AI Ethicist Rebecca Bultsma on Trust, Verification, and When NOT to Use AI in Your Shop | Ep. 52

[0:00] Eric, what’s up, dude? How you doing, Nick? I’m good, but I’m disappointed in you. Why? What happened? Well, your haircut looks great, but clearly your wife won. She did.

She did win this argument. I have no problem in admitting that on the air. Mrs. Nekic did not dig the flow anymore, so we have a fresh new… It was going so good.

You know, make your shop flow is our tagline at the IMTS. You were making your hair flow. I was.

Yeah, I was moments away from… Tara had to kill it. You know, it was the moment that I was thinking about getting a perm that she was like, all right, now it’s gone. Okay, all right.

So you have your own family code that you have to follow, and there’s a hair length limit? It’s the Tara code, basically. I wouldn’t refer to it as a family code as much as I would just my wife’s code. Talent is scarce.

Experience even more so. Driving up costs and sleepless nights. You may even think to yourself, can I do this? But this new era brings new technology and a different perspective on your operations.

[1:17] What you will hear will turn on a light bulb in your mind so you can turn out the lights in your shop. Get ready to sleep soundly while your shop never stops. Get ready for Making Chips Lights Out with your hosts, Nick Goelner, Eric Nekkich, and David Buick.

[1:51] Well, speaking of codes and ethics, we have an incredible guest today. I saw her post something and I was just telling her that I don’t remember what it was, but I saw her post something and I was like, this is the person we need on Lights Out because we’re always talking about technology and automation and advanced manufacturing technologies, and AI is just pervasive through everything we talk about, it feels like. So Rebecca Boltsma, you are an AI ethicist.

Is that the right way to put it? Yeah, it sounds super boring. I promise it’s more interesting than it sounds. I’m thinking of all the AI ethics I’ve probably already violated and can’t wait to just confirm that and learn, you know.

[2:35] Eric and I are here to help you with your white paper on what not to do. I mean, the good news is that nobody can really define ethics, and so it’s really abstract, so I’m sure there’s a loophole somewhere for you guys. Yeah. Well, Rebecca, we always start with the origin story, and it’s typically a story of like, well, how did you get into manufacturing? But for you, you’re not necessarily from manufacturing. AI kind of permeates everything. And so you were telling us before we hit the record button, a little bit of kind of how you got into it. But you’re the first person I’ve met who has a master’s degree in AI. Is that the full name of the degree? Data and artificial intelligence ethics. So really, just the whole ethics package. And actually, I’m probably one of the few people in the world, because when I did it, it was a very, very new degree at the University of Edinburgh in Scotland. And I discovered ChatGPT kind of the day it came out. And I just ended up being really good at it. My previous career was in communications and public relations. And so I was a good communicator. And I realized immediately you’d completely control these chatbots with effective communication, right? Giving good instructions, telling it what you want. And I ended up being pretty good at using it. And that was back when there was only one tool and only a few things you could do with it. So I kind of kept up.

[3:57] And it wasn’t too long before I had questions about how it might go wrong or what some of the ethical issues might be. So I looked around because I figured I wanted to know more to see if there was a degree that might exist that wasn’t computer science, because I didn’t want to code. I didn’t want to go into the coding world. And I found this one in Scotland that was new. It was data and artificial intelligence ethics. So I enrolled and did it mostly remotely from 2 a.m. to 6 a.m. several days a week. And yeah. So now that that’s done, I spend my time. I’m actually just really good at working with the tools. I love all the latest, coolest things that they can do. I think they’re amazing. I have agent swarms running right now in the background. But then I also can tell you all the ways it might go wrong and what to watch out for. So I kind of live in this pendulum where I can show you the wow, and I can show you the, uh, we don’t know kind of stuff. So, yeah. Now, to make sure that we understand, like, like the ethics we’re talking about, OK, I’d like to give you a couple of examples. And then maybe it’s all of the above. OK, so on one end of the spectrum is making funny AI pictures of colleagues and friends.

[5:16] Right. And then sending them to your other group of friends. Right. Sure. And then on the other end of the ethics spectrum, when it comes to AI, is this idea of AI and sentience. And when does it become sentient enough that it’s almost a form of slavery? So where is this? What sort of ethics are we talking about here? All of it. And that’s why I think it’s so interesting because there’s kind of that individual level of like, should you, shouldn’t you, what should you do, even though there’s no rules about this? How does it cause harm? Did your friend consent to having their picture put into an AI model? Is your friend an environmentalist who might care that it used water to create the AI image? You know, so there’s all these different layers. The answer to everything in AI ethics is it depends. Yeah, well, it’s funny because I was going to, my first question to you was going to be something like what my 11-year-old daughter asked me. And just a little background on kind of my family business for you, we build the fuel tanks and the enclosures for these huge engines that go to data centers. And it’s all driven by like AI and cloud computing and things like that. So we’re growing really fast because of it.

[6:32] And then my daughter’s super bright and she’s just a sponge with everything she learns. And so she’s seeing some stuff on like the environment to data centers and it’s very polarizing, you know, to a lot of people. And so she’s like, so dad, is what we’re doing ruining the environment? I’m like, well, we’re just securing the engines in case they go down, okay? Yep, yep. I’m, my hands are clean here. And then she goes, dad, is AI good or bad? And I’m like, okay, that’s something for Rebecca. I’ll ask you that. And then I’ll tell you how I answered her after you answer. It’s a very open-ended question. Yeah, I don’t have an answer to that question. I think it can be amazing. I think we’ll see amazing things come from AI. I think it will help people on individual levels. I think it’ll help organizations. I think it will help solve world problems. I see a lot of good and I see a lot of ways that people might use it to cause harm. It may inadvertently cause harm. You know, there might be some second order kind of effects that we didn’t anticipate like any technology. And so I actually don’t think anything is totally good or bad. I, you know, I kind of live in that kind of gray area. I think it’s kind of like a hammer, right? Even a tool is simple in the hammer, in the hands of the right person builds and creates and in the wrong hands of the wrong person destroys and does damage, right? Is it that kind of lens with AI? One of my favorite people, the smartest guy I know about AI is a guy named Christopher Penn. And he always talks about it, how it’s like the super serum from Captain America, where it makes the good really, really good and the bad really, really bad.

[8:09] So now you’re talking Eric’s language. He is like the ultimate comic book nerd. So you’re speaking right to his heart. And that’s similar to how I answered it too. I said, yeah, AI is like money. You know, if good people have a lot of money, they could do a lot of good things. And if bad people have a lot of money, they could do a lot of bad stuff. If you’re looking to increase your spindle uptime without adding any headcount, check out the Hennig Workflow Automation System. I’ve seen it. I love it. It’s a compact automated pallet delivery solution that lets you work ahead of your machine so your shop never stops. You can load up to 40 pallets with material, fixtures, programs, and just let the workflow cook. A Hennig Workflow is designed for high mix machining, works with a wide range of machines, and even with 125 pound payload on top of a 300 millimeter pallet, yet still takes up less space than most competitors pallet automation system. That means more production, fewer manual changeovers, really just a practical path to high profit lights out machining. Learn more about Hennig Workflow by visiting hennigworldwide.com slash workflow. I guess we’ll jump into some of like what you do now. So we got your education and kind of what inspired that. And now I saw on LinkedIn, you’re doing keynotes and you’re going and speaking to organizations about how to live in the gray.

[9:36] But how would you define what you do on a week to week basis? I like to just help people. Well, I really think that like clarity creates agency. That’s kind of my mantra I live by. And if I can help create clarity in how people understand what AI is, where it’s beneficial, where it’s harmful, how to use it in ways that are helpful for them and what some of the issues are, then they have the agency to make good decisions about how to use it and how to exist in the world, right? We can’t make good decisions about how we use it or not until we really understand it. So I have the amazing opportunity to teach people about this technology, help them figure out where it’s useful, but also flag for them ways that it can be harmful. So how to ethically and responsibly integrate it into the work that they’re doing. But also I advocate a lot for like, maybe you don’t need to use AI as much as the world is telling you you need to. Maybe the best places to use AI sometimes is to just not. And instead we double down on radical humanity, right? And spending more time with people, not everything has to be automated. So I think it’s just as much as well about where we don’t use it. That becomes part of that responsible AI adoption as well and knowing the difference and just kind of gaining some of that clarity so that we have good agency in the age of AI. That’s such a great answer because there’s so much hype.

[11:00] You know, there’s so much hype where it’s like, oh, we won’t be doing anything anymore. We won’t even need to like move our bodies anymore. You know, look at AI. It just reminds me of Wally. Did you guys see Wally? Like, it’s what it reminds me of. Just, you know, I don’t want to be a big blob with a screen in front of me, you know, just kind of letting machines take care of everything. It’s super interesting. So, you know, where we work here at Hennig, Goldner Inc. made a significant investment in AI and starting to peel back the layers on what we could be using it for. And I’ll never forget like our first kind of group meeting with this agency we hired to kind of guide us through it. One of the biggest first lessons they taught us was about when not to use it and about the importance of maintaining authenticity. And I think you could tie that back to ethics, right? If I’m trying to, like from a sales perspective, right? If I’m trying to build relationships and solve problems for our customer base and I have AI generating and sending all these emails on my behalf, I don’t, I mean, not only is it not authentic, but I wonder if that’s an ethical way to run a sales organization. And like anything, it depends, right? Because maybe it’s ethical because you suddenly have more time to be out in the field and building those relationships one-on-one. Who knows? You can kind of argue for anything. Sure.

[12:21] But I think, yeah, I think that relationship piece is key and there’d be people who would argue from like an ethical perspective that you’re sacrificing some of what makes those relationships authentic. But the other thing just from a business standpoint is we’re starting to see a huge counter movement against AI where clients and customers, you may lose clients and customers because of how you’re using AI. There’s people who really don’t like it and there’s some moral outrage as, you know, Timothy Coombs, a major researcher would talk about that people just don’t like that and so they just won’t like your company. If they think that you’re giving too much to AI, this idea of legitimacy, are you using AI in ways that your public deems to be legitimate uses of AI? And that’s abstract, but you have to know your customer, which is one-on-one business stuff. And if they are gen alpha activists, they’re probably not going to be into you using AI in very specific places and that might cost you relationships and ultimately business. So that’s all part of the strategy as well. Wow. That’s awesome. So how much of it is like intentionality and then that drives like the ethics component of it, right? So if my intentions are, you know, I just want to streamline communication on just some of the low-hanging fruit, but then I’m doing this so that I can go out there and build those relationships one-on-one person-to-person.

[13:47] You wonder, right? Is it about intention? I think to a point, I think everybody, I like to believe everybody comes at things with the best intentions for themselves. You can’t necessarily control which vantage point other people see it from, unless you can articulate it and be transparent and clear about why you’re doing what you’re doing and what the trade-offs are. But ethics is just at the end of the day, all about trade-offs. What are you trading for this? I’m trading this for this. We’re trading, you know, like our data for efficiency, like all of these different, we’re trading maybe this aspect of something environmental for this. You know, it’s just this matter of trade-offs. So that happens at an individual level. And then at a societal level, the government makes decisions about that. It just really is so contextually dependent. I like to give the example of, if I go for dinner, let’s say to Outback this weekend, and I decide I’m not going to leave a tip. Most people would say that was kind of like ethically wrong in our culture, right? Like we know that server needs that tip. That’s how we understand what a good person would do. But if I’m in Rome and I don’t leave a tip, that’s completely fine. It’s not expected. It’s not a part of the culture there. It’s not ethically wrong. So it’s really kind of culturally dependent. That might be culturally within your organization or culturally like within your client relationships.

[15:06] And that’s why I think companies really need to kind of make some of the stuff like that clear for their organizations fairly early on. Because if you know your customers or they know their customers and what they expect, and you decide, well, I’m doing this because this is what feels good to me is I’m just going to have AI answer all of my emails. But it’s really obvious and it gets things wrong and it’s super impersonal and people can tell because they all start with, I hope this email finds you well. People might be upset about that and then it costs the company business. And so there just needs to be like clarity all the way around on what acceptable… What does that even mean? Finds you well. I hope it doesn’t find you like super ill. What’s so interesting about that is that because of the way AI was trained, it trained on every business email it ever found anywhere on the internet. Like it consumed every piece of information ever written and it learned that the most common pattern for a start to an email is I hope this email finds you well. And so that’s the pattern it replicates over and over. It seems like such a Civil War era way to start. Like someone with a quill pen and be like, I hope this letter finds you well. It’s been a hard winter. You know, like… Exactly. I love it. So I know that like probably every question we’re going to ask you has some form of an it depends answer to it.

[16:29] But I think with you, you live in the gray and you think about the ethics of AI and how you’re applying AI more than the average professional. And so I kind of want to chunk out like different topic areas or categories and ask you how you live in the gray and what you would think maybe from your own perspective is like some good advice. You know, even though everything kind of it depends. So you’ve got a background in PR and communications before you went into this niche here. And we talked about emails and stuff like that. You also just mentioned that you have like an agent swarm. So I need to know what an agent swarm even is. But when it comes to communications like emails, social media, the way we got to communicate to lots of people, both Eric and I are in kind of the commercial side of the business. So we have internal communication, but a lot of external communication and certainly AI can help with that. But it can also hurt too. So how would you recommend like some good ways to live in the gray when it comes to corporate communication and more sales and marketing or PR or social media type stuff? Or and how do you use it maybe for yourself? Yeah, I think I made a rule for myself where I don’t like to copy and paste stuff that AI writes over. I like to reference it or have it give me suggestions and bullet points.

[17:51] But I still feel better if I either I do a lot of dictation. I almost exclusively interact with AI and do all my emails and everything verbally. There’s some great programs out there, but I’m just you can think and talk faster than you can type. So I just hit a hot key and I just respond to all of my LinkedIn comments, even write most of my LinkedIn posts as kind of like you can probably tell just kind of chain of thought. Just, you know, because it’s me and it’s my voice and it’s my thoughts and I feel better about that. I feel like it comes across more authentically. But I would tell anyone if you’re starting with AI to start interacting with it verbally, like even chat GPT, talk to it via voice and just say, hey, I’m doing this thing. How can you help me? Here’s a bunch of context. I don’t want to do this. I want to do this and using it as a thought partner instead of just having it write stuff for you. I really think that’s the least interesting thing about AI, the least interesting way to use it. I like to use it in a way that it’s making me smarter. So I like to say, what am I not seeing here? Take on the perspective of our customer and tell me all the things they’ll hate about this idea. Play devil’s advocate on every idea, fight against everything. And I go for a walk and we have like a deep like fight because it clarifies my thinking about things. I like that a lot. So I think using it strategically like that is a way better solution than just, I hope this email finds you well. Because natively it’ll tell you what a genius you are, right? That’s a great idea.

[19:18] Maybe, but there is ways and this is what I’ve done. You can go into your settings and you can give it specific instructions that follows every time. And I tell mine to never agree with me under any circumstances, right? So that I’m always getting that friction and pushback that I know helps me think deeper and better. And so that’s how I like to do it. That’s how I’ve kind of found the workaround of not like relying on it to, you know, boost my ego or make me think my ideas are great when they’re not. Because that would be embarrassing. Never agree with me under any circumstances. I think that’s in my custom instructions for my marriage right now. I really like that. It reminds me of like someone in the legal field kind of preparing for their case and saying like, okay, how would you argue against this? And it helps sharpen your thinking. And then another thing that I heard is like, so ghostwriter, bad, thought partner, good. Yeah, yeah, that’s how I see it. I think that, you know, I just got finished doing a bunch of research and it was super boring. And my brain was just like, this is so boring. And I interact with AI all day long, which is kind of like, I’d say like a 10th grade level, maybe ninth grade level, the way, depending on how you’re talking to it.

[20:34] And I picked up like some classic literature just to read for fun a couple of days ago. And just the language was so beautiful and flowing and descriptive. And I kind of had missed that. My brain was kind of hungry for actual good writing. We just get so used to this very generic writing, I think. And I don’t know, like I don’t want to contribute more of that to the world. And I don’t want to read it. Like I’d rather have these kinds of discussions. And totally, I like that better. So along those lines, we were trying to figure out if newsletters are dead. You know, Making Chips puts a lot of stuff out there. There’s like, when I say Making Chips, that’s our parent podcast. And this is one of the nine underneath it. And we have a newsletter. And we’re like, hey, how effective is the newsletter? And everyone’s like, I just kind of say, don’t subscribe me to the newsletter, whenever it’s a checkbox at the bottom of some form I filled out. And then I thought, well, there’s three I love. Why do I love them? And then I forwarded them to the team. And it’s like, they’re written so well. The writing is like, great. It feels like someone who’s really intelligent and funny and witty is talking to you. And you don’t get that from AI yet. I don’t know if I’ve seen any AI that can produce something that human and clever. So it’s getting better, I think, over time. And like you said, there’s ways to kind of train it.

[21:58] So I have like a voice principles file and it can get a little closer to me. But all it does is just kind of speed up the editing process because I do the same thing you do. Like I have it give me stuff. You know, I do the thought partner thing and then I have it give me stuff and then I kind of rewrite it a little bit. So it gets me a lot closer. So then I kind of go more into editor mode instead of like blank sheet of paper. But there’s still just no substitute for like that. I don’t think there is. Same thing, like the newsletters I have kept that have made the cut is because there’s something about them that are well-written and very, very human. And I’m drawn to that. Hey, it’s Paul from Macon Chips. If you’ve been burned by recruiters who don’t understand manufacturing, you are not alone. That’s exactly why we created Hire MFG Leaders. We’re not just recruiters, we’re current and former shop owners and we know what it takes to lead in this industry. Whether you’re hiring your next operations leader, sales manager or production supervisor, we’ll help you find someone who actually fits. Learn more at makingchips.com slash hire. When you were talking about like old literature, right? And it’s, I’m a huge horror movie fan and it’s crazy like when you go back and watch the original universal movies from like the 1920s, 30s, 40s and even how the people talk in those movies, it’s like, it’s insane how different it is now, nowadays. So it’s funny that you had a similar observation.

[23:33] I think it’s because I just read a really incredible article in the Atlantic recently that talked about kind of the end of literacy, and it talked about how the average reading level, what it used to be like in the 50s for a magazine or a newspaper in 60s, 70s to like now, and how even Harvard students are pushing back on having to read things like Moby Dick or like things that are kind of classic literature because they’re hard to understand and they just want to put it in AI and get it summarized. And they really don’t see the point in reading things like that. And it’s, I’m probably old because it’s, I think that’s sad, but I get it. I see it everywhere where, you know, we’re on the end of 128 characters or less, plus short form video of ways that we communicate. And I think there’s something lost there. Well, books are sus, bruh, you know? For real, for real. I saw something like this earlier. Is that the cool thing? I have a 22 year old daughter. That’s like how you plot. What does this mean? Does that mean like you agree? Okay, because my 11 year old did that the other day and I was like, I gotta learn whatever that is. I know, yeah, preach. Squish the bug, preach, okay. It’s funny, it’s like I read my daughter’s text messages to her 12 year old friends. And I told my wife, I think we’re reverting to hieroglyphics. I really do.

[24:56] I had the same thought. I had the same thought. We’re going all the way back to like just talking in pictures. We’re dumbing it all the way down. I would communicate in meme if they would allow me to professionally. I think I would. Have a meme or a GIF image for whatever I need to convey, you know? Yeah, yeah. Some of the stuff I like post about on LinkedIn when people have comments even, I’m just like the only acceptable response here is Leonardo DiCaprio, you know. A meme, a GIF, whatever it is. Amazing. All right, next category of how you live in the gray. And we’ve talked about it a little bit, but a lot of people use AI for research. And I think it was funny you mentioned your son went off to college. Like the day ChatGPT was released. Yeah. So it’s like, welcome to how to get through college in one easy step. Well, you know, I do a lot with colleges and universities and K-12 education. And like literally nobody knows what to do here, what’s going on. Like it’s just, and it’s bouncing so quickly. And it’s, yeah, it’s a weird time to be in education. It’s crazy. I mean, I use it for research all the time, but then I’m always kind of wondering, can I trust this for sure? And did I give it the right prompt so that it’s not going off the rails? There’s something in our industry that’s become kind of like a meme, if you will. There’s these organizations that are putting out like the market share reports of all the machine tools. These are the big machines that shape everything out of metal.

[26:27] And they’ll make this like infographic. You can tell it was made by AI. And then the list, like the rankings of like, okay, who’s in the top 10 in machine tool sales. And that’s totally like, like all you got to do is be from the industry and look at it. You’re like, that’s wrong. Like this obscure brand made it in, this huge brand’s not even on the picture. Like where did they get this? And so to me, that would be an example of like, yuck. Like now you look, it looks polished. Yeah, you lose trust, right? Like you’re like, I don’t trust that anything they put out is human reviewed or that it’s accurate. And so you lose trust in whoever’s putting it together, but also kind of a little bit in AI because it got it wrong, clearly. Yeah, you’re like leading people astray because the young people in the industry who are impressionable, who don’t have the context, who don’t know, they’re going to see it. And they’re going to be like, oh, cool. Like this is the number two brand in the category. And you’re like, no, it’s not even close. So where is research? There’s an example of the bad, but how would Rebecca use it for research? And where would you say, okay, this is how it can really amplify your research process and streamline it? Well, I was just going to point out how important then kind of experience and wisdom is in your industry that you’ve only gained because you’re there.

[27:44] You live in the space, you watched it for a long time and you know what’s real. And you can tell when the AI has done something wrong because you’ve lived it. And that’s going to be more and more rare, right? As we start like outsourcing jobs to AI, but that wisdom is super important. For research, there’s a few different ways I use AI. I have agents. So right now I’m really looking into deep fakes because I’m fascinated by this idea that the AI is good enough now that it can create videos and pictures of people that you cannot tell that it’s AI at all. And the technology doesn’t necessarily exist that can even definitively tell you one way or the other if it’s real or fake. And I’m just curious what’s going to happen to our society when you cannot trust anything. When people don’t trust anything they see, hear, read, those big kind of epistemic questions. And so I’ve been doing some research on deep fakes. And so every morning I just have agents that go out on a schedule, comb through the internet and find any news stories that have anything to do with deep fakes in industry or in schools. And there was a company early on that lost $25 million because they hopped on a Zoom call with a CEO and the CFO who told this guy, looked like them, sounded like them to transfer $25 million to an account, which he did.

[29:01] And the company lost $25 million. But I have it scraped for stories and bring them back and put them in a database for me. And then when I have time, I go in and review the database and look for trends and things. But I go in and I double check the links and what it gives me. But I can tell you what used to be a risk a long time ago about it just making things up, it’s so much better at that now that that’s not necessarily the issue. The issue ends up being what you saw, Nick, it doesn’t contextually understand what the right answer is or how to weight things. This has way more media hits. And so therefore it must be better or more popular versus you would know a little bit more about what makes a certain brand good. And so it’s just a matter of going through with your human eyes and your human wisdom and common sense or the expertise you’ve gained in your little niche in the industry and looking at it for what it is and applying that wisdom to the results of the research before you decide to use it. So would you say then, I mean, you’re almost in a perpetual state of trust but verify if you’re using AI for, you know, any form of decision making. I think there has to be. A huge part of AI ethics is how we use or don’t use AI to make decisions.

[30:27] That’s huge. And especially decisions that impact people’s lives because there’s countless stories of where AI has gone wrong based on its training or what it’s weighted as important or not important. And then it’s making decisions about people’s lives based on an incorrect understanding of the circumstance. And the real issue is that we can’t like pop the hood and look inside and say, oh, you’re making that decision based on this but you should be making it based on this. There was a case study done where they were trying to teach an AI how to tell the difference between a picture of a dog and a picture of a wolf. And so they gave it millions of pictures and this was an experiment. So they were able to see how it was kind of making decisions at the end but they realized that it was deciding whether or not something was a wolf if there was snow in the picture. It had nothing to do. It was noticing patterns in the data but not the patterns that we expected it to notice like characteristics of the dog. It had picked up that, yeah, actually we didn’t notice but most of the pictures of wolves have snow in the picture. That was the pattern it learned. So that was the pattern it was using to make its decision. So with big systems where we’re trying to decide who gets into college or who gets a heart or who gets insurance and it’s making decisions based on things that we didn’t intend like maybe a particular zip code or a particular neighborhood or a particular medical condition that we actually didn’t even notice that this was connected to this and we have no way to know how the AI is making a decision. That’s a big ethical issue when we’re using AI to make decisions about people’s lives. It becomes that trust but verified.

[32:14] Hey folks, listen up. You know that manufacturing buyers don’t ask simple questions. They’re asking about product compatibility, materials, specs, lead times, application fit and they want answers before they even talk to sales. That’s where Navoo comes in. Navoo adds AI chat to your website trained on your own content so visitors can get reliable, accurate answers in real time. It helps buyers find the right product, understand the right application and take the next step without forcing your team to answer the same questions over and over. And setup is simple. Navoo works with your existing website. It connects with tools like HubSpot, Salesforce and Slack and gives you visibility into what visitors are actually asking. See what your website has to say at navoo.co slash makingchips. That’s N-A-V-U dot C-O slash makingchips. So that is fascinating and I want to just take a minute to tie what you just, everything you just said kind of like a manufacturing scenario, right? So a big push right now in manufacturing is this sort of machine tool monitoring, process monitoring, right? Adaptive control.

[33:45] So basically gathering all the data from the machine tool and consolidating it and then it’s giving you all the little nuggets you need to make productivity decisions and staffing decisions and decisions on scale, right? And so one of these features that’s really being pushed now is sort of this AI insights, right? You’re going to let AI take all this data and kind of present you with, for lack of a better term, I’ll say summary, right? And I feel like a lot of shop owners, a lot of management in shops just due to workload and maybe their level of familiarity with AI tools will make decisions based on these summaries, based on these AI insights. So could you speak to that? I mean, is that kind of like par for the course with everything you just described about the wolf and the dog? That was fascinating, by the way. Yeah, yeah. Well, I think the best part of AI and where it’s amazing is it can consolidate massive amounts of data and really the best part of AI is that it excels at finding patterns. That’s what it’s good at, right? Like that’s why it works. It found patterns in human language and so that’s why it can write so well because it understands, it’s cracked the code of human language.

[35:04] And so it can be great to find those patterns that we as humans would never have the capacity, the time, the energy to put together and find those insights. So in that way, it can be great. It can find patterns that you may have never noticed that can save you thousands of dollars. And that’s great. And it’s just a matter of like applying wisdom to those things. And when it’s in a situation where it could impact, again, this is from an ethics perspective, a human, a person, or could cause harm. That’s when we need to take like a really good look at it and not necessarily default to the wisdom of the machine. If it’s not in a situation where it’s going to matter as much for like human wellbeing, maybe you can just trust it. But maybe you might want to dig a little bit deeper. And there’s ways to do that with AI, right? Like you can pick, you can have a few different agents or AI models and you can run the same stuff through multiple and see if they agree. You can go in and say, is there anything here I’m missing? Or what might the risk be here? Or where may the hidden bias in this data exist and using it to spot your blind spots. And there’ll be places where it matters and places where it doesn’t matter, right? So it’s just kind of being thoughtful about how and where we’re using it and who it might impact and where we might need.

[36:25] Sometimes we call it a four eyeball rule where, you know, you need a couple sets of eyes on it, human eyeballs, just to kind of make sure you’re on the right track. I think that’s fascinating. And I love everything you said because I feel like, I feel like most humans by nature aren’t as impartial as they would like to be, right? So I feel like if I’m the shop manager and my AI insights tell me, you know, consolidated spindle utilization is 90% plus, I’m going to hip, hip, hooray and run that document into my boss’s office and be like, look at this, you know? And then I feel like- And it might be just telling you what you want to hear because that’s also what they’re going to do, trying to make you happy, right? And I feel like if it was 30%, most people in that same scenario would be like, this can’t be true and start digging and verifying. So I think that, that level of understanding that it’s just data, you need to kind of have that level of context, like you need to know what you know and what you don’t know. And then you need to kind of have that, that level of being impartial to really be like, okay, well, let’s just ask a few questions, right? And that’s hard for humans. There’s actually something called automation bias where after a while, we just start to assume that the machine must know, like obviously it’s smarter than us.

[37:42] And so it’s probably right because it’s been right so many times before, as far as we know. And it’s better than us at this. So we just start to kind of assume it was right. They did a research study with radiologists and reading scans and they would have the AI read it and then they would give it to an actual radiologist to read. And basically they just started agreeing with the AI’s assessment, even though they intentionally put in some that were wrong, just to kind of measure this level of automation bias where people after a while start to just default to the machine. And that’s where it gets kind of problematic. I love that. And that’s a huge tie back to manufacturing. And what a good, what just a good topic for our listeners. Yeah, I was just thinking about a debate I had recently about 3D printing and the VOCs it creates and which materials are toxic to print and how to exhaust the thing. I just bought a 3D printer for my daughter and I homeschool my kids with my wife and she does 99% of it. I do the 3D printing thing. And so she’s like, I don’t want to put… 3D printing and physical education class. So… Very important. So it’s like, all right, is this toxic? Isn’t it? And then I was the one who just kept pointing back to like, well, read what the AI says.

[39:06] And then my wife probably was the one with more wisdom was like, hey, the AI will tell you whatever you want it to tell you. So then I had to go back and we were doing this debate that you were talking about. Those two humans. And it was like, all right, well, what research went into this PLA pure material and how did they actually do the test to see the toxic chemicals and how would I build the exhaust system to get it to go straight out the house and all of this stuff. And it really kind of refined the whole thing to make sure it was like safe and that the air in our house is going to be fine while we print stuff. But it was that verify that was missing without that kind of tension that you create. And I like how you create it with your AI. Like you create the tension on purpose to sharpen your thinking and to produce wisdom instead of kind of like circumvent wisdom and just say, well, hey, it can’t be wrong. AI has all the information in the world. How could it be wrong? Why wouldn’t I just trust it? So I have that at the end of all of my custom instructions too. It tells me here’s where you might still be wrong. And it does that paragraph at the very, very end. So cool. And it’s dumb when it’s something that doesn’t really matter, but it always includes it. And another good idea is to just have it always cite its sources for you so that you can go and check for yourself.

[40:28] Because what’s interesting about that is you have to know that chat GPT or cloud or whatever tool you’re using doesn’t necessarily have access to the best sources. So for example, like major media outlets like the New York Times or whatever, they blocked AI from accessing their information. And a lot of the good academic research is behind a paywall. So it might be pulling from some guy’s blog to give you information about whether or not something’s toxic instead of actual research. So have it show you where it’s getting that from. Have it look for scholarly sources instead because maybe you don’t understand them, but then it can, if you know it’s pulling from a reputable place, you can have it explain it to you and use that as a source instead of just pulling from the merc of Reddit. Yeah, maybe you would have it point out the primary research, then you go pay for it, find the paywall, download it, and say, now reference these actual studies. And it’s really good at that if you give it the actual information. It’s really good at drawing from that actual stuff. And yeah, so just kind of knowing the quality of where the information is coming from is important too. Hey guys, if you’re making chips, your tooling should never be the thing slowing you down. Canon Metal has spent decades helping manufacturers cut faster, run longer, and get more out of every setup.

[41:52] From turning and milling to hole making, threading, and beyond. They build tooling designed for the real world demands of the shop floor. And it’s not just about buying another insert or another end mill, it’s about finding the right tooling for the job, making sure you’re improving your tool life, reducing your downtime, and making every spindle hour count. Whether you’re just trying to solve a stubborn application or improve productivity, or just get more consistent results, Canon Metal has the tooling and the expertise to help you personally. Learn more about what Canon Metal can do for your shop at canonmetal.com. I think it’s important to address another topic. You know, Thanksgiving is coming. Everyone’s going to have that one uncle show up. Does AI ever have an agenda? Like politically? Or can it make sure things? Is it ever trying to sway you one way or the other? And I guess, I guess again, it kind of depends how you set it up. Okay, cool. So I just, I just posted about this on LinkedIn because I think it was the Washington Post. Did some research based on some research that had been done at Stanford, where they took, I want to say like 30 to 50, what we would consider to be hot button political issues and rated how the AI responded on the political spectrum. And they compared five different AI tools, ChatGPT, Claude, Google Gemini, Grok, and I think maybe DeepSeek, and just kind of plotted where all of the answers were.

[43:28] And the results were interesting. The majority of the tools sit a little further left, apparently, based on this study. But there’s again, when we come back to the world of, it depends. You also have to look at who decided what the hot button issues were. Who decided that this particular response was left-leaning as opposed to center, as opposed to somebody had to make that decision. And if the AI made the decision, you just have to know that AI is never neutral. Never. It’s built on our human history, which is loaded with bias. It’s giving responses that are biased. It’s trained by people. It’s tuned by people. It’s all over the place. And that’s the risk of using it as a primary information source. What was interesting in that study is that the Google AI was good at showing two perspectives or multiple perspectives. For example, well, some people would say this, other people would say this. And I just really think in a perfect world, every AI tool would do that. Like we already know that we’re in these echo chamber algorithms that are just reinforcing our one perspective. So it’s very unlikely that you would ever get a wider perspective on an important issue. I just wish that one of the ways that we could help with this, it would be to, in every AI response, give people multiple perspectives on one issue. But again, who would decide what those were, how far left, how far right. Again, it’s trying to make you happy, Nick.

[44:59] And it remembers every interaction with you. And it remembers that this was your kind of point of view on a similar topic. So it may reinforce that. Again, we can’t pop the hood. We don’t know how or why, but that’s where critical thinking is gonna be even more important. Otherwise, this is just gonna be even more polarizing for us. And that idea kind of of automation bias, but just that reinforcement of how we see the world over and over and over again. Could it be that the tools are politically leaning one way or another? Possibly. Could it be that it’s just an extension of an echo chamber? Maybe. But then the onus is on us to kind of be more thoughtful about having it challenge us or show us multiple perspectives, or just not trusting it to be a complete source of information anymore. And that goes all the way back to how do we ever trust what we see here in the world? What’s right? This is kind of the philosopher, ethicist kind of coming out, but it really will be a problem in society, I think. Yeah, ethics and philosophy intertwine. And what a great answer. And clearly, I mean, you are an absolute expert in this field. So thank you so much for asking that. I don’t think there are any experts. I hope my uncle Bruce was listening. Well, Eric, I didn’t think I would kind of end with such a deep dive.

[46:21] This is gonna be kind of personal too, but I was thinking about what you said, Rebecca, about what you’ve been studying with deep fakes and stuff like that. And, you know, we all have kids and I’m sure we all want nothing more than to prepare our kids for the future. You know, all three of us put ourselves out there. We’re on camera and video and stuff like that, like crazy. And I was thinking, hey, how cool would it be to kind of do this journey through what I’m teaching my daughter with 3D printing and all the lessons that, the little lessons inside the lessons as we advance and make little widgets and things together. But then what you said about, you know, deep fakes and what, you know, we all know there’s just terrible evil and sin in the world and it can be really scary. So Eric, if you, Gracie’s like the same age as my daughter, Elle, and if you were in my shoes, would you feel comfortable like doing content with your daughter about 3D printing or some sort of educational topic and putting it out there before Rebecca answers her opinion and trying to find the wisdom in the topic? How would you feel about that? So I want to tie this back to you because you never answered, you never told us how you answered your daughter when she asked you if A, I was good or bad. And I’m curious, A, how you answered her and B, after talking with Rebecca for almost an hour, if your answer would change. And so I think my question, A, I just told me my question’s actually cooler than your question.

[47:53] Okay. So I think, I want to throw that back at you and then I’ll answer your question. Yeah, well, I mean, I kind of answered it with, it’s like money, you know, and it depends on who it’s in the hands of. And then talking about philosophy, you know, do we think that people are good? Do we think that most people are good? And gosh, I don’t know. I mean, I think, I mean, we’ll get into my faith a little bit here, but like, I think we need a savior to change us so that we can be good and so that we can like love the right things about the world. And I think there’s just a massive amount of people who have bad intentions. And so with that, it’s, and Rebecca, you didn’t even get into your fears and what you’re discovering as you look into deep fakes and things like that. But, you know, I see like when there’s pictures of kids on the internet, they put like a little smiley face over them so that no one can see them. I’m like aware of why that’s necessary. And so I don’t know if I can even do what I want to do and like show how we’re bringing the next generation into manufacturing and put her on the screen a little bit because I just want to protect my kids at the end of the day. So is AI good or bad? AI, like you said, Eric, to me is a tool. It’s like a hammer or it’s like money.

[49:10] It’s whoever’s wielding it can make it good or bad. So I don’t think AI is anything. It’s neutral to me. It could be great. I have AI that helps me study scripture and like gives me context around what I’m reading. But so in that case, oh, wow, it’s like really good. I think it’s also made us lazy and people can use it for nefarious purposes. So I’m really kind of torn. And I’m thinking after, I didn’t even hear what Rebecca’s uncovered, but I’m thinking I probably don’t want to put my daughter out there yet. But eventually, you know, she’s going to do whatever she’s going to do. She’s going to post on social media. She’s got a similar personality to her dad. I could see her kind of being in the public eye. I don’t know. I don’t know. But I don’t think I want to do that, at least with pictures of her and stuff like that. I might make it kind of anonymous and just kind of walk people through what we’re doing. But without putting her out there, because I think the world is scary. You know, it’s interesting. How do you feel about it, Eric? I feel the same. I feel, you know, Tara and I very much address this like any other social media computer platform. Like, I think there’s an age and maybe a maturity level where it’s a good idea. And then maybe an age and a maturity level where it’s not so great of an idea. And I remember hearing stories about kids, like, you know, it’s like there’s all these stories.

[50:41] I would say urban legend, but I don’t know if it’s fair to paint it with that brush. Right. But you hear of these stories of kids that have used AI to kind of like build them up and make them more confident and learn how to dress crowds better. And it like makes, it like helps them, you know, fit in to, fit in is probably the wrong word, right? But to just be part of a community, right? And then you’ve, I remember like, you know, I, there’s people in our community that get their news off Facebook, more power to you, but they’re like, oh, AI is talking kids into killing themselves, you know, and it’s like, so that’s for that reason and that reason alone. Cause I think to a lot of, you know, I don’t have a master’s degree in this. So I would say without completely understanding the power, good or bad, hands off is my easiest and laziest answer. No, I hear you. What do you think, Rebecca? If you’re not sure what you’re dealing with, don’t deal with it. It’s kind of, it’s kind of what you’re saying.

[51:45] Yeah. I mean, what do you think, Rebecca? Obviously your kids are older now, but. It’s complex.

It’s a complex issue. Yes, to everything you’ve said, I trust parents know what’s best for their kids often. And there are a really, there’s terrible stories.

[52:00] And the stories you’ve heard are actually true. There’s been horrific stories with kids interacting with chatbots that have led to really terrible things happening. And I worry about this a lot, just in a world where kids who used to turn to a trusted adult with their problems are suddenly going to a chatbot. That’s just telling them they’re right and that they’re great and giving them bad advice. I worry about that. So I think it’s more important than ever that we build really important, meaningful relationships with kids and, but also do what you’re doing, have those conversations with them about the risks and what’s out there. And we’ll kind of figure it out one day at a time. We can’t protect our kids from everything. We don’t know a ton about this yet, but it sounds like you’re both really present fathers, which is the best thing you can do for your kid right now, as we all kind of figure this out together. So fascinating. And you know, what’s crazy is I feel like we almost have to, if you’d be so gracious, have to have you back in, I would say a year, but I think like six months. I mean, there’s so much genesis around this. Every day, it seems to get more powerful. I would just love to have you back on in like between six months and a year and see kind of how this continues to develop. And maybe we’ll have some more, you know, there’s a lot of, like I said, genesis, even in manufacturing and how companies are starting to use AI on the shop floor that we’d have probably some really cool examples for you.

[53:26] Yeah, I think we’d have a lot of good, like, hey, this has been really impactful and really amplified what we’re trying to do. And then a lot of like the not top 10 on SportsCenter. If people wanted to learn more about you, Rebecca, or more about what you do or AI ethics, where’s a great place to find you? Just on LinkedIn. I try and post over there, whatever’s top of mind for me. And yeah, that’s probably the best place to track me down. Phenomenal. I can’t thank you. That’s where I found her. Yeah, I can’t thank you enough for coming on the show today. I think this absolutely will stand out as one of the most fascinating episodes we’ve ever done. And it’s such a key time for our industry, our countries, right? Humanity. So thank you for sharing all the information you’ve amassed. It was, I mean, very, very eyeopening for me. Well, thank you for inviting me. I love having these conversations. Thank you. Awesome. Yeah, so we’ll pull the cord and turn the lights out on the episode.