Ep 127: Restoring Trust by Advancing Human-Positive AI with KVJ

[00:00:09] Hey everyone, welcome back to Elevate Your AIQ, your go-to source for insightful conversations on human-centric AI readiness, talent transformation, and the future of work. Today I'm excited to be joined by Kathryn Von Jan, better known as KVJ, CEO and co-founder of Tough Day, an AI workplace advisor that gives employees and managers confidential, personalized coaching to navigate real workplace challenges.

[00:00:32] KVJ is a veteran technology executive whose career includes Lotus Development Corporation and IBM, where despite both of us being involved in social technologies and the future of work, somehow we never met. And then Salesforce, where she was the chief strategy officer for their innovation division before co-founding Tough Day.

[00:00:50] KVJ, we get into what it actually takes to build AI that reflects an organization's values, why ethical review and red teaming matter just as much as the innovation itself, how thoughtful research design can uncover what great management really looks like. KVJ, if you care about building AI that earns trust rather than erodes it, this one's for you, so go give it a listen. I always learn a lot from my conversations with KVJ, and I love this open dialogue with her, I know you will too.

[00:01:16] Thanks, as always, for spending time with us on Elevate Your AIQ. Hey, everyone. Welcome back to another episode of Elevate Your AIQ. I am your host, Bob Pulver, and with me today, I am excited to talk to Kathryn Von Jan, better known as KVJ. How are you today? Hi, Bob. I'm well. Thanks for having me. Absolutely. How are you? My pleasure. I'm doing all right. Doing all right. We're just talking about the Westchester County weather and hoping it drops below 90 one of these days.

[00:01:46] I think it's hot all over the country. I just talked to somebody in Boulder, Colorado, who said it was in the 90s as well, so can't seem to escape. We got a lot to talk about, but I wanted to start with a bit about your background. You can go as far back as you want, but I'm particularly curious about your experiences at Salesforce and some of the innovative work that you were doing there. So, yeah, I'll let you kind of kick things off. Well, I think it's relevant to go a little farther back. Sure, sure.

[00:02:15] I've been working in tech a long time, since the 90s, and started my career at Lotus. And I kind of, you know, I guess it's helpful context. I got into tech not because I thought I would work in tech. I was interested in solving big world problems. My undergraduate was in business and international affairs. But I was really fascinated by culture and how people come together to tackle big problems.

[00:02:45] And went back for a master's, really focused in anthropology and understanding the power, how power is shared and utilized in tribes. And I was working, it was sort of like this combination of theology, anthropology, women's studies, and leadership.

[00:03:08] And trying to figure out, like, what's different about tribes where everyone has a voice and it's less hierarchical and all of those things. So, anyway, I was doing this master's program and my advisor said, you know, you really ought to go into the tribe at work and understand this firsthand. Go be an anthropologist. And I went to look at the top places for women to work at the time.

[00:03:37] And that was Lotus Development in Cambridge. And I was like, I'll just go. I'll try to go work there. It seems like I might observe some things that are relevant. And I ended up working for Irene Greif, who is the mother of collaborative technology. First woman to get a Ph.D. in CS from MIT. An incredible organization, really founded by Mitch Kapoor. Incredible values.

[00:04:03] And I was really lucky to start my career there because the culture was so empowering and designed to take care of people and help them to do their best work. And my role there, I started as her assistant and then got hooked and started to apply technology to knowledge management and storytelling in the tribe at work.

[00:04:26] And fell in love with this idea that technology can enable and empower us and help us thrive in the world of work and build better businesses. So that's how I got into this space. And whether it was at Lotus and then IBM or doing my own thing, working in mobile, social, AI since 2012.

[00:04:50] I joined Salesforce in 2013 and left about three years ago to start my own thing, tough day. So there's a lot of context. But I say all of that because the thin red line throughout my career has been culture and tech and innovation in the world of work. And I care deeply about it. So I can go, you know, into lots of things at Salesforce or in any of that. But what did you have in mind? Yeah.

[00:05:20] Well, first of all, I know Irene. Oh, I love that. Lotus. I started at IBM about three months before IBM acquired Lotus. And then it took me through a circuitous route. I wound up working on some projects and getting involved with basically connecting IBM's biggest clients to IBM Research.

[00:05:47] So I spent a lot of time working in the social analytics and social technologies space when Irene had sort of moved to run, basically run the Cambridge Lab.

[00:05:59] And so between the Cambridge Lab, the Hawthorne and Yorktown Lab here in Westchester County, Armonk, of course, and then the Almondon Lab out in San Jose, and even the Haifa Lab in Israel, there was so many amazing technologies, many of which never actually saw light of day or at least not outside the walls of Big Blue.

[00:06:25] But all of those things had an enormous impact on what I wanted to do throughout my own career at the intersection of people, technology, processes. I spent a lot of time around when you left to go to Salesforce researching concepts of collective intelligence, collaborative decision making.

[00:06:48] That's how I got introduced to prediction markets way before the current fad of prediction markets and all the trouble that people are getting into with those. But there's a lot of use cases for those, not to go completely off topic. But prediction markets have a lot of value inside a company because inside information is good.

[00:07:10] It helps you uncover, you know, knowledge and insight that would previously go sort of untapped so you can make more informed decisions. But anyway, a lot of the technology, a lot of the concepts that I understand and are now sort of resurfacing in the age of org, you know, redesign and humans plus AI, you know, synergy and collaboration.

[00:07:33] All of that work from, you know, 10, 15 years ago is as relevant now as ever. So it's just amazing to hear someone else who, you know, was interested and influenced by, you know, some of those early, you know, IBM experiences and some of the, you know, fellows and both, you know, technical and business, you know, leaders that were really showing us the art of the possible.

[00:08:02] Absolutely. The researchers there are incredible. And it's an exciting place to start. And then when you go to other places and you start to see, like, it doesn't all work that way. Like that was also like an eye opener to me. Well, Salesforce, though, was, I feel like that was probably more in IBM's realm of, you know, invention and innovation perhaps as well.

[00:08:32] You know, software as a service became a thing, but I'll let you frame it differently. It's actually a funny story how I got to Salesforce and why I went there. It does have a bit of that IBM flair. But going back to 2013, I had a startup called Rad Matter at the time, and it was a virtual internship platform. And we were bootstrapped and profitable.

[00:08:55] And we built basically a platform for students from their freshman year on to do virtual internships in the form of games. We called them winternships. And they could do different missions and kind of level up. And they would level up ultimately to a paid internship, a virtual paid internship, and then an on-site paid internship and ultimately into their first job.

[00:09:21] And the idea was, like, students need to understand the world of work and apply a lot of what they're learning in school and get real-world experience and figure out by the time they graduate, like, what do they actually want to do? What are they actually good at? All of those things. So I was doing that. I was also simultaneously, I was like a beta tester of LinkedIn and way, you know, way back when. And I had this incredible network. And these two stories come together.

[00:09:51] So working on this platform, we are partnering with Manpower Group. We were also in talks with LinkedIn at the time about how we could work together. And one day I woke up and my LinkedIn profile was gone. And I was blown away. It was like, I got an email that was like, you've been discommunicated. Like, you're no longer in this community. And then I sent some notes to some friends. And I was like, can you look me up? Like, am I really gone? And they were like, yeah, we can't find you.

[00:10:21] And so, like, they wrote to LinkedIn and asked the question, like, why am I not there? And they were like, yeah, she's broken the rules and can't be here. Something to that effect. Whatever they said. And I was like, this is crazy. The reality is, like, our platform was trying to become, like, the student version of LinkedIn. Like, the early phase, the early stages of LinkedIn. And then, you know, ultimately, we thought we could get acquired by LinkedIn or something like that.

[00:10:50] But something weird had happened where they found out, this is what I heard through my research. Like, we partnered with Manpower Group. We started a huge campaign to get all of their clients on board our platform for young talent. And then, like, as soon as we announced that, like, a week later, my profile was gone. So, anyway, that was weird. And I put up a fake profile. It was just, like, rad cat at super cool startup.

[00:11:20] My title was benevolent troublemaker. No, yeah, it was, like, benevolent troublemaker. And based on that and no connections and no skills and nothing, based on that, Salesforce reached out to me about an innovation job. And I was like, oh, my God. This is, like, the – this is why LinkedIn should not exist. Like, this is a mess.

[00:11:42] Like, so I went to my friends at Manpower Group and told them, like, this funny story about getting outreach to this profile of nothingness. And they were like, oh, my gosh. Like, we know the head of innovation at Salesforce. You have to meet him and tell him this story. I met him. We totally hit it off.

[00:12:02] And I decided to shut down the company and join Salesforce based on pretty much on that conversation and on reading Mark's book into the cloud and understanding his values and how he thought about the opportunity in the market and all of the really awesome sort of, you know, deviant ways he broke into the market initially. Like, there were such great stories in that.

[00:12:27] So I fell in love with the whole culture, the entrepreneurialism of a large company and decided to join. And it was 12,000 people at the time. So it was, you know, still early. A fraction of what they have now. Yeah. It was like 80,000 when I left. Yeah. That's a really funny story. It's funny also because LinkedIn is still, you know, fighting off everybody who tries to do anything, you know, that they deem, you know, unscrupulous, right?

[00:12:55] Like, you know, throttling your usage or, you know, trying to kick you out if you're, you know, scraping data. I mean, we could spend the entire podcast just on that. Yeah. Unfortunately, it is a reality in our world. Yeah. Yeah, for sure. Well, I mean, just to sort of go back a little bit into IBM and Salesforce's of the world versus sort of other.

[00:13:21] I mean, I left after 22 years and went to a big media company and they were trying to build themselves as a media technology company. Company today, maybe that that moniker would hold. But back then it was like all of those all of those toys that I had at IBM were gone. I mean, the entire toy box was gone. That's the other interesting thing about working at IBM and Lotus. I know exactly what you mean.

[00:13:50] Like we had the Internet when nobody had the Internet. We had like, you know, instant messenger and all of those things. Because laptops, I remember having to convince another leader in another company I worked for, like, we need laptops instead of desktops. And so we have to be out in the world and meet customers and things. But the moving back to Salesforce, for me, it wasn't like I always had all the tools. Like I brought the tools wherever I went after Lotus and IBM.

[00:14:19] And I was always working in tech in some way, even in the mobile space. It was like I was leading an innovation team and kind of at the cutting edge of working on, like, 5G concepts and the 3G timeframe when 3G was brand new. The first UMTS license was out there. So I've always kind of worked at the frontier of these technologies.

[00:14:42] And I haven't really had to work, other than that one place where they were, like, on desktops instead of laptops, haven't had to deal with not having technology. But, you know, I've worked with a lot of clients that don't have all of the tools and the bells and whistles and had to help enable them or bring them along and help them understand the value of them.

[00:15:06] And I guess because I was playing with a lot of different things that were, you know, very early stage. I mean, I wouldn't even call them pilots. They were just experiments, right? And so some of the stuff was not, it wasn't like technology for technology's sake.

[00:15:25] It was even things around, like, if you think about the social network analysis space and organizational network analysis, expertise discovery, you know, being able to visualize how knowledge is sort of shared across a network within an organization. Not just a formal team, but, you know, informal networks and online communities and things like that. Not all extensions of the Lotus platform that you probably saw right before you left.

[00:15:52] But some of it was just some of the things we talk about now, right? Like, how can we make individuals and teams more productive? How do we improve productivity by having this knowledge shared in shared spaces? And so, you know, they were sort of quaint concepts back then, sometimes looking for a business case. And now it's just like these are, you can't really live without them.

[00:16:18] I think there's, and even back then, like, there's the core infrastructure to enable sharing. And then there's, like, the application. And I've always been application-centric. Like, what's the problem we're trying to solve and then how to solve it? So even in the world of knowledge management, it wasn't, like, knowledge. And it wasn't really about communities. It was coalitions.

[00:16:44] Like, a community of knowledge can have all the knowledge and sit together and, like, pontificate. But a coalition is designed to go solve something. So how you build technology to enable people to go actually, yes, get the knowledge that they need. But the workflows, the processes, the things that are going to enable them to do that work differently. Like, that's the point of knowledge management is putting it into action and getting an outcome from it.

[00:17:13] And I think that's even today, you know, where a lot of organizations are struggling with AI. It's like, okay, I can have the AI. And, like, a lot of people are still having what I call a 2012 conversation with AI. It's like, can I write a job description with AI? It's like, oh, my God. Like, yes. And you could have done that before, too. But the real magic is when you think about redesigning and restructuring.

[00:17:43] Think about the outcome, the job to be done. Like, what do you need to get to? And what is the best way to achieve that today? Leveraging your people and this technology together. And, yes, a whole lot more knowledge that we have access to than ever before. But it still, for me, comes back to what are we solving and how can we solve it? And putting those workflows and the new technology in place instead of just putting old technology or new technology on old process and problems.

[00:18:13] It comes up a lot in the conversations when we talk about what should AI do? Not what AI can do, but what it should do. Yes. And the impact on how human, individual sort of human critical thinking, but also the team effectiveness, you know, moving past these individual, you know, metrics and measures of productivity or success. You know, we generally work in teams.

[00:18:42] We're working across hopefully breaking down, you know, organizational silos and things like that. And so I think people are getting really tripped up for good reason of like where do I draw the line? Because people may have different sort of thresholds to say, you know, I'm comfortable letting AI do this because of X, Y, and Z. And those could be sort of human centricity kinds of factors or it could be governance kinds of factors.

[00:19:12] I mean, there's so many, you know, this is a pretty expansive, you know, mind map. Yeah. I mean, I was, I had a conversation with a leader this morning that revealed, so they were using AI for skills mapping. It's your point. Like we want to find the right people in the organization for any given role, project, what have you. Noble, noble cause and would be great to have that like instant matching and give people opportunities. Like you can make the case for that.

[00:19:40] But 60% of their people opted out. They gave the opportunity, they explained it to the company and told them what the purpose was. And 60% actively said no. And then it's shifted into more of a research project to understand why. But it comes down to trust. Like the reality is a lot of people don't understand what are you really going to do with that data.

[00:20:08] They don't really, and we know from the Edelman Trust Reports, they don't generally trust business. They don't generally trust these AI companies. And so they're like, I'm just going to say no unless I can understand, you know, until you build the trust. And it's fair, right?

[00:20:27] Like there are reasons not to trust certain companies and certain, you know, whether that's the AI, your own company, your employer or the AI companies that they're deploying, the solutions that they're deploying. And, I mean, I'll just give you an example. Like I met a couple weeks ago a company. Someone said, you know, we should meet and consider working together. I met them. And they're like straight up workforce surveillance.

[00:20:55] And they've created all of these agents. Each agent does something very, very specific. Goes, let's say, into your email, your teams, your Slack, your projects, you know, your HRIS everywhere in the company, your finance system. And like one will be looking for fraud. One will be looking for are you stealing from the company? One will be looking for toxicity.

[00:21:22] Like there's all these things that they're looking for and that they're trying to predict. So almost like minority report inside your organization looking for signals that you might be a bad player. And then like, you know, some companies will just say I'm automatically doing like AI reviews of the workforce and people will just be exited. Without any, let's say, grace or deeper research or understanding.

[00:21:52] It's just a methodology for thinning out your workforce. And I think those technologies are super dangerous. I personally was like, I'm not interested in supporting this. I actually gave the team a whole bunch of feedback. I'm like, well, I think they should stop their business.

[00:22:11] But I think that's we're living in this world where on one hand, leaders, whether it's the CEO, CHRO, COO, CIO, CTO, all of them are pressured to go faster in AI. They're like, I know we need to be using it, adopting it, changing our workforce with it. They're assigning responsibilities. They're buying technology. They're buying enterprise platforms.

[00:22:40] They're building lots of things. But there isn't really a thoughtful process around problem solving or like researching and understanding what problem we're solving, researching and understanding like how we do it today. Is it maybe the way we'd want to do it? So let's not just pour AI on top of a process that already exists to repeat it. Like maybe that process shouldn't exist.

[00:23:05] Like there's all of these design challenges with deploying this technology that those muscles haven't either they haven't been appreciated and built and celebrated, recognized or or we're just moving too fast and they kind of go out the window. Like the ethics and design thinking are friction. They slow you down.

[00:23:32] Like I remember even at Salesforce, like we, my research team, you know, it would have been nice to do a six month ethno research design research project on some of the problems we were solving. But, you know, we were like, you get you get six to eight weeks and we're going to have to do really wild research in really different ways to get insights that quickly because there's not we don't have six months.

[00:24:01] And so now I think people can like start to turn to AI and ask a question, assume like it sounds pretty confident, like let's assume that's right and we'll just run with it. And and so we're getting that's why we're getting AI slop. And I think we're getting process slop and we're getting application slop.

[00:24:21] And so what so in the spirit of speed and using AI and building adoption, they're they're they're actually creating things that at the same time reduce trust, create fear and put people into what I call functional freeze, which is just like they're going through the motions. They're doing the work, but they're burnt out.

[00:24:47] They're afraid and they're just like, I just need I need to not lose my job. So I'll just like go through the motions and do the thing and quietly productivity and performance get killed. I don't mean to be so dark about it because I think there's a positive way out of this, but I think that's the state we're in in 2026. No, no, I think you're raising a really important point.

[00:25:09] You know, I often use this phrase responsible innovation, but like that is really difficult in practice when you've got incentives that are pointed towards things that don't take that responsibility part of it into into account. So I feel like I also because everyone is a builder now, you know, I use the phrase responsible by design, which, of course, Irene and crew would have used that long ago when they talked about human computer interaction.

[00:25:38] And, you know, privacy by design and, you know, privacy by design and all of these concepts that you need to think about as you're designing these solutions. So it's some combination of are you to one of your original points? Are you solving a real problem or is this or is this a money grab and trying to hit while the market is is hot and, you know, build up a user base? Or are you going to do something the right way from the beginning, taking this sort of human?

[00:26:08] I use the phrase human centric quite a bit, but I know you've talked about like this human positive. Yeah. And so so I want to sort of dig into into that a little bit. But I just feel like as you know, you're not it's not just, you know, engineers and developers that are building these things. It's it's the every it's the every man.

[00:26:29] It's that any student, any worker, any freelancer can be, you know, whipping these things up as a side hustle or just a pet project just to see what happens. And unfortunately, we just don't have the right, you know, guardrails in place as a as a society or from a governmental perspective. Yeah, it's it's interesting because it's magical. Like I would like I love the idea of everyone can be a builder, but I also like and this is not a new problem.

[00:26:59] Like we have seen technologists build technology for technology's sake. Like there have been a lot of apps and things that were built that was, you know, somebody had an idea in a closet and they happened to have technical skills and they built it and it was not relevant or useful or successful. Right. So now we have more lay people creating more stuff. And the other my other pet peeve is like, what are we measuring?

[00:27:27] And like in the world of HR, the example I always use is time to hire, like time to hire is always the metric for recruiters. And yet, like if time to hire is the metric, you're not going to, let's say, look look under unusual stones for great talent. You're going to go find easy talent to win very fast. And then they may or may not be a fit. They may or may not be in the company in a year from now.

[00:27:55] But you were measured and you close your you close, you know, close the deal, the talent deal fast. And I think we have all of these broken metrics and token maxing is one of them. The companies were like, we just want everybody to start using AI. So we'll just like measure how many tokens they're using to prove that they're using AI. And I know now a lot of companies have backtracked on that and they're not doing it.

[00:28:21] But we have to be thoughtful about which metrics matter in an organization. And that that was really the start of this human positive company framework that we've been building to help guide companies to understand what is it that they're actually building or buying when they make these decisions.

[00:28:42] And I think it's to help talent choose companies that they truly want to work for, to help buyers, including IT, decide who do we want to buy from and and to help investors figure out who do we want to invest in. And this this human positive company framework has three pillars. I can just give you the high level and we can dive deeper into it if you like. But the first the first pillar is workforce ingenuity.

[00:29:10] And the idea is not just, you know, my my workforce is innovative and and creative. It really starts with the basics around trust and psychological safety and caring for your people, compensating them fairly. Like all of the sort of Maslow's hierarchy needs are met. You're you're empowered. You are learning. You're growing. And ultimately, you have agency to create in your organization. You have a voice you're listened to.

[00:29:39] So that is that pillar. And it's an example of a metric there is like what percentage of your revenue is spent on reskilling and upskilling? Like, I don't I don't know what the answer should be, but I know that we should be thinking about reinvesting our profit into or reinvesting revenue into learning.

[00:30:03] The second pillar is what I call positive sum prosperity, which is how you operate in an ecosystem. Fair play. It is the one that says, like, we grow the pie together. This is not a zero sum game. And we all win and we are all stronger when we cooperate and, you know, without antitrust and monopolistic behavior. Like we should we can win together.

[00:30:29] And there's a whole but and there's also media narrative control as part of that. Like, are we manipulating the media? Are we telling the truth? Are we honest about what our intentions are? And and then the last one is the ethical and humane use of tech.

[00:30:46] And that's all your guardrails and thoughtfulness around governance and including including your workforce and your stakeholders, your consumers in AI decisions. Like what this is going back to design, how you design the technology to protect people, do no harm, that kind of that kind of thing.

[00:31:08] So we're able to give organizations essentially an audit and say we think like, you know, we're building technology that meets these criteria. We're operating our company in a way that meets that criteria. Not everyone necessarily agrees with all of those things. But we think if you if you operate generally through this lens, you will be human positive and create opportunities for people to thrive and create value together.

[00:31:39] And that's going to be good for the economy. It's going to be good for society. It's going to be great for the world. Yeah. No, I love I love the framing. I mean, I think it's hard for people to it's either hard for them to get their head around all of that and and then decide, OK, that that all sounds great. But how do I how do I track it and how do I translate that into, you know, actual, you know, quantifiable, you know, growth and and what have you.

[00:32:04] So I think that's where we get tripped up a lot in even in some of the metrics like you're talking about in in talent acquisition, like how can't just look at time to hire in isolation. If if you're if your quality of hire, assuming you're measuring that at all, if if that's not going up for tensions, not going up, if engagement's not going up, if morale's not going up. What does it matter how fast you hired? You have a revolving door that's spinning faster.

[00:32:32] Exactly. And and that's what we're seeing. Right. Like burnout is out of control. Like disengagement is is growing. And, you know, the great places to work organization just came out with data that shows that if you treat your people well, those companies that treat their people well are in great places to work are three times more successful. I don't remember if that was revenue or profit or what have you, but are three times more valuable than the others.

[00:32:59] Right. Like it does take time to be able to measure consistently how these practices actually result in in outcomes that that boards and and shareholders care about. And so, you know, you can look at the framework like for each of these pillars. There's there's seven criteria and we're very clear and prescriptive about like what areas to look at and and at least debate them.

[00:33:29] They all seem they're all pretty well tested at this point. Like we we go through with organizations. It is it is a it is a moral leadership question, though, like you can be mercenary and transactional and be wildly successful. You can lie. You can cheat.

[00:33:46] You know, like for for a lot of and I mean, you know, I would say we also do this war games like exercise where we have a workshop where people role play being in different kinds of companies. And, you know, it's really fun because one has no constraints and one has all these human positive constraints and they're both confronted with the same problem and they have to go solve it. They solve it in very different ways.

[00:34:14] And at the end of each one of these scenarios, we learn what old companies have done and then everyone gets to vote. So if if if you are now called by a recruiter and asked you, which company do you want to work in, which one would you work in? If you were if you were in the position to buy from one of these companies, which company would you buy from? And if you were an investor and you had to bet on one of these companies, which one would you bet on?

[00:34:39] And consistently rooms of executives will say like 75 percent choose the human positive one, even though there's more friction and it's harder to navigate through those constraints. You know, but 25 percent are like, nope, I really like full on. It's all profit and I'm going to be a trillionaire kind of approach.

[00:35:04] And and that's I think that's the conversation that we need to have in that, like every employee is responsible for understanding, like what what am I complicit in? When I make a decision, when I create a tool, when I use a tool, buy a tool, like what am I supporting? And that's that's we need to do a lot more education and training around that.

[00:35:29] I'm actually on a panel tomorrow where we're talking about about that, about the decision, not just what general tasks should be given to AI. But when you know, AI is part of your workflow, such as in talent acquisition through the recruiting file. What is what does it really mean to a human in the loop is is like it's almost meaningless at this point. What is the loop? What and where is the human in said loop?

[00:35:58] And what does that really mean? And you realize that every every stage gate through a recruiting funnel. Decisions are made. Right. So if there's no humans involved in even those top of funnel decisions, if you excluded people, if you're not giving them an interview, if you're not giving them an assessment, if you're not even giving them a recruiter, you know, a 20 minute recruiter phone screen. That is a decision.

[00:36:26] And if a human didn't look at that and make that decision, you have outsourced decision making to an algorithm. And that's a problem. So what are we supposed to do with the thousand applicant per open role ratio? What are we supposed to do? This problem existed before AI. And, you know, it's interesting.

[00:36:49] So this is why organizations often only interview Ivy League or Tier 1 students in schools. Like that was that was part of what we were tackling with Red Matter was like even getting students who didn't go to a Tier 1 school some visibility because there's amazing talent that that just doesn't break through.

[00:37:13] And and what I learned at the time was, you know, like the Accentures of the world, the big companies were afraid to even look at an interview like a an application from someone who didn't go to a Tier 1 school. Because if they interviewed someone, let's just say like University of West Virginia. And their resume looked exactly the same as the person who went to Yale. But you've got Yale on the list.

[00:37:42] And if you interview that person from West Virginia now, because of the EEOC, you actually have to interview everybody else who has a resume that looks like the person at University of West Virginia. But, you know, what is a differentiator, unfortunately, is like the brand of the school that you went to. So they use these screens sort of as or proxies to determine who gets through the interview.

[00:38:07] And so, like, on one hand, I would say AI has a great there's a great opportunity here to look at more talent, to find more talent, to give people more like AI interviews to or new ways to prove themselves and get on your radar and earn the interview, if you will. Like, but you have to think about all of those things to design into your process. And I just want to give you one other example of that on the flip side.

[00:38:37] And this was something from Salesforce, actually. It was probably 2017-ish when we created an AI to predict attrition. And this was like our HR analytics team, data science team working on this problem of like, could we predict attrition? And lo and behold, we really could.

[00:39:02] And we could go to a manager and say, you know, this person is very likely to resign in the next month or two. And a lot of companies, like, thought that was awesome. Like, you know, especially leaders, HR leaders, they're like, wow, we would love. And in fact, that surveillance tech company that I talked about, they also had an attrition predictor agent. Here's the problem.

[00:39:29] As soon as you start going to managers and telling them that someone is going to leave, they don't want that person to leave. So they take all the bonuses, the RSUs, the promotions, like all the good stuff that should be distributed to the person who is most deserving of those things. And instead, they give it to the person who is one foot out the door and threatening to leave to try to keep them. And then those things are not available to the people who deserve them.

[00:39:58] And then eventually what we found by doing it is that people would start to look for jobs and start to try to leave just so they could get those things. So you've, by doing this thing that you think would be good to try to keep people around, actually eroded trust. And we, because we had an ethical review process for innovation, we could look at it and say, this is eroding trust.

[00:40:27] We must kill this innovation project and not do it. And I think a lot of organizations do not have a group, committee, team, or process for looking at what is being created and what are the implications and what are the potential harms inside the organization. And they're just happening. So that's where we need more thoughtfulness and education.

[00:40:52] It doesn't mean you don't do the experiment, but you at least are thoughtful about measuring it and understanding it. And by the way, we turned that AI into a re-recruiting tool that did not predict attrition, but it did predict who was likely ready for a new role. So internal recruiters would call them and treat them like a candidate for the next cool job at Salesforce. It was a huge win, right?

[00:41:19] So I'm not trying to stifle innovation at all, but we do need more thoughtful design process and governance inside organizations. Yeah, for sure. So when you think about organizations that are reasonably sort of mature in their AI journey, at least from an oversight standpoint, to your point, they've got some committees.

[00:41:47] Maybe they have an AI ethics committee, responsible AI committee trying to really assess, is this something that we want to invest in? And if so, how much, what are the metrics to gauge success? And would that be different for like the sort of innovation pipeline as opposed to just sort of more normal projects all together, all in one review?

[00:42:15] I think it's the same across the board. I think, first of all, it is about the whole organization having a voice, right? So it's like that see something, say something. Like, you know, it doesn't mean that every organization is going to go, anyone who wants to innovate anything has to go through this whole laborious process. But it's there. So, like, if you are tuned into the values of the company, I think it starts with values.

[00:42:45] So, you know, our company values innovation, but they also value trust, also value fairness, like whatever those things are. The first job is to make sure everyone in the company understands those values and the behaviors associated with those values.

[00:43:01] And then when someone has an ethical question, a values-based question, there's a place to go to ask, you know, how does this, you know, is this thing I'm working on or is this idea actually aligned to those values? Like, what are the potential implications or problems with this?

[00:43:24] And think, do the mental gymnastics around that a little bit and have a place to go, which, by the way, can be AI as a first step if you've trained that AI really well to represent those values. That's something that we do at Tough Day. And give every employee a path forward to raise a red flag and bring it before someone.

[00:43:50] So, you know, at Salesforce, we had the Office of Humane and Ethical Technology, and any employee could go to them at any time and say, I have a concern. And actually, it was encouraged and celebrated. It wasn't like a bad thing to say something. It was a good thing to say something. And they could put together a team of diverse people from across the company to actually think about it.

[00:44:14] Like, when employees said we shouldn't sell our technology to gun manufacturers, like, there was a group of people that came together to explore, like, what are our guardrails around this? And we, at the time, decided not to sell the technology to those who sell bumper stocks. But, like, that's where we landed. But it was not one person or a small team thinking about it. It was like, that's a process. That team runs the process.

[00:44:41] But it is anybody in the company and different people, different stakeholders across the company chiming in on it to come to a decision really quickly. Yeah. So how did you build the AI, I guess, that early stage sort of vetting AI? Like, what knowledge from what people sort of went in to that to give it a once over? So I'll just give you a little bit of background on Tough Day.

[00:45:09] Like, so we've been building Tough Day to essentially augment managers. And so the AI is smart about management and HR and employment law and your organizational knowledge of your company.

[00:45:26] So we can ingest everything from, like, your values and your strategy to your HR policies, how things work inside the company, written and unwritten rules, cultural things, to, you know, accessing or orchestrating, like, getting into different systems that are relevant.

[00:45:44] And the idea was really, like, what, like, if you think about the best manager you've ever had, like, they were probably someone who was, like, really good at the human side, but also really knew the organization really well and could help you navigate it. And so in terms of getting those values right, every organization has different values or many are the same.

[00:46:09] But, you know, we worked with one organization in Hawaii where they said, you know, we want to imbue what they call aloha spirit into the AI and had very strong cultural values. Like, what does that mean, aloha spirit? And just to get very tactical, what we did was used redacted transcripts between people having difficult conversations in the style that really represents those values.

[00:46:38] And we were able to adapt the AI to sound like and to have to represent those values. We also did a ton of red teaming and trying to make it, make the AI say or do something that was against those values and identifying those gaps and then, like, solving for that.

[00:47:01] So, but at the end of the day, like, we saw the helpfulness rating of this AI go from 88%, which we thought was really good, to 99.6% because it aligned with the values. And I think, you know, it's not hard for organizations to, well, let me say this. A lot of organizations have values written on the wall and that's as far as it goes. That is not what we're talking about here.

[00:47:30] We're really talking about living values, being part of everyone's, even their review process. Like, you know, do I represent the values? Do I behave in a way that showcases those values and celebrates those values? And am I, you know, a light bearer for those values? And I think that not every company is doing that. That's an opportunity for them. And of course, there have to be consequences when you don't live by the values, right?

[00:48:00] Like, if you are sociopathic and that is not okay with your company, like, you know, you don't keep around a salesperson who really treats people terribly, even if they're driving great sales. Like, there have to be consequences. So I think part of it is the AI and part of it is just, like, culture in general and deciding to have, and go back to moral leadership, like, deciding to live by these values.

[00:48:28] And then the process part, the AI part, is not as hard as you would think. Yeah. No, I like that approach a lot. As you were talking, I was thinking about some of the conversations I've had around, it sort of ties to the collective intelligence and collaborative decision-making that I mentioned at the beginning that I sort of got hooked on from some of the IBM research work. But it was really about, well, what does that mean in the age of AI?

[00:48:54] So, you know, I've, in the past, was talking about collective human intelligence, obviously, with some analytics to derive, you know, some insights and things like that. But even some of the IBM projects, I remember, it was like, they had, like, these, like, debater technologies where different people from different perspectives would come in and it would, you know, chime in as well.

[00:49:16] And now you have, you know, the potential to, you know, create these agents who are constantly learning and ingesting very specific, a very specific sort of corpus of information to help understand the nuances of some of that. And how that could be injected into your operating model, but also your innovation management sort of funnel, right? Like, we can help you sort of parse this out.

[00:49:45] We can help you be responsible by design. We can help you think about users that you, that weren't on your sort of target list, but you have to try to anticipate technology used in ways that were not originally, it wasn't originally designed to do with users who may have perhaps, you know, ill intent, right? AI is trickier these days.

[00:50:08] And performing those types of, you know, like you said, red teaming or evaluations is trickier than, you know, actual coded, you know, software that is, that is rules-based. There's only so many buttons. There's so many, so many things that a user could possibly do with traditional software. And, you know, the game has changed. It has. It has.

[00:50:30] I think like some of the, I think new research methodologies are being created that can help with all of that. When we started building Tough Day, the first research we did, we called Wizard of Oz testing. I don't know if you've heard of Wizard of Oz testing. It's where we had humans playing AI.

[00:50:54] And so the first phase was we invited our, you know, research subjects, our target users to engage with AI. And we told them like, you'll be, you'll be interacting with this AI, but just want you to know there's humans in, in the AI too. So you're engaging actually with humans and AI. So it's all confidential, but, you know, please come in and engage with this thing.

[00:51:22] And the reality is we had sitting around the table, really amazing leaders, managers, HR, an employment lawyer, a therapist, a coach, who am I missing? Like a couple other archetypes or like real humans. And we told them, okay, when someone types in, like we, we told everyone it's going to be slow. Like, don't, don't think it's going to be as fast as like chat to PT and whatever. It's going to be slow.

[00:51:50] And we, we had them decide like, who's going to take this question, who, who's going to answer this question. And one of the things that we learned very quickly through observing this was none of them give any answers. What they all do is they ask more questions because they all need more information to figure out what's really going on here and diagnose what the problem is before they can give any advice.

[00:52:17] And so by observing that we did design our AI to be very curious and to emulate that behavior. And I think just by being a little more thoughtful, when you go to build your AI solution, whatever that is, like think about different ways where you can learn how that thing should behave for the, for the outcome that you're looking for.

[00:52:44] And I will say, and then with respect to AI, like we got to the point where we were like, we're going to use synthetic humans, synthetic workers to come in and start challenging our wizards behind the curtain. And then we got to the point where we had the AI and it was like synthetic users and synthetic and, and, and the AI answering them having conversations. So we could learn from all of those conversations as well.

[00:53:14] And, and, and, and do the red teaming and do all these other things. Right. So it's, if safety is a value and for us, trust is, and, and safety is part of that. Like we would never want to give an employee bad advice about their workplace. Like that just cannot happen. And so we designed all the different research as part of the process. You don't just go build the app. You build the research first.

[00:53:39] You know what you need to know, and then you start to design and build the solution and test it. Yeah. No, it sounds like a very pragmatic approach. It's fun too. It's not just pragmatic. It's really thought and it delivers better results. Yeah, no doubt. Well, I think that's common advice these days, right? We need people to ask better, better questions, right? And I know we should be spending more time making sure we truly understand the problem before we try to come up with a solution.

[00:54:08] So I think that's good advice for anyone, no matter what industry or, you know, rung you are in your company. So, KVJ, we covered a lot of ground. I feel like we could talk for at least another hour.

[00:54:24] But I want to be respectful of your time and I wanted to make sure, you know, if there's anything else you're working on, any advice that you have for the audience in terms of, you know, AI readiness or the human and AI collaboration opportunities, you know, things like that. Anything else you want to share? Oh, gosh. I would just say very happy to engage with people on this topic. I'll send you some links. Maybe we can just post some links to go with the podcast.

[00:54:52] But I just I think that right now we're in a moment. It's a call to action. We are creating our future and we have to be thoughtful about what future we want to create. We're all creating the future with every decision that we make, every application we launch. And that's a very powerful and amazing opportunity for everyone. And it's a huge responsibility.

[00:55:16] So I would be very happy to talk with anyone that wants to dive into that and needs a little help. Yeah, I appreciate that. I'm sure my audience will as well. So we'll put your contact information in the show notes as well. And KBJ, thank you so much for spending some time with me. This has been great. My pleasure. Thank you for having me. And I'm looking forward to staying in touch and hearing about your panel tomorrow and what comes out of that. Yeah, absolutely. All right. Well, thank you again. And thanks, everyone, for listening.

[00:55:46] We will catch you next time.