifts at the center of the change.You need to see new possibilities and let go of the past. And yet the article, when it gets into recommendations, is all about the leaders at the top. And you talk about how do you help people get through change? How do you use change to even energize people? And when change is thrust upon you, you're going to get exhausted.And so, the whole mechanism for change, for reinvention, I think personally, is you have to involve people in it. So that it's not being thrust upon them, which is a very, very different approach than the typical hierarchical top-down where McKinsey and a lot of other firms thrive. So, I'm very interested to see how this works.[00:05:40] Brian Ardinger: The last thing that I want to throw about this is, again, when you talk about reinvention, a lot of it comes down to understanding your customers and that. And it'll be interesting to see what McKinsey customers and clients want from the next change reinvention of what consulting is. What are the expectations, and I think that alone will be some interesting articles and things that come out in the future of what does it mean to have a consultant look at your stuff, and help you along that path.What are the things that the customer actually demands and needs when a lot of the tools are now easily accessible by the people that were paying millions of dollars before to have that delivered to them. [00:06:14] Robyn Bolton: The cobbler's children have no shoes, is a popular phrase for very good reasons, so we shall see.MVPs, Innovation Bias, and the IKEA Effect in Product Development[00:0...

On this week's Inside Outside Innovation, Robyn and Brian sit down to talk about the AI question that no one wants to answer, the power of a good value proposition, and why industry veterans are building tomorrow's billion-dollar startups. Let's get started.Inside Outside Innovation is the podcast to help innovation leaders navigate what's next. Each week we'll give you a front row seat into what it takes to grow and thrive in a world of hyper uncertainty and accelerating change. Join me, Brian Ardinger and Miles Zero's Robyn Bolton. As we discuss the latest tools, tactics, and trends for creating innovations with impact. Let's get started.Podcast Transcript with Brian Ardinger and Robyn BoltonOpening Reflections on Innovation and the Year Ahead[00:00:30] Brian Ardinger: Welcome to another episode of Inside Outside Innovation. I'm your host, Brian Ardinger, and we have Robyn Bolton, our co-host from Mile Zero. Welcome, Robyn.[00:00:53] Robyn Bolton: Thank you. Great to be here as always.[00:00:55] Brian Ardinger: We've got the number of different articles we're going to talk about today. As everyone knows, this podcast is about giving the real insights of what's going on in the world when it comes to innovation. What's going on in your world? [00:01:07] Robyn Bolton: It's funny, it's time of year, so kind of feels like everyone is both wrapping up and gearing up, trying to bring things to a close. So, we can all effortlessly and go on the holidays, but January we'll be here before you know it. And so people are already starting to think about what's going on with AI in 2026, and what does the new world of work look like?[00:01:29] Brian Ardinger: I'm looking forward to my inbox being filled with the best things that happened in 2025 and what to look forward to in 2026, and like kind of year-end wrap stuff that you get. It's interesting times, especially like on the investment front, you know, a lot of things slow down at the end of the year as people start planning for it. I kind of love and hate this time of the year from the standpoint of, gives you some time sometimes to do that stuff that you don't always have time to do and remap what you're going to do for 2026. [00:01:55] Robyn Bolton: Yes. And speaking of the emails, wrapping things up. Spotify's Yearend rap came out I think a couple days ago, so also getting a lot of those in the old inbox.[00:02:07] Brian Ardinger: Alright, well let's get into it. We've got a couple of articles to talk to today. The first one that we came upon was from KP Ready. It is called the AI Question. Nobody wants to Answer and KP does a good analysis. He basically says, is the juice worth the squeeze when it comes to AI. And I think a lot of people are asking that question right now.You know, as more and more enterprises, you're hearing about more and more experiments, more and more people using the technology, and you're getting conflicting results and feedback on is this really paying off. All the money that's being spent into ai, all the things that we're doing around it. Are we seeing the returns and when will we see the returns?You know, from my understanding and what I've seen, the question is not like, will we receive returns, but when and how do we get through this exploration phase so that we can be effective with using the dollars and the time and the resources around this to actually find the value that's created. And so let's start with that particular article. What was your thought on it? Is the Juice Worth the Squeeze? AI ROI and Experimentation[00:03:04] Robyn Bolton: There's always this level of uncertainty around new technologies of is the juice worth the squeeze? Are we gonna get ROI? When are we going to get ROI? Running lots of experiments, but it definitely seems like AI has kind of amplified that. I actually just wrote a blog post asking, like, did your AI strategy, was it developed by the underpants gnomes?And just in case there, we have listeners who don't know who the underpants gnomes are, they're from South Park, and basically their business plan is phase one, collect underpants, phase two, question mark, phase three profit. And it just seems like there are so many AI startups, companies, experts, consultants, et cetera out there who have become underpants gnomes. And kind of just have this like, hi, put it on top of everything, and profit, and no one's kind of slowing down to kind of like, well, do we even need AI? How do we need it? Like, what makes sense here? [00:04:04] Brian Ardinger: I think a lot of people are not necessarily thinking. They feel the pressure to start doing something with AI, and so they start immediately deploying and doing things without looking at, well, is this a particular area that really would benefit? Or could we create real value if we can get this right? And they oftentimes overlook some of the other kind of hidden costs when you talk about it, deploying technology that's new or different.And I think more importantly, how it affects the culture of the people deploying it. So, you've got the, you know, the data infrastructure costs, you've got the integration complexity, you've got the change management, ongoing maintenance, all these kind of hidden costs when you're dealing with a brand new technology that you don't necessarily know.And some of the things I saw in the article that were interesting and I've seen in real life too, is how you can kind of maybe think through this process of, you know, which particular project should we deploy and that? Don't be afraid to kill experiments quickly if you're not seeing ROI in a particular area, you know, maybe shelve that particular idea and focus on one that is showing some value to it.Don't focus on having to deploy it everywhere, all at once. Again, try to find the particular areas or the particular people that are more willing and able to make those steps. And then thinking about that, you have to build everything yourself.I think that's another place where I'm seeing a lot of brand new tools and folks out there that are trying things that have already built some things that maybe you can go out and purchase and buy and experiment rather than having to come up with your own team to do and make all the mistakes that they're probably already going through it and made that tool in the first place.[00:05:00] Robyn Bolton: So, all great advice, and I just want to underscore the people aspect is you have people who are going to use this, people that you hope will benefit from it. Some people who will be resistant, and so don't underestimate the people, the human. AI interaction and all of those dynamics as part of the rollout, the implementation, the change management, all of that.Sharp Value Propositions in the AI Era[00:05:30] Brian Ardinger: All right. The second article is from our good friend Ben Yoskovitz. He always puts out some great stuff. His article on his substack called Focus Chaos is called the Real Differentiator in the AI era, A sharp specific Value proposition. Ben talks about how a lot of folks are falling into particular traps of, again, thinking that, well, I've got AI so hot, I'm going to be deploying AI, and let's go out there and build a startup around that, and without thinking about, well...

On this week's episode of Inside Outside Innovation, Robyn and I talk about creating flywheel effects through customer obsession, the landscape of AI startups and what's real and what's not, and why workers don't trust AI. Let's get started.Inside Outside Innovation is the podcast to help innovation leaders navigate what's next. Each week we'll give you a front row seat into what it takes to grow and thrive in a world of hyper uncertainty and accelerating change. Join me, Brian Ardinger, and Miles Zero's Robyn Bolton as we discuss the latest tools, tactics, and trends for creating Innovations with Impact. Let's get started. Podcast Transcript with Brian Ardinger and Robyn BoltonStartup Ecosystem Building, Travel, and Early Observations[00:00:45] Brian Ardinger: Welcome to another episode of Inside Outside Innovation. I'm your host, Brian Ardinger, and with me, I have a co-host Robyn Bolton from Mile Zero. Welcome, Robyn.[00:00:50] Robyn Bolton: Thank you. Great to see you, Brian. [00:00:52] Brian Ardinger: Great to see you. Both you and I have been on the road quite a bit for the last couple weeks, so it's nice to actually say hello again and get back into the swing of things.I spent last week in Savannah, Georgia, talking about startup ecosystem building with the Savannah Harbor Innovation Partnership. And they're looking for new ways to spice up and kickstart a lot of the startup activity there. So it was quite interesting. As an innovator, I think it's always important to get out to other communities and see what's going on and share war stories and best practices and all that kind of fun stuff. So, it was fun to get out there.I give a shout out to what they're trying to do, trying to get the right people across all parts of their ecosystem together, whether it's founders or investors, university, and things along those lines. I think they're doing a good job of trying to kickstart a lot of stuff going on down there. I'm excited to see where they go to. [00:01:41] Robyn Bolton: Savannah is home to one of my favorite innovations, the Savannah Bananas. But yes, I was on the road too. I went a little further afield. I was in London for the Thinkers 50 Conference, which as you would imagine was extremely interesting, especially these days where everything is so volatile and uncertain everywhere across the globe.Thinkers 50 Takeaways and Early Reflections on UncertaintyAnd there was a lot of discussion around how now more than ever, is a time for courage and to be brave. There was a lot of discussion around what the future holds and several brave souls who just said, we don't know. It could be anything. One of my favorites was Daniel Pink. He said, basically, we're living in the era of Schrödinger's cat, that the future will be radically different and the same.All at the same time. And I'm like, okay, that sounds totally fair. So lots of really interesting ideas. Lots to think about, as you would expect from a conference's called Thinkers 50. Fascinating, fascinating conversations. [00:02:47] Brian Ardinger: Now we're going to open the box and what do we find when, when it's opened? The box is already open. You know, we're going down that path no matter what. And you know, it is kind of interesting. You, you're seeing a lot more bubble talk and things like that, but yet yesterday Nvidia had their quarterly announcements and $5 trillion valuation and blew out their estimates.And so it's like, well, it's not a bubble yet, or people are at least spending money NVIDIA's getting paid for this stuff. So, we'll, we'll see where it all shakes out. [00:03:12] Robyn Bolton: Yeah. Even if it's a bubble, I mean, we had an internet bubble back in the early part of the century and it didn't mean the internet went away. It just means we resorted ourselves. So even if AI is a bubble, I don't think it's going away. The Flywheel Effect, Customer Obsession, and Human-Centered Touchpoints[00:03:26] Brian Ardinger: Got three articles to talk about today. First one's called the Flywheel Effect: How Customer Obsession Creates Self-Reinforcing Advantages. This is from Wildfire Labs. They've been putting out some great content. The Flying Wheel Effect talks about building a startup isn't about growth hacking. It's really about how do you create customer experiences and these flywheels that generate self-reinforcing momentum because there's so many things out there taking aways attention and competing for dollars and mindset.So how do you build into your process different ways that you can create a flywheel to differentiate and build it up? So, one of the examples they talk about is chewy.com and how Chewy built into their experience the idea of when a person's pet passes away, they actually sent them flowers. And created an experience and a touchpoint as part of that to relate to their customers but also create a means for further conversations. And, you know, creating a positive experience with that brand and with that company. [00:04:27] Robyn Bolton: Yeah, I think it's so interesting and this idea of flywheel, it's one of the things mentioned in the article is the first step is creating an exceptional experience that creates emotional impact.Rising Customer Acquisition Costs and the Decline of Consumer TrustAnd you just think of our world right now, which is focused on cost cutting and on driving waste out of the system and on AI. The first step of creating this flywheel and this incredible loyalty and retention and word of mouth is so very human.You know, I think more and more we're going to start seeing that the more human you can be, you know, a handwritten card, sending flowers. That's something probably you could get AI to do, but it's a very human thing to acknowledge the grief that comes with the loss of a pet.The more human businesses can be, especially at key moments, that's where they're going to win. That's where they're going to differentiate themselves, in a world that is increasingly kind of more robotic and lean. [00:05:27] Brian Ardinger: The ironic thing is, startups have the exact same tools to create and reach customers as in the past. So it's being commoditized, the ability to actually grab a customer and have that first interaction.But the cost of customer acquisition, I think some of the statistics they had in the article talked about what costs, you know, a hundred dollars in 2019 now costs $160. You know, a startup that spends 61% of their new capital on customer acquisition up from 28% in 2018. So those are massive jumps and it's hard to make a business model work if you have that much to acquire the brand new customer and you can't keep them from churning or going somewhere else.Community Building, Trust Signals, and Real Startup DifferentiationAnd the other things, the idea of trust and how consumer brands are, the ability to trust brands is going down 73%, I think they said reported increased skepticism toward marketing claims, and that makes sense that you're seeing a lot of AI slop and a lot of things. Who do you trust anymore?And then finally, this idea of how do you build a community around your company, not just a product or service, word of mouth being so important. 78% of successful seed stage companies ...

On this week's episode of Inside Outside Innovation, Robyn and Brian sit down to talk about divergent thinking, peak college towns, and how as little as 250 documents can poison your AI. Let's get started.Inside Outside Innovation is the podcast to help innovation leaders navigate what's next. Each week, we'll give you a front row seat into what it takes to grow and thrive in a world of hyper-uncertainty and accelerating change. Join me, Brian Ardinger, and Miles Zero's Robyn Bolton as we discuss the latest tools, tactics, and trends for creating innovations with impact. Let's get started.[00:00:40] Brian Ardinger: Welcome to another episode of Inside Outside Innovation. I'm your host, Brian Ardinger, and I have my co-host, Robyn Bolton. Hello, Robyn.[00:00:47] Robyn Bolton: Hello, Brian. How are you today?[00:00:49] Brian Ardinger: I am doing well. Episode 338. We're excited to talk about innovation. I'm so glad you're on the journey with me here. [00:00:56] Robyn Bolton: Wow. 338. There should be a cake or streamers or something. [00:01:01] Brian Ardinger: Well, as usual, we've got a number of different articles and things that have caught our attention over the last couple days. So, I figure we'd dive in. And the first article I wanted to talk about today was for my friend Audrey Crane. She works at Design Map, and she posted on the Design Map blog an article on Divergent thinking.The article, I'll give it a little preface. So, it was looking at how you generate better ideas. Looked at a lot of things. First of all, looked at the research and the fact that there's a famous study by George Land and Beth Jarman looking at how five-year-olds were asked to come up with as many uses as possible for a paperclip, and nearly all of them, 98% could generate 200 or more ideas.They continued to do this study and looked into adulthood, and by the time the participants became adults, only 2%. Of adults could actually do the same thing and generate 200 ideas about a paperclip. It's pretty crazy, and it points to the fact that, while it's a super important skill to have, to be able to generate new ideas and think about different things. We are losing that ability as we get older. [00:02:08] Robyn Bolton: It was surprising and sad, but reminded me of two things. So, one is a book that came out a long time over a decade ago now, the Innovator's DNA. Where the authors looked at, you know, across hundreds, thousands of successful, both entrepreneurs and intrapreneurs, corporate innovators.And they found that the one thing that they had in common, or most had in common, was associative thinking, which. Is being able to put two very different ideas together to make something. And so divergent thinking actually reminded me a lot of the associative thinking. Because it just, your brain works in different ways.And the second thing is, you know, I teach at Massachusetts College of Art and Design, and in talking to my actual art colleagues there, one of the things that they say that's very consistent with what we say in innovation, is that your first idea, your first work, your first version is always the worst.They have different language around the idea of constraints and don't accept the first version, but that is a very common discipline within the art and design world. That it's just your first idea is going to be terrible, so you've got to push past it. And that was another aspect of this article that I found really interesting.[00:03:29] Brian Ardinger: It fits into startups as well. I was mentoring the new batch of NMotion companies are going to be announced here soon that are going through it. And did a little round table and met each of the teams and talked about what they're building and that. And the piece of advice I left them with is, you are here on day one with your idea. Be open to other ideas that happen when you start talking to customers and trying to understand if, if you really are onto something.Because a lot of times you get into an accelerator or you get some early traction and you think, okay, I've got all figured out. I got the solution. I'm just going to barge ahead. But you have to be open to that divergent thinking and different ways of doing things just so you don't necessarily leave opportunity on the table. [00:04:08] Robyn Bolton: It's the old adage, right? The fall in love with the problem, not the solution. And it's so easy to fall in love with the solution, but the divergent thinking article was a great reminder. Data-based reminder of why we've got to nurture the skill to go beyond that. [00:04:26] Brian Ardinger: Yeah, and I love the article too, because Audrey goes into a number of different techniques, brainstorming techniques, and that. Mm-hmm. So, I encourage people to check that out, to learn some tactical things about how you go about building up that divergent thinking muscle.The second article, this one made me a little nervous. It's from the Neuron Daily, which is a fantastic newsletter, and this one came out a couple weeks ago, but the title was Poisoning AI Models just got scarier. 250 documents is all it takes.And it goes on to look at some research that came out of Anthropic that looked at the number of documents that you could ingest into the LLM that would make it start spewing out nonsense and making it bad. The original thinking was, well, the more larger models that you have, you'd have to have more malicious documents put into it to make it go off the rails. What they found out was that's actually not the case.It takes a very small amount of wrong data to be put into it to actually make it start going the wrong direction. Actually, I think in the article talks about a 13 billion parameter model trained on 260 billion tokens, got backdoored with the same 250 documents as the 600 million model trained on just 12 billion tokens.And previous research assumed that attackers needed to control 0.1 of the training data. And so for larger models you need millions more documents. What actually happened is the math was 250 poison documents was, is only 0.00016 of the training data was enough to actually poison the model itself. Crazy stuff to think about it.When we think about. All the things that'll be putting into the web and everything else. It's a wonder why we're not seeing more and more hallucinations. I guess [00:06:07] Robyn Bolton: it was terrifying, and I'll be honest, the numbers went so over my head, but it had an analogy in there. It's like putting a teaspoon of a toxin into an Olympic sized pool. Yeah. And you'd kill all the swimmers. And I was like, uh oh. That is. So much more awful than I thought when you were just saying 250 documents.And you know, I think it just goes to pis. There's incredible promise with Gen AI and there's risks that we don't even know about. It's one of those things of use with caution.[00:06:39] Brian Ardinger: It'll be interesting to see what other research comes out to either debunk this or to fill up the back holes of this. But it's just something to keep in mind and follow the research and see what's actually going on out there. Because we are in brand new territory and the exploration mode, we need to be wary of the pitfalls that are ahead as well.[00:06:57] Robyn Bolton: