Subscribe
Share
Share
Embed
What's your AI success story? What made you realize you had to change? And what does a hopeful future with AI look like to you? AI Hope with Geoff Livingston asks three simple questions of people doing the hard work of adapting and thriving in the AI era.
AI Hope with Camille Stewart Gloster Interview
===
[00:00:00] Hey everybody. My name is Jeff Livingston, and this is AI Hope. My book, Now Is Gone, is about never-ending change. This podcast is about what comes after, the stories, the people, and the reason for hope.
Geoff: ~Hey everybody, it's Jeff Livingston and I'm with AI Hope ... Excuse me~ Hey, everybody. This is Jeff Livingston and it's AI Hope. And I am here with Camille Stewart Gloster, best-selling author of-
Camille: The Insider You Built. That's right. Thank you for having me.
Geoff: And, uh, behind us is a poster with the book, but I will, ~uh,~ make sure to include all the links and all the information for both her consulting firm and the book- Yeah
Geoff: in there. ~Um, ~before we get going, you and I are both in this sector a little bit, but I don't know if people know who you are, my people. Okay. So I, I wanna give them a little bit of a background. Can you talk to them about your, ~uh,~ prior government experience before going out on your own?
Camille: Yes. So now I am the CEO of CIS Strategies, a cybersecurity and emerging tech advisory firm.
Camille: I started this after leaving the White House as the first deputy national cyber director, focused on [00:01:00] technology and ecosystem security, which is a long way to say that I got to focus on emerging tech like AI and quantum, supply chain security, future of work, all the future resilience things. And prior to that, I was the global head of product security strategy at Google, and I've worked-
Geoff: Okay, badass.
Camille: I've served two presidents.
Geoff: Dropping things here.
Camille: And, um, yes, yeah. And I just, I mean, this is an issue. Emerging tech and its intersection with cybersecurity is an issue I've worked on a long time.
Geoff: Very good. And by the way, we agreed we were gonna do this, ~uh,~ subway take style. There you go. So I hope you folks are enjoying this.
Geoff: So this podcast, we talk a little bit about, like, the positive, ~uh,~ and like what AI can do, but obviously, ~uh,~ you're helping people prevent major issues from happening. Before we get into some of that, what was the moment that you said to yourself, "Oh my, this AI stuff is something I need to keep my, ~uh,~ finger on"?
Geoff: What was your kind of like epiphany moment, if you would?
Camille: Oh, I mean, you know, AI has been around a long time, so my entire career [00:02:00] we've been using it in cybersecurity for anomaly detection and a number of other things.
Geoff: But- Role-based access, everything, yes,
Camille: yeah ... yeah, exactly. So the moment though that ChatGPT was released and my mom called me to be like, "What's this ChatGPT thing?"
Camille: I was like, okay, generative AI in the hands of everyone is something that we really will have to pay attention to because it's game-changing.
Geoff: Right.
Camille: ~Um, ~especially as we saw the capabilities improve so rapidly. And when I transitioned out of government and started helping companies adopt AI, I realized just how much support they needed to actually achieve their goals in terms of operational reliability and business value.
Camille: They were just kinda sprinkling some AI on things, which is not the way to do it.
Geoff: Yeah. Let's put a little chrome on that.
Camille: Yeah. Yeah. You know, people love the sound of a glitter party where you just, like- ... throw glitter everywhere. That's so true. But if you're the person who has to clean it up, it is literally the worst, and that's kind of how I feel about AI right now.
Geoff: Yeah, I mean, I personally make a joke every time I talk to somebody about implementation and I say, "I'm really an AI therapist- ... to kinda help you work [00:03:00] through some of these issues," right?
Camille: Yes.
Geoff: There you go.
Camille: It's so- Everybody's got just a little bit of PTSD.
Geoff: Right. They're definitely traumatized. Yeah. I think so.
Geoff: Oh, my God. Yeah And I, and I don't think the guys at Silicon Valley are helping at all with some of their, um-
Camille: They're not thinking about how we are processing data.
Geoff: Bombastic statements maybe? Yeah. Did you ... Oh, by the way, before we get going, this, this is being recorded right after the Anthropic, ~uh,~ filing.
Geoff: Did you have- Oh, yes ... any hot takes on that that you wanted to share?
Camille: I mean, I think you should read it. I think you should understand what they ... The, what is ... How many pages was it? Like, 300 pa- Oh ... something very long. There were some thoughts in there, man. I mean, it's a lot to read. Yeah. But you should
Camille: There's a lot of synopsis.
Geoff: Yes.
Camille: Go understand what it is that they think is coming and how they frame the reasons that they should go public, and the risks that they're anticipating. I mean, it will impact your life for sure.
Geoff: Very good. So what is your favorite AI use case that's really kind of, like, made your world better?
Camille: So I have created [00:04:00] basically an AI chief of staff that, ~um,~ draws on a few data sources to give me a task list, a view of the day, ~um,~ prep materials, and kind of just a different look at my calendar that helps me show up prepared that I really enjoy.
Geoff: That's awesome.
Camille: But in using it, I have been very thoughtful about what tools it has access to, and it does not have access to my email.
Geoff: Right. It's interesting. The connector issue is a massive one-
Camille: Yes ...
Geoff: for enterprises, and they're always worried about these rogue agents. And I, I guess we're gonna get right into the book pretty quickly here. The rogue agents issue or data leaks. Yes. And, and they never really think through the controls or the systems and how they're allowing people to basically act without thought.
Geoff: And then on top of it, there's no, like, real training-
Camille: Yes ...
Geoff: that's going on.
Camille: Yes.
Geoff: And that's a huge issue, too. And I actually ... My biggest pet peeve ... So this is gonna be our hot takes here.
Camille: All right.
Geoff: My biggest pet peeve, so I'll, I'll lead [00:05:00] by example, is executive teams that give directives on AI but don't know what they're talking about.
Geoff: They don't train themselves. What's your hot take on, I guess, maybe whether it's enterprise or organizational adoption of AI?
Camille: Yeah. I mean- Agents in particular are akin to hiring a new employee.
Geoff: Right.
Camille: Because they, while not people, and I don't want you to anthropomorphize them, are actors in your organization.
Camille: They have access-
Geoff: No parasocial? No
Camille: parasocial? No. Yeah. They have access. They have authority. They are making decisions on your organization's behalf. They're taking actions on your dec- your organization's behalf. And you've kind of been like, "Here are the keys to the kingdom," in a way you wouldn't do for an intern, much more the level of access you've given these tools, and, uh, then are surprised when they're not getting value from it, when it does something they never expected.
Camille: The list goes on. That is my big pet peeve. I'm like, just a [00:06:00] little investment upfront in understanding the capability and mapping it to your goals and understanding what it... Not only what its capability is, which is where we all seem to be focused, like, ooh, it's capable of doing these things, but what should it be doing?
Camille: What should it not be doing?
Geoff: Right.
Camille: Like, what are the boundaries of where we want it to take action on our behalf?
Geoff: So I am one of those people that uses that intern analogy. Yeah. And so to me, it's like the world's smartest intern, but like any intern, it can go off the rails. And since we just spoke with marketers, when I talk to marketers about it, I'm like, "If you ask an intern to write you a story about cheeseburgers, it's probably gonna...
Geoff: The intern's probably gonna g- come back to you with a bad story, like a crazy story with all sorts of weird use of cheeseburgers. If you ask an AI agent to write you a story about cheeseburgers, you're probably gonna get a ring on your doorbell in 10 minutes with DoorDash showing up with a bag of cheeseburgers."
Geoff: And part of that, and like, I didn't ask for actual cheeseburgers, I wanted you to write a story. And, and I think so much of that has to do with- Several [00:07:00] aspects. One, not understanding your own workflow.
Camille: Yeah.
Geoff: So the specificity necessary to tell the agent what to do-
Camille: Yes. Yes ...
Geoff: is missing, and then also the human checks are missing.
Geoff: So, like, if you just let this thing go off the rails, and I think in particular with LLMs-
Camille: Yeah ...
Geoff: right? Like, this type of AI is probability-based, so once it makes one error, if you have too many steps in a row, it's c- creating compound errors.
Camille: Yes. So you're very generous to say it's the world's smartest intern.
Camille: I wouldn't even go that far. I think it's just like any capable intern in that it's smart enough to be dangerous. Like, you took some classes, and now you think you know the subject.
Geoff: Right.
Camille: And so you show up, and the first answer might be spot on, 'cause I saw it exactly like that in the textbook. But as you get deeper, or even to your point, as you ask more complicated things, the r- responses don't compute.
Camille: Like, they- they don't actually meet the mark in all of the ways. So maybe the essay actually does kinda [00:08:00] sound, to sound good in the first and second paragraph, but by the third paragraph, it is wildly different from what you thought or just has some factual inaccuracies. It is literally predicting the next word.
Camille: So if you, if it lacks that specificity, if it lacks the, not only what you should do but what you should not do in the instructions, it might predict things that are gonna sound good to you. "Well, I know you like burgers, so I'm gonna make sure that burgers sound like the best thing in the world," rather than giving a balanced argument about, I don't know, maybe if you've got heart disease, you shouldn't eat a burger.
Geoff: Well, burgers are kinda like pizza, so don't put pineapple on them in
Camille: general. Yeah, ew.
Geoff: Anyway.
Camille: I don't
Geoff: like popcorn. I don't know where that came from. I'm sorry. But we're having a good time. ~Um, ~last question for you. Yeah. What's your hope for this AI revolution? Like, where can this take us if it goes the right way?
Camille: Yeah. Well, I really want people to feel empowered because the decisions we make every day, how we show up in our organizations really will determine where it goes. And so when we think about the [00:09:00] spectrum of realities or potential realities that we've heard, it's everything from, like, Terminator to this tech utopia where they've cured cancer and all the things.
Camille: I hope we land closer to that. We're gonna land somewhere probably in the middle, and I hope it's in the middle but towards this tech utopia where we have harnessed it in a way that it helps us solve complex challenges like climate change and h- heart disease and health outcomes, that it's helped us be more efficient so that we get time back to be human, ~um,~ and that we understand it enough to make informed decisions, risk-based decisions that align to who we are as individuals, kind of like we do with everything else.
Camille: When we think about our physical safety, you decide whether or not you're gonna do something based on your skills and ability. Am I gonna, you know, go skiing? No, I'm really clumsy versus someone else who is. And neither of those is wrong.
Geoff: I had you picked off as a snowboarder anyway.
Camille: I mean, yeah, [00:10:00] maybe. I don't know.
Camille: We'll try. But I feel like people are making those decisions very easily in other aspects of their lives, but with technology we just kind of take it, and e- with AI even worse. And so I just want people to be more intentional, and I think it would lead to better outcomes.
Geoff: Very cool. ~All right. So Camille Stewart Gloster, author of The AI Insider You Trust.~
Geoff: ~Let's find her stuff online. Get the book, follow her. Brilliant mind. Thank you for joining us ~
Camille: ~today. Thank you. Thank you for having me. Theinsideryoubuilt.com. ~
Geoff: ~Ooh. Let's rewind that, uh, exit. Sorry about that. ~
Camille: ~That's okay.~
Geoff: So Camille Stewart Gloster, the author of The Insider You Built. Brilliant mind, consultant.
Geoff: Follow her, get the book. And with that, thank you for being here.
Camille: Thank you so much for having me.
Geoff: Appreciate it.
Camille: Great job