Welcome to the forefront of conversational AI as we explore the fascinating world of AI chatbots in our dedicated blog series. Discover the latest advancements, applications, and strategies that propel the evolution of chatbot technology. From enhancing customer interactions to streamlining business processes, these articles delve into the innovative ways artificial intelligence is shaping the landscape of automated conversational agents. Whether you’re a business owner, developer, or simply intrigued by the future of interactive technology, join us on this journey to unravel the transformative power and endless possibilities of AI chatbots.
Some colleges and universities are hesitant about students using AI, while others have embraced it.
The University of Cincinnati has created its own version of an AI chatbot, BearcatGPT, becoming the first Ohio university to incorporate their own AI platform for their students.
For most people, networking is vital to the progression of their careers. But is networking equal for all?
We will talk about the factors that play into successful networking and if networking sometimes leads to nowhere, especially in the tech industry.
And we’re discussing two recent lawsuits. The first one is Elon Musk’s against Open AI, which was lost due to the timeliness of the suit. The second one is against many kids’ dream vacation, Disney, and what it means for the parks.
It’s all part of the discussion for this week’s Tech Tuesday.
Guests:
This transcript is generated with AI. To ensure its accuracy, review the audio file.
Amy Juravich: Welcome to Tech Tuesday from All Sides with Amy Juravich. This is a show where we share stories about science, technology, and the future of our environment. The use of AI tools for college students has become unavoidable. College campuses have turned to creating their own chat box to comply with their own privacy standards. The University of Cincinnati has become the first Ohio university to offer an AI platform tailored for its students and staff. Joining us now to tell us more about the program at UC is Vice President and Chief Digital Officer Bharath Prabhankaran. Welcome to All Sides.
Bharath Prabhakaran: Thank you, Amy. Thanks for having me.
Juravich: So you’re calling this Bearcat GPT. And it’s like, it’s a limited use instance, but similar to OpenAI’s chat GPT, and can you explain to me what can Bearcat G-P-T do?
Prabhakaran: Sure, yeah, really what you just alluded to, right? So I think the goal behind creating this platform was to make sure that our faculty staff and students have access to these AI tools that are evolving on a daily hourly basis sometimes in a private and secure fashion, right. Our biggest goal behind this was making sure that any data that’s shared that’s UC proprietary or specific in this environment stays within the environment.
Think of it like a walled garden. So in some ways it can actually. Full data like OpenAI does or ChatGPP does from the corpus of the internet, right? But, you know, the data that we share within the BearCatGPP instance is not pushed outside to train these public large land blocks, right, so that allows us flexibility to train it on, you know, data that, we need specific to our research or student use cases and so on and so on, right. So that’s really the model or the goal is to find a private secure instance of, for now, OpenAI, but other models in the future of these tools in the BearcatGPP.
Juravich: Okay, so what made you decide though to make your own? Because your students could just use the OpenAI one. Was it really just the privacy and the security aspect? You feel like, I guess you wanted more control?
Prabhakaran: Uh, I wouldn’t say control specifically, uh, but, but in some ways, yes, I mean, we have, you know, controls that double that store, right? So we’re certainly not controlling what people are doing in it. I think the goal is to control the data is like me to look at it.
And I think secondarily, the public instances of tab GPT allow you to do things which are not UC specific. So we have actually designed things that I can talk about within bearcats GPT for instance, that we wouldn’t be able to do. Within, you know, public stack GPT because we’re leveraging data from our internal system like Canvas, we use our learning management system and other systems, right, so we can leverage student data from that and our ERP platforms and so on and so forth within their CAD GPT to do cool, you, know, agentic stuff, which we can’t in the public instance. It’s sort of double-edged, right? I mean, we want privacy and control, but we also want the access to internal systems.
Juravich: Okay. So give me some examples of what people are using your Bearcat GPT to do. What can they do with it?
Prabhakaran: Sure, so I can give you quite a few examples. It’s been live for about, I would say, six months for faculty and staff got it first. We’ve actually had a pilot for over a year. So we wanted to make sure we kicked the tires, test the loads and usage and capacity and so on. About 50,000 faculty and stuff got it in the fall of last year of 2025. And then in the spring of this year in February, specifically, we rolled it out to all our students. There are about 54,000, right? So if you think about it, about 70,000 user base across all those, you know, stakeholders.
So even in the short time that it’s been live, we’ve had over a hundred plus agents that have been built on the platform. So the usage is obviously there, the demand’s there. And some of the sort of interesting use cases that have built on top of it are we built our own personalized tutoring platform. We realized there’s a big need for tutoring, especially our freshman students. I was an engineering student way back when and calculus was always my Achilles heel, which is a bit of a problem, but this calculus is a prerequisite for engineering. And that’s where really most of our first year students struggle is the math courses.
We started with calculus and statistics. At Claremont College, which is one of our branch campuses, we had students really struggling with these courses. So we built a set of three agents that basically provide personalized tutoring on math courses right and so that leverages the data obviously from you know student information system from a learning management system and so on and so forth to provide them tutoring so like I said it’s three agents there’s one called I think Bearcat StudyPal there’s a Bearcat Genius and then one that’s a bearcat test so those are the three agents that sort of work hand in hand to really take a student through the process and this is not my chat GPT so a big difference there is you’ve been to chat GPTs that I have a homework here at five o’clock this evening You need the answer, it’ll give you the answer.
Here we’ve actually built it to have some Socratic guardrails in place. Really it’ll be like a real tutor saying, okay, great, but let me help you understand the concept. What do you know about differential equations or integration or whatever it is, right? And then it’ll sort of walk you through step-by-step. So if you were kind of at the beginner level, maybe this party pal will be the agent that’s interacting with you. If you’re a little more advanced, it senses that and sends you to genius. And then finally, if you’re ready to prep, then it will do some quizlets and test prep and stuff like that, right.
So that’s just an interesting use case we built. There are several others that are in the works right now in pilot phase, but I would say that one of the big reasons we have it is you can actually experiment in a safe and secure fashion, have access to internal data, and then, you know, see the results that are really hitting where it matters. We’re all about student success, right? Are we supporting student success using these platforms?
Juravich: Yeah, okay, so the example you just gave with the tutoring, so on your Bearcat platform, it won’t give you the answers to the quiz, right? It’ll tutor you and teach you and help you arrive at the answers. But I mean, there’s nothing stopping your students from going to the other version of Chad GPT and putting the test in and saying, give me the answers, right, so.
Prabhakaran: No, there isn’t and at the end of the day, that’s the bigger conversation around AI usage in courses and the push and pull between faculty and students, right? That’s a whole other topic, but yes. Yes.
Juravich: Okay.
Prabhakaran: I hope the students use this, I mean, the right way. And really the goal here is we can’t ever have enough tutors for all our students, right, to provide personal ice cream rings. So how do we use the platform for things like that? Now-
Juravich: Yeah, to use to use open AI at a really high level, you have to pay. I mean, there’s the there’s everything that we we all know that’s available for free, you know, just via the Internet. Is this a part of like, is there an additional student fee or anything to be able to use the Bearcat GPT?
Prabhakaran: No, we made an investment entirely to provide this platform to our stakeholders, so no extra fee for anybody, the university is paying for the cost.
Juravich: This is Tech Tuesday From All Sides on 89.7 NPR News. We’re talking about Bearcat GPT with University of Cincinnati Vice President and Chief Digital Officer Bharath Prabhankaran. So this is a private version. It ensures that you see data isn’t being used in other AI models. But is the University of Cincinnati gonna use the data themselves? Are you guys gonna use the data to learn, to build more things? I suppose there’s things you could do that the university can do with the data.
Prabhakaran: Sure. I mean, with the caveat that all within the regulations that we have to comply with, right? So the student data, that’s FERPA, we have to make sure that the privacy of the student data is protected, right, for other kinds of data. There’s other compliance regulations. So you don’t want to be even used for misusing data or doing things for other types of stuff, right.
But outside of that, I think, yeah, absolutely. Like. With tutoring, you know, we can track trends or patterns, right? So I mean, if I know we can tie it into our learning management system, if a student is, you know, failing quizzes or the other things that are happening, we can maybe predict some of those trends and maybe proactively do a human intervention for a tutor. Right. For example. Right. So there are things we can do with the data we collect from it and adhering to the compliance and privacy policies, which we hope to leverage to get better and also build difference.
Juravich: If the students can the students have conversations with their professors or you know through the model to Is it possible to like get to connect with your with your to with your teachers the faculty through it?
Prabhakaran: So that’s the next step is adding on that functionalities. Right now, it’s just the agent that’s tutoring them, right? But actually, you know, we’re gonna build in this concept of a digital twin. Really, so that’s a broader companion agent, right. So every student that comes in will get a companion agent. That’s the vision, right, and that walks them through the entire academic career, if you will, right and so that will include interactions of, you now, contact a tutor or, you find a study group or, contact your professor.
And if you look at it from the flip side, we can also have agents for professors. Where, you know, when they’re not available, for instance, the agent can run 24-7, right? So the student can interact with the corpus of the body of knowledge that the professor provides, right. So yeah, that’s in the vision. It’s not there yet, but not yet.
Juravich: Are there, is there any worry that students, you know, if the university can see that they are using the tutor a lot, is any worry the students will be like, oh, the university knows I’m not very good at this or something like that? I mean, they’re kind of being watched in a way, right?
Prabhakaran: Um, we don’t really track specific queries of what things are being done by students. So like, so if you don’t track what each specific student is doing on the platform, right? We do, all we do is keep track of trends and then, you know, there are certain things in these bodies of not AI that you, you know, terms that you shouldn’t be using, right. How do I do that thing? I mean, we try some of those keywords, but we don’t track specific usage patterns.
And even if we did, right, I mean we would never penalize the student for actually asking for help, you know, using a tool to help them get better. Right. I mean I we would view that as a positive student is actually trying to get better as opposed to doing it. But we don’t t we don
Juravich: So, I mean, it’s obvious that your university’s leaning into AI, you know, even though it’s changing every day, it is the future. How much are you leaning into it? Because there’s the idea of professors using it for presentations, for grading, for more. Are you an advocate of using AI for faculty to make presentations, to do grading, to
Prabhakaran: Uh, not, not all in, right? So I think that we are certainly, uh, on the firmly in a camp of responsible use of AI and ethical use of the AI, right. So that’s the foundation of everything we do. Uh, in fact, you know, just to digress a little bit, um, we have a structure that we’ve set up called the AI community of practice, or you can think of it as center of excellence. So we have different committees focused on different areas of AI.
So how does the app plan teaching and learning for us or research? Policies and guidelines around AI. And then the responsible use of AI is actually a focus group that really looks at how we use this responsibly, right? So we all constantly evaluate solutions that are available. We are definitely not yet there with automated grading. I don’t think we’ll ever get there. We may leverage it for some levels of user assignment grading, but never without a human ability.
We are, we definitely believe in responsible use of AI and we are not, I don’t think we’re quite there yet or leaning in that direction at all. Automated. I mean, the faculty, sure, they may use it to build presentations or use it as a tool to assist them, but not in the grading, not in the application process, right? Because, you know, as we know, AI has inherent bias, right. I mean, there’s issues where it does hallucinations and things like that. So we don’t want that to creep into any of those students facing us.
Juravich: There are critics of AI programs being used at universities. I read one, Ronald Purser is a professor at San Francisco State University, he wrote this lengthy piece. The title was, AI is destroying the university and learning itself, a very polarizing title. But in it he said, students are using AI to write papers, professors are using it to grade them, degrees are meaningless, and tech companies are making a fortune. Welcome to the death of higher education. That was the quote that struck me. I mean, how do you feel about this? I mean you’re like, are you worried about promoting AI rather than limiting it?
Prabhakaran: He could be one of our faculty members for all you can. You could talk to some of our faculty who probably sit in that camp, right? And so maybe there’s some marriage to some about those arguments, but then reality is this, right. I mean, AI is here. I mean the genie’s out of the bottle, right, how it’s used and so on. It is really causing high rate to maybe introspect. In terms of how we deliver education and what we’re actually delivering. I think that the model that higher ed has used is this whole ivory tower model, right? We’ve been doing things the same way for 400 years.
There’s faculty who’ve been teaching the course the same day for 30 years. We will not pivot, we won’t do anything different. I mean, the reality is, you know, these tools are here. So, I mean what each faculty member should be asking themselves is, what are we actually trying to teach the students? What’s the learning outcome that we’re hoping to get out of this particular course, right. If it’s a creative writing force, for example, and you suspect all your students are using chat GPT for assignments, put them in a room without any technology, then, you know, make them write on a piece of paper and you figure out who’s been cheating and who hasn’t.
So I think in some ways, you know, That’s the dystopian scenario, right? Where, you know, AI running everything, the agents are running amok, and there’s no humans in the loop. And we all end up like the Terminator movies, like with Skynet killing us all, I ended up dating myself with my ex with those movies. Or we’re like in the movie Bali, right, we’re all fat and lazy, lying on a couch with universal basic income, and the robots are serving us, right. I mean, that’s why Elon Musk scenario. He says, you don’t need to stay for retirement, blah, blah, block.
But I think higher education has to just adapt and pivot, I think. The traditional degree programs and the curriculum that’s been taught for many years has to be revised. So we are actually actively doing that in many of our schools. They work with our industry partners. So I sit on also on the other side of the CIO, and we have a CIO round table as all the corporations, CIOs and others. So we work with school of IT to give them feedback in the curriculum, right? Hey, you shouldn’t be teaching this topic which was 20 years ago, right. Let’s bring new stuff in the program.
So I think there’s a constant effort to sort of refresh our curriculum. Of, you know, provide the tools to our students. And, you now, we would be doing our students a disservice, faculty member, by not teaching them, that equipping them for the, you, know, aid of the future and these jobs of the future, which none of us really has a crystal ball on what they will be. But one thing is for sure, I mean, critical thinking will be required. All those human skills that are, you know, part of our portfolio is what’s going to distinguish us in the future.
And, you know, these tools should be what they are today, our productivity assistance, where we get to that, you know, AGI concept they keep talking about where AI is more intelligent than humans. We, I don’t think we’re quite there. I think we need to learn how to coexist and add. So what’s uniquely human about us, teach our students and cultivate the skills and give them, you know, access to these tools to use and then, you know, historically, I mean, even before AI cheating is always been there, right? There’s students find creative ways to cheat. If they want to take a shortcut, they will do it.
So AI maybe is taking that to a completely different level, but I think we, and I’m not a faculty member, so I can’t speak for them, but they need to pivot. The education system needs to pivot in terms of what we offer, and also maybe chunk up learning into credentials or be prepared for some, you don’t need a four-year degree for everything. So we chunk things in and do different types of degree programs. So I think those are all conversations that are being.
Juravich: Well, in the minute or so that we have left, talk to me about the future of Bearcat GPT. You talked a little bit about improvements you plan to make. When students come back to school for the next semester, what can they expect? So what are you hoping they’ll be able to do?
Prabhakaran: So right now, like I said, it’s primarily OpenAI, which is the chat GPT model, right? The large language model that supports chat GPTs. So we are planning to roll out additional models. So Gemini is on the hook. So hopefully by the time they’re back, we should have access to Gemini, which is Google’s large language tool for AI. And then we continue to add a set of other models, with the vision being really, this is sort of their one stop shop for most AI tools that they would be using outside of UC. So we provide them those within UC. And then in parallel, What they can also expire to some additional use cases, you know, agents that we’re building that will hopefully streamline their life and make it better for them.
Juravich: We’ve been talking about Bearcat GPT, an AI platform for University of Cincinnati students and staff, and we’ve been with UC Vice President and Chief Digital Officer Bharath Prabhamkaran. Thank you so much for your time today.
Prabhakaran: Thank you, appreciate the conversation.
Juravich: And coming up, we’re gonna talk about gender inequality in the tech industry. It’s been there for years and it still persists today. That’s when Tech Tuesday from all sides continues on 89.7 NPR News.
You’re listening to All Sides with Amy Juravich, a show where we share stories about science, technology, and the future of our environment. Networking is a vital tool for developing a person’s professional career. It builds and maintains relationships, expands professional networks, facilitates the exchange of ideas and thoughts, and often leads to job opportunities. Ethel Mickey is a sociologist who studies why inequality persists in workplaces that claim to have fixed it. She also is an assistant professor at California State University, San Bernardino. Her research and her book, “Networking to Nowhere: How Gender Inequality Persists in Tech,” reveals how everyday workplace relationships quietly shape who gets ahead in the tech industry. Welcome to All Sides, Ethel.
Ethel Mickey: Thanks for having me, Amy.
Juravich: So congratulations on the book. It’s called “Networking to Nowhere.” It’s about gender inequality in the tech industry specifically, but I wanted to know what inspired you to write a book about this.
Mickey: Yeah, so the book is coming out this June with University of California Press. I’m very excited. It’ll be out in the world. And I was a graduate student in Boston many years ago and had a lot of friends and colleagues who were in the tech industry and working and starting their careers and coming home with these stories about how they were trying to get ahead and make it in this highly competitive, powerful industry and.
Everything from these grueling schedules of attending happy hours and networking events and conferences to perfecting LinkedIn pages to joining soccer leagues across the river in Cambridge so they can hang out with MIT graduates. And I just became really fascinated by how relationships were being managed in this industry. And of course, as a gender scholar, I want my research not only to think about understanding what’s going on in workplaces for women, but how do we?
Improve workplace conditions for women. And so why not start at the top? Tech is the most powerful industry, both economically and politically, but we know it’s remained stubbornly dominated by men, despite some really deep investments by the government, by nonprofits, by corporations themselves to train women in STEM and to infuse them into technical roles. These efforts haven’t paid off. So I was trying to see, is there a connection here between the way people are building relationships and these patterns around gender?
Juravich: Yeah, you mentioned a paradox of modern tech culture because networking is hailed as the key to opportunity, but it systematically is also working against women. So tell me more about this. Like, why is networking the key when it comes to tech, but why does it also work against women in that way?
Mickey: So we, excuse me, we tend to think about networking that works when it leads to something real for us. We think about network being successful when it needs to maybe a job lead or a referral, best case scenario, you get an interview out of it. But what I found is that the networking that pays off that usually comes from relationships that you already have. So thinking about relying on your friends.
Family members we don’t often think about as part of our professional networks, but I find that a lot of people are leaning on family. Former co-workers, people who can vouch for you. And then the kind of networking that doesn’t work is what a lot people are told to do. We’re given this kind of blanket advice that we need to go to conferences or send out cold emails, talk to strangers. And women especially are given this advice by career coaches. Because the whole lean-in movement was actually… Kind of on this premise that if we just lean in and network harder and connect with other women that women will get ahead. And this can feel productive, but what I find is is it rarely turns into opportunity. So networking networks is built on trust and networking that doesn’t is often built from scratch. And so it’s not about how much you network but whether your network actually has the power to open up doors for you.
Juravich: Well, talk to me about this idea of the boys club. I mean, it dominates many industries, but the tech industry certainly, like it seems to be a boys club, I mean how did this become such a significant force? How do women and minority get around this? But like, let’s start with why did the tech industry become a boys’ club?
Mickey: You know, it’s interesting in the book I write about the history of computer programming, it actually started off as a women’s space. It was seen as a soft skill developing software. And so it was designated to women in the 70s and 80s, whereas the men were tinkering with hardware. And so if we look at the history and the legacy of computing, women were there first.
Juravich: Well, it made me think of typing, too. I mean, women had all the typing skills. But continue, go ahead. Exactly, yes, yes.
Mickey: And over time, as more money became infused in these types of companies and this industry, the men started figuring out, well, we should take claim over this sector. And so we’ve seen a really rapid shift, if you look at the numbers, to this complete swing. So now we see that about 80% of technical jobs are held by men in our country. And That is paired with this culture of masculinity that’s embedded in firms. You mentioned the old boys club. Some people talk about the bro grammar culture, which is this kind of hybrid masculinity of a tech bro and this geek masculinity with also this kind of fraternity work hard, play hard culture.
And it was really interesting because the way I was doing my research was through something called an ethnography. And this is when I was really immersing myself in the industry and in the company that I was studying. And so I was there as a young woman in a very men dominated space. And so, I was talking to people about their experiences but I also was able to experience it myself. What did it feel like to be the only woman in a meeting and what would it feel to try to speak up? What did feel like to be only woman at the happy hour?
The happy hours at the Irish pub. The research was done in Boston became sort of an extension of their research because a lot of workplace conversations happen outside of the office. And so I would tag along and just get to see these really subtle ways that exclusion can happen. It’s not necessarily done intentionally, but by people just hanging out with who they feel comfortable with, bonding over shared interests, this can create a sort of invisible barrier for women.
Juravich: Well, and it’s not just the boys club at work or at the Irish pub for the happy hour. I mean, sports has also become a very male dominated gathering spot when it occurs to like when we’re talking about natural networking. I mean we always think about what you know everything that happens on the golf course. So like more and more women were trying we’re taking up golf because deals were made there. Talk to me about this challenge where with because I think it’s not just golf but it’s also like watching sports together too. Yeah.
Mickey: Yeah, so the company that I was embedded in and studying, they actually celebrated something called the athlete culture. And the HR professionals would tell me we look to hire athletes. And that at first kind of was surprising to me. I’m like, what do you mean by this? You’re a tech company. You’re not a sports management company. And what I came to understand is that it was a certain type of cultural fit that they were looking for. They wanted someone who was yes, familiar with sports, but also was a team player.
Their ideal kind of hire had maybe played college athletics, you know, when they were undergrads. Sports was actually like this metaphor used to think about the way work was done. A lot of the work was team-based, but you’re also kind of talking about more informally how relationships happen in this industry and a lot of industries. It’s not just in not just tech, certainly. Where guys are getting together on a Friday afternoon to golf and then they talk about work. You know, they’re on the course for hours at a time or there was one man I interviewed who, as I mentioned, very intentionally joined a soccer league in an area where he knew he could connect with MIT graduates and Harvard graduates in Cambridge because he was hopeful that just by playing on a team together that… Work would inevitably come up and he could get the inside scoop on what their startups were doing. And so men were very quick to tell me, oh, we don’t network. I don’t network. But then they would list off all of these kind of strategic ways that they would socialize in order to build out friends and buddies who they could lean on when things got, when times got tough.
Juravich: So would it be a part of like the interview process where in the interview they would just casually be like, so what do you think the Boston Bruins chances are in the Stanley Cup next year, right? Just to see how the person would answer.
Mickey: And you know, women were highly aware of this. So some of the women would tell me that they learned how to do fantasy football, for example, because a lot of their clients and customers, they played fantasy football. So if they weren’t part of the league, they were missing out on all of this kind of socializing and bonding that happens. And a lot the networking, especially around sales and selling the tech products to clients involved going to sports games. Going to Red Sox games like late September rainy games and then coming into the office the next day So it the the sports culture was definitely infused in a lot of a lot this networking
Juravich: This is Tech Tuesday from all sides on 89.7 NPR News. We’re talking about the networking challenges that many women and minorities face in the tech industry, and we’re talking with author Ethel Mickey. You also wrote a paper called “The Game of Chutes and Ladders: Gender and Aspirational Recourses During the COVID-19 Pandemic.” It examined how the pandemic shifted career aspirations for working fathers and mothers. Can you talk to me about how working parents experience networking differently. Because I certainly don’t have time to sit at a bar for hours after work, ever. Continue. Yeah.
Mickey: Yeah. So, yes, my other kind of hat of research, I’ve been looking at the impacts of the pandemic on working professionals and just, you know, the time void that happened for folks juggling parenting with remote work and managing a crisis. But there were several women, you know, tech skews younger. So most of the folks in my study did not have kids and were not married at the time, they were in like their early 20s. Uh, old in the industry, at least in the company I was studying was 40.
Um, but there were some parents and the working moms, especially just had this kind of this ethos of like, God, I just, I don’t want to do this. I don’t want to go to a happy hour. I can’t, I have to get home. I have. You know, school pickup. I have do, uh, go to practice and the, and I want to be with my family. And that was like a trade-off and a lot of guilt around this too. It wasn’t just that they couldn’t. But that their priorities had shifted necessarily so. And so the women in this kind of situation were really strategic with their time.
So they would find networking breakfast, for example, they would try to meet people during the day and go to lunches, but those were harder opportunities to locate. And so they did feel like they were being left behind or left out. And the folks that they were working with, their coworkers, their team members didn’t really think about it. They would say, okay, five o’clock on Friday, like let’s all go bowling or go to the Red Sox game. And women had to say no. And they felt that they knew they were damaging their careers in that way.
Juravich: Yeah, I feel compelled to say, like, my networking would be, let’s talk about a six-year-old soccer game that I have on Saturday, right? I can talk sports. Let’s talk that. So, your book also has a chapter called “Down to Earth Regular Guys.” And in it, you say that tech workers must not only establish their competence, but also be culturally similar to the majority group. So, in this case, the white male engineers who are apparently under 40. So what, what standards are you talking about here? Because this feels like a fraternity sports team mentality to me. Tell me more about how you can try to be a down to earth, regular guy.
Mickey: So we know from a lot of social science research that we as humans are attracted to people who we view as similar to ourselves. And there’s inherently nothing wrong with that, right? You just mentioned that you would probably bond with someone who also has a six-year-old and going to soccer games. And then there’s something about that shared experience that creates connection. Now, the problem is, is when organizations use that idea of shared culture, shared backgrounds, shared experiences, to think about who makes a good fit for their company and who they should hire.
So the way that hiring was happening, people would very explicitly say to me, we’re looking to hire someone who doesn’t just have the skillset, but it’s someone that I wanna hang out with for 50 hours a week, someone I wanna go grab a beer with, somebody that I can be friends with. And so what ends up happening is that people wanna hire people who they think are like themselves. And so if it’s a company that’s 80% made up of white men. Then they’re going to hire more white men.
And other social scientists have been, management scholars have been writing about how the use of referrals and the use culture as a hiring mechanism is problematic. It creates this unequal kind of system of relying on networks and relationships and it’s subtle. And so what I’m talking about in the chapter that you referenced in “Networking to Nowhere” is that a lot of this is happening through comfort, through familiarity, through existing relationships. People wanna help people that they already know. They wanna be seen as doing favors, building trust. But the problem is that those relationships in tech are still heavily made up of men. So one man would tell me, I really love my team. We have a great fit. We have great vibe together. We collaborate really well. And then in the next sentence, he would describe it as a brotherhood. So he wasn’t intentional in excluding women, but even in the language he was using to describe it, the bond was around their shared gender and their shared identities.
Juravich: I wanted to ask you about one more chapter before we run out of time. It’s called “Men Schmooze Women Strategize.” And you give an example where a woman said she preferred not to know anyone at the company prior to getting the job because she wanted to make it clear she got the job on her own. But that seems to go against this whole networking idea. So what are women afraid of being like that they only got the job because they knew someone? Tell me about that.
Mickey: Yeah, so women spend a lot of time doing what I call strategic networking. And this is where they’re going out and trying to meet new people. And this through conferences, through events. They want to build relationships without leaning on their dads, without leaning on their husbands. They want you know, really prove that they have made it for their own worth, that they weren’t, you know like the token woman who was fired as a diversity higher. We see this also for people of color in tech who are are even more outnumbered than women and and then at the same time men spend a lot of their time schmoozing so so the the foil to women’s strategic networking, men are strategic socializing They’re spending their time not at the conferences, not at industry events, but instead reaching out to people that they used to work with.
Going to the bar, one man told me he does bar stool interviews where they sit at the bar together and he’ll conduct interviews there. And this really changed how I think about networking because it showed me that some of the most valuable opportunities were happening kind of outside the formal systems. One man told he had been in tech for 20 years and never formally applied for a job. Every opportunity instead came from people he knew. So who has access to these types of relationships? I found that it was men and I found that the way that they were building their relationships, but also maintaining them over time was leading to more opportunities. So it’s not that women are failing to network, they’re just doing more networking and getting less out of it. The most valuable type of networking in tech often doesn’t look like networking at all.
Juravich: Just to end on, what advice would you give to someone who may be struggling to network and is trying to get an in, whether it be in the tech industry or any other industry? Can you give some advice?
Mickey: Sure, I hope that workers realize that if networking feels like a second job, you’re not imagining it. But as someone who studies workplaces, I really wanna give advice to companies. I want the companies to change to address these really large scale issues. You can’t just tell people to network more. That’s not going to fix inequality. You have to change who gets access to opportunities in the first place.
The biggest lesson from my research is that careers are shaped by relationships but access to those relationships is unequal until we address that until we change the way that work is organized until we changed the way hiring is done that people are assigned to teams and projects, then inequality will still keep reproducing itself in ways that are easy to miss but incredibly powerful.
Juravich: We’ve been talking about why inequality persists in workplaces that claim to have fixed it with sociologist Ethel Mickey. Her new book is “Networking to Nowhere: How Gender Inequality Persists in Tech.” Thank you so much for your time today, Ethel. Thank you.
And coming up, we’re gonna talk about two recent lawsuits. One involves Elon Musk, the other involves Disney. That is when Tech Tuesday from all sides continues on 89.7 NPR News.
You’re listening to All Sides. I’m your host, Amy Juravich. This is Tech Tuesday, a show where we share stories about science, technology, and the future of our environment. In a $150 billion lawsuit against OpenAI and Sam Altman, Elon Musk lost. He lost due to the timeliness of the suit. To tell us more about the lawsuit and what it means for the future, we have Russell Holly, Director of Commerce Content at CNET. Welcome back to All Sides, Russell.
Russell Holly: Thank you.
Juravich: So Musk’s $150 billion lawsuit against OpenAI and their CEO was rejected by a federal jury. Talk to me about what was, before we talk about why it was rejected, tell me why was Elon Musk suing?
Holly: Elon Musk was suing because OpenAI is the single biggest competitor to the tool that he is building called Grok. And it’s, you know, not hard to see that that was the reason for this. He’s been very, very public about the approach that he has taken in order to, you know, build his own AI tools and see them as being, you know, competitive, both in image creation and in answering creation, you know, limited to Twitter, where OpenAI is available to everything else. But. Elon Musk was actually an early investor in OpenAI, and that was where a lot of the discrepancy here came, was that the promise for OpenAI’s entire platform was that it would be this kind of open public system, and it became quite private and profit-driven over the last couple of years. And so that’s the legal standing, but the justification that Musk made quite publicly was, you know, that it was. Uh, it was no longer fair, uh, you know, to have this competitor because the origin of that competitor was supposed to not, you know, by its very nature be competitive.
Juravich: Well, so no one ruled on that, though. The lawsuit was rejected due to the timeliness. Tell me more about that.
Holly: Yeah, so jury deliberation took less than two hours for a court hearing that had them sit and listen to arguments over the course of several days in which documentation was produced that made it very clear that the early origins of OpenAI was in fact to be this kind of open tool that others would have access to and be able to benefit from in their own ways. And to have it be this sort of central pool of knowledge for other organizations to tie into.
Then there was quite a bit in the way of conversations that were released around the time that Sam Altman was originally removed from the OpenAI board for issues that were encountered there and then reinstated as the head of OpenAI in this very kind of whiplashy kind of way through the help of Microsoft’s Satya Nadella. Over the course of several conversations were released, but the jury deliberation when it started ended very, very quickly because at the beginning of this is one very simple fact, and that is that Elon Musk was more than a year too late for this to be a meaningful lawsuit filed against OpenAI.
Juravich: So there was a deadline on claiming it wasn’t open or…
Holly: Well, sure, the contract that they had, you know, come into an agreement to the statute of limitations was expired by if memory serves more than 18 months. So.
Juravich: If he would have submitted this lawsuit 19 months ago or in a more timely manner, did Elon Musk have a case? Could he have won?
Holly: There’s a very real chance that some sort of settlement could have been found, you know, where a payout would have been made to Elon Musk, who is now very much a competitor for OpenAI. But that is something that would have had to have happened, like you said, 19 months ago. And he was very busy 19 months ago and focused on other things.
Juravich: I perhaps working for the federal government. I don’t know. I don’t. Yes, that’s right. Yes. Okay. Um, so open AI was planning an initial public offering to open the company up to like public stocks and the lawsuit was kind of a roadblock for them. So now that that’s resolved, does that mean that, you know, we’re going public and other AI companies might go public to
Holly: It is very likely that OpenAI will go public either by the time this summer is over or by the the time we get to around Thanksgiving is the speculated timeline among experts. It’s unlikely that the IPO will happen during the summer although not impossible, but given the… Very fast way that this lawsuit wrapped itself up, it’s entirely possible that OpenAI will decide that now is the appropriate time, given the additional attention that is currently focused on the company. But if you pay attention to financial experts who look at this sort of thing, it is speculated that it will be much later in the year.
Juravich: And what does it mean for them to go public? What does it means for the future of AI, the future technology? There’s a lot of money to be made here and a lot interest in this.
Holly: For OpenAI, it is a lot of money that would be very useful for leveraging those stocks in order to make good on a lot the promises that OpenAI has made, but not been able to deliver on. OpenAI had a pretty challenging 12 months in being able to meet the expectations that it set out for promises made to Microsoft and several other large companies. You start to see stories about Apple and Microsoft both distancing themselves from the relationships that they have made with OpenAI.
Apple is actively looking at using Google’s Gemini for its AI partner over OpenAI, and that is because the results gained from OpenAI are just not what was originally promised. It hasn’t made Siri smarter, it hasn’t provided additional information in a way that is useful to readers. Apple doesn’t exactly help in that case, but that is one of the biggest reasons that Apple is trying to move to Gemini. And that puts OpenAI in a pretty delicate situation where it can either start making good on some of the promises it has made to some of its businesses or it can go public and use the revenues gained from that in order to make other promises to other people in order recoup some of that loss.
Juravich: So they’ll make some more money, but they’ll also make more promises. So we’ll see if they can fulfill any promises. This is Tech Tuesday from all sides on 89.7 NPR News. We’re talking about some recent tech news with director of commerce content for CNET, Russell Holly. I wanted to talk about another lawsuit that happened recently. This one’s against Disney and the way that Disney uses facial recognition in its parks. So can you tell me about the facial recognition system and how it affects visitors to a Disney park?
Holly: So this is something that started rolling out to not just Disney parks, but also their their kind of mall adjacent things like Disney Springs and downtown Disney and even on some of their cruise ships, if memory serves, where there are now facial recognition systems that bypass in many ways the old like metal detection systems. So if you walk into one of these systems, there are there’s a camera array that you walk through and it does, you know, kind of scans of your body to make sure that you don’t have any dangerous things on you or things that are against the Disney Code of Contact when you enter their facilities.
But that keeps a record of you and that record is used to track where you are throughout the park at all times. The benefit to Disney is that it shows where the most popular areas are, which helped Disney kind of optimize against that and make sure that there is appropriate entertainment in the correct areas to make sure that that sort of thing. So the folks aren’t walking around, you know, bored at any point or to make sure that if there is inappropriate conduct that that person can be flagged and removed from the park and, you know, either banned permanently or, you, know, just sort of removed from the day depending on the infraction. But this is something that isn’t just new. It’s been a part of Disney parks in the U.S. Now for nearly five years.
Juravich: And so the lawsuit was filed by a woman named Summer Christine Duffield. And it was filed in a US court in New York. What is she claiming that Disney is doing, that she wanted to file the suit?
Holly: The plaint ultimately comes down to privacy. This is a federal complaint, and it accuses Disney of violating competition in consumer protection laws by not disclosing the use of this technology and the data that it collects. Uh, you know, so if you walk into a Disney park, you know, if you go to Disneyland right now, if you go, go to Los Angeles and go to Disneyland, uh, you know, when you stand in front of the gate to hand someone your pass in order to get in, the very first thing that happens is someone takes your picture. You are, you are instructed to stare directly at a camera and a picture of your face is taken and it is attached to the ticket that you have provided. Um, so. It’s it you know what they don’t make clear is the cameras that are available in the rest of the park that you know take that facial recognition information and uh you know kind of follow you through the day. There are that is the information that is you know kind of the the privacy issue there that this person is alleging.
Juravich: But there are other companies that do scan you. I mean, like the Intuit Dome, Dodger Stadium, are there lawsuits against them too? Or is Disney just, it’s throughout the whole park instead of like at Dodger stadium, it’s just at the front end.
Holly: Yeah, it’s not immediately clear why this is being limited to Disney. Universal Studios, they’re a brand new theme park in Florida, Epic Universe. They’re also quite clear when you come in that facial recognition is a part of what’s being used, and that is that Universal has been quite clear about this being used to identify what parts of the park are more popular on what days and why.
So, this information is fairly common. I think part of the issue from this particular consumer’s perspective is Disney had no public facing information about what happens to that data after it has been collected. Disney has since released statements that made it clear that this is something that is deleted within 30 days, except in cases that, you know. Legally require them to keep information on for maybe a crime that has been committed or, you know, a fraud complaint or something like that. And, you know… But it’s also true that this is facial recognition that applies to every single person in the park, and that includes kids under 18. So there is, you, know, parental awareness that has not been made there as well.
Juravich: I wanted to pivot, we only have a couple of minutes left, but we had one more topic. It seems like AI chatbots have been hallucinating more recently. So I’m not sure if it’s that we’re using more AI and so therefore we’re coming across more made up stuff or it is that the chatbot are hallucinating more. But there was a piece in CNET that said that AI was making up citations in research papers. So can you tell me more? Maybe first say what a hallucination is. I mean, it’s just AI making something up, right?
Holly: It’s exactly right. It is exactly what it sounds like. When we refer to an AI hallucination, we mean that we have asked an AI chatbot for a question, and the answer that we’ve gotten seems factual if you just read it. But if you follow up in any way, shape, or form, if it has cited a legal document or, you know, based on a report that happened in 2012 and it, you, know, provides examples, that that information just doesn’t exist, that there is nothing to base this on. And we’ve had stories like this over the couple of years.
There have been attorneys that have been disbarred because they have filed legal briefs that have included AI hallucinations that referred to cases that have never existed. There have been medical journals that have published and then retracted because the data sources that have been added in there just never existed. The question that you asked at the beginning of this was really interesting, because the idea that this is something that is getting worse over time is something that was predicted by AI scientists four or five years ago.
There was a paper that was put out that said that because AI learns from its users. So it just keeps, you know, it’s an ouroboros of stupid, kind of. It just kind of keeps, you know picking up on this information. And as soon as somebody takes bad information and shares it somewhere else, that information is now available for the AI to scrape as a fact source. So it is now fact checking itself based on data that it has made up. So it’s getting worse over time in some cases because it has no idea how to take information from the internet and distinguish fact from fiction.
In a lot of cases, Gemini started using Reddit as a fact-based source very early on, and that’s how we got to Gemini as an AI platform telling you how much glue was safe to put on cheese pizza in order to get the optimum cheese pull. There’s all kinds of little things like this, but this is the first study from Cornell and UCLA found almost 150,000 AI-generated citations across major four research databases that were all hallucinations, that did not exist at all.
Juravich: All right, we’re gonna have to leave it there and talk about AI hallucinations again next time. We’ve been talking with Russell Holly, Director of Commerce, Content at CNET. Thank you for your time, Russell.
Holly: Thanks for having me.
Juravich: This is Tech Tuesday from All Sides on 89.7 NPR News.