Latest / Bloomberg Talks / Cisco Systems President & Chief Product Officer Jeetu Patel Talks AI Industry
Transcript
- 0:02Speaker 1: Bloomberg Audio Studios.
- 0:04Speaker 2: Podcasts, radio, news. So this is the basic idea. Switches, routers,
- 0:10Speaker 2: wireless networks, data center, AI fabric.
- 0:14Speaker 3: How about that? It must be Cisco Systems. Good morning, Mr. Chambers,
- 0:18Speaker 3: if you're listening.
- 0:18Speaker 2: Jita Patel is Chief Product Officer at Cisco.
- 0:22Speaker 3: The Royal Bank of Canada says we don't trust our walls.
- 0:27Speaker 2: Is the next step in AI this massive security risk
- 0:31Speaker 2: that Paul Sweeney's checking account is unclogged.
- 0:35Speaker 1: It is, Tom, the central issue of our time at
- 0:38Speaker 1: this point because these agents that we are moving into,
- 0:41Speaker 1: so we're in the second phase of AI where you've
- 0:44Speaker 1: got this moment just like chat GPT that's happening with
- 0:48Speaker 1: personal agents. The difference between a chatbot and an agent
- 0:51Speaker 1: is a chatbot, you just ask it a question, get
- 0:53Speaker 1: an answer. With an agent, you actually give it tasks
- 0:57Speaker 1: and jobs it can conduct on your behalf fully autonomously.
- 1:00Speaker 1: And one of the biggest challenges you have with agents
- 1:03Speaker 1: right now are people are concerned about how much you
- 1:05Speaker 1: can trust these agents because there's a paradox of a
- 1:09Speaker 1: sort because with agents, they're not that useful if you
- 1:12Speaker 1: don't give it access to your tools, your email, your calendar,
- 1:15Speaker 1: your credit card. But the moment you give that access,
- 1:17Speaker 1: your risk goes up quite dramatically. And so we have
- 1:21Speaker 1: to make sure that we address the trust, safety, and
- 1:23Speaker 1: security issue.
- 1:24Speaker 3: Okay.
- 1:25Speaker 4: How do we do that?
- 1:26Speaker 3: And who does that?
- 1:28Speaker 1: Well, we are spending billions at Cisco. Cisco is one
- 1:30Speaker 1: of the largest security and safety companies in the world,
- 1:33Speaker 1: and we're doing that. But the way you should think
- 1:35Speaker 1: about how we do it is pretty important. The difference
- 1:40Speaker 1: is that these agents are like teenagers. They're supremely intelligent.
- 1:45Speaker 1: They have no fear of consequence, and they exercise poor
- 1:49Speaker 1: judgment far more frequently, Paul, than you and I would like. Now,
- 1:53Speaker 1: the way that you have to think about this is,
- 1:56Speaker 1: It's going to get harder and harder to distinguish when
- 1:58Speaker 1: there's a malicious attack that's altering the behavior of an
- 2:01Speaker 1: agent versus an agent just exercising poor judgment because it's
- 2:06Speaker 1: very fastidiously just following a goal that you gave it
- 2:10Speaker 1: to do. And so the way that you have to
- 2:12Speaker 1: do it is you have to monitor an agent's behavior continuously.
- 2:16Speaker 1: And when you start seeing drift and misalignment between human
- 2:20Speaker 1: expectations and the agent's behavior, you have to be able
- 2:24Speaker 1: to intercept right away. And so there's a whole set
- 2:26Speaker 1: of technologies around that. And then the other area that
- 2:29Speaker 1: people don't talk about as much is this notion of
- 2:34Speaker 1: your ability as an organization to efficiently manage and use
- 2:39Speaker 1: tokens is directly proportionate to your security posture, which means
- 2:45Speaker 1: the lower and the more efficient you are with using tokens,
- 2:49Speaker 1: the better your security posture.
- 2:51Speaker 4: Where is the average corporate American enterprise with AI just
- 2:58Speaker 4: in general, including using it wisely, protecting its safeguards, all that.
- 3:02Speaker 3: Where are we today?
- 3:03Speaker 4: It feels like we're super, super early.
- 3:05Speaker 1: We're very early. I think what has started getting diffused
- 3:10Speaker 1: is a lot of people within the enterprise now use
- 3:13Speaker 1: chatbots just like they use.
- 3:15Speaker 2: The search engine.
- 3:16Speaker 1: So that's happening at scale. But when you start thinking
- 3:19Speaker 1: about agents, we're still in the very, very early days.
- 3:22Speaker 1: And I think part of the reason for that is
- 3:24Speaker 1: just people being fearful of trust.
- 3:26Speaker 3: G.D.
- 3:26Speaker 2: Patel, have you put your checking account on Muse?
- 3:30Speaker 1: I have not put my checking account on Muse, but
- 3:33Speaker 1: I use Muse regularly. The reason I haven't put my
- 3:36Speaker 1: checking account on Muse is that's the area that I
- 3:38Speaker 1: feel like we need to make sure that we're actually
- 3:41Speaker 1: being careful before we go out.
- 3:43Speaker 3: And give too much access. Fancy Cisco phrases. John Chambers
- 3:46Speaker 3: would have never used these languages. Ancient hallucinations. Alert fatigue. Translated,
- 3:53Speaker 3: we don't trust the computer. That's all there is to it.
- 3:56Speaker 3: How does your industry, with great respect to Mr. Chambers
- 4:00Speaker 3: and all that he invented and sustained, I should say,
- 4:03Speaker 3: across some tough years, how do we get the trust
- 4:07Speaker 3: so we do transact given hallucinations and fatigue?
- 4:11Speaker 4: I think.
- 4:12Speaker 1: What ends up happening in this market is you have
- 4:15Speaker 1: to make sure that the safety and security technology is
- 4:18Speaker 1: moving and innovating as fast as the innovation in AI.
- 4:23Speaker 1: And once that starts to happen, what you start to
- 4:27Speaker 1: see is the comfort level goes up with people. And
- 4:29Speaker 1: that's just, it's a matter of time. And, you know,
- 4:33Speaker 1: I was just with some of the other folks on
- 4:36Speaker 1: a TV show over here with Bloomberg, and they were
- 4:39Speaker 1: asking me, like, who's going to pay for this? Well,
- 4:41Speaker 1: I think The revenue is there. If you look at
- 4:44Speaker 1: just what Anthropix revenues are, that's the end user revenue
- 4:48Speaker 1: that's being paid for. And then if you look at what,
- 4:51Speaker 1: you know, both Azure and Google reporter has revenue on
- 4:54Speaker 1: the cloud, that shows you that the infrastructure is also
- 4:56Speaker 1: getting paid for. What you need to have is the
- 4:58Speaker 1: trust and safety and security being, once that goes up
- 5:02Speaker 1: a notch or two, you'll start to see a massive
- 5:04Speaker 1: kind of upsurge that happens within usage.
- 5:07Speaker 4: I'm going to say maybe two years ago, I was
- 5:09Speaker 4: speaking with a management consultant and he was saying, His
- 5:12Speaker 4: clients every day are calling in saying, this is two
- 5:15Speaker 4: years ago, how do I implement AI into my business?
- 5:17Speaker 4: Is it a friend? Is it an enemy? And he says,
- 5:19Speaker 4: before we engage anybody, we have them sign a document
- 5:23Speaker 4: that says their C-suite, their board, commits to implementing AI
- 5:28Speaker 4: on a morally just basis and with all the guardrails
- 5:31Speaker 4: and things like that. That kind of blew me away.
- 5:35Speaker 4: But now it seems like that's kind of what you
- 5:37Speaker 4: have to do. Are you seeing corporate America think about
- 5:39Speaker 4: it that way?
- 5:40Speaker 1: You have to make sure that the way that the
- 5:42Speaker 1: guardrails and the value system that your organization has gets implemented.
- 5:46Speaker 1: And that is a prerequisite for success for enterprises, especially
- 5:51Speaker 1: the regulated ones. And so that's not really that controversial
- 5:56Speaker 1: or debatable anymore. That's what everyone does. The question is how.
- 5:59Speaker 1: And how do you implement them? Because these agents and
- 6:02Speaker 1: these models tend to be unpredictable, where you don't really
- 6:06Speaker 1: know how they're going to behave. And so you have
- 6:08Speaker 1: to make sure that there's a constant monitoring. And the
- 6:11Speaker 1: moment you start seeing drift and misbehavior in some ways,
- 6:16Speaker 1: you have to make sure that there's implemented guardrails that
- 6:19Speaker 1: actually stop it from doing something you don't want it
- 6:21Speaker 1: to do.
- 6:21Speaker 4: So when you see a company like Anthropic, file an S-1,
- 6:25Speaker 4: and then the risk section is the world's going to
- 6:29Speaker 4: come to an end, AI is going to destroy everything.
- 6:32Speaker 4: I can't even imagine being the lawyer of the banker
- 6:33Speaker 4: who was tasked with writing up that silly risk factor,
- 6:36Speaker 4: but it's not silly, is it?
- 6:38Speaker 1: Well, I think there is a considerable risk of things
- 6:42Speaker 1: going wrong. I think we have to make sure that
- 6:47Speaker 1: we're careful on not being hyperbolic about it either. So
- 6:52Speaker 1: while there might, two conditions can hold true. You can
- 6:55Speaker 1: have risk and humanity does not need to come to
- 6:57Speaker 1: an end.
- 6:58Speaker 2: Do you see a roll up here? You mean if
- 7:01Speaker 2: there's going to be transactions and combinations down the line, success, failure,
- 7:07Speaker 2: how do you see that unfolding into anything but monopoly
- 7:11Speaker 2: or even monopsony characteristics?
- 7:15Speaker 1: I think there's a number of, The number of players
- 7:18Speaker 1: that are participating in this market today is not one
- 7:22Speaker 1: or two. There's a fair number of players. Now, as
- 7:29Speaker 1: agents come about, will the business model shifts occur?
- 7:32Speaker 2: Absolutely.
- 7:33Speaker 1: Think about a retailer. In the past, a retailer would
- 7:36Speaker 1: typically go direct to a consumer, influence the consumer through
- 7:39Speaker 1: some advertising, and make sure that they can actually upsell
- 7:42Speaker 1: or cross-sell. Now you might have an agent that you
- 7:44Speaker 1: have to talk to. That agent aggregates the information across
- 7:47Speaker 1: a multitude of retailers and then provides an algorithm to
- 7:50Speaker 1: give it back to the consumer. That changes the business model,
- 7:53Speaker 1: and we have to make sure that we all think
- 7:54Speaker 1: through that.
- 7:55Speaker 3: Thank you so much. Greatly, greatly appreciate it. With Cisco
- 7:59Speaker 3: this morning, Jita Patel.