Katherine Forrest: Well, I just sort of thought that this was a time when we just have to alert people to some things that are happening out there, because we’re not going to be able to dive into any one of them deeply. But there are a couple of things that are really happening fast, and we don’t want our audience to walk around to their cocktail parties for the weekend and like not know what’s really happening. So let’s try to name a few of them.
Anna Gressel: Okay, we’ll do it like a game. Do you want to go first, Katherine?
Katherine Forrest: No, you go first.
Anna Gressel: All right. So I am going to pick the announcement from OpenAI this week about their AI agent. It’s a ChatGPT agent mode and it’s deployed and it can do real life tasks for you. And I think people have been waiting to see this. So it’s finally here.
Katherine Forrest: Right. I mean, it’s really interesting if you go onto YouTube right now and you look at some of what people are doing with this. So we’ll talk about that more in a minute. But I want to talk about,—and we’ll both talk about,—the recent White House announcement of an announcement. You got to love it. You know, it’s an announcement of an announcement. That’s going to come out on July 23rd. So it’ll come out like right after or right around the time that this episode airs. And Trump is going to lay out a speech that they’re characterizing as a plan that the White House says is for, “winning the AI race.” So we don’t know a lot about it, but we’ll tell you today what we do know.
Anna Gressel: And I think we’ll explore what that means for folks. And one thing I think we should also talk about, Katherine, because it came out today, is the guidance from the European Commission on general-purpose AI models. And that’s in addition to the General-Purpose AI Code of Practice, which came out last week. And so we’re seeing companies sign on or not sign on to that. The guidance on models is different and more generally applicable, and we’ll talk about that too. I think we’ll do it in a future episode. But also, can we pivot and do a little on quantum? I think, you know, we should talk a little bit about the June announcement that the leaders of G7 set out, which had a vision for the future of quantum technologies.
Katherine Forrest: Right, right, right. And so we have to, we’ve got to stop there. So there are so many more, like I also am dying to talk about this paper that I was just reading yesterday that came out. It’s called “Chain of Thought, Monitorability…”
Anna Gressel: You and I were both reading that.
Katherine Forrest: Yes. You know why? Because Keith Richie, our extraordinary researcher, had sent it around. But it’s got lots of folks signed on to this from METR and Anthropic and OpenAI and Google DeepMind and Redwood Research and Magic and Amazon and Scale AI and the UK AI Security Institute and the University of Montreal. You’ve got Yoshua Bengio and Geoffrey Hinton and Sutskever. Anyway, it’s a really interesting paper.
Anna Gressel: Katherine, I interrupted you though while you gave the title. Do you want to tell people the title again?
Katherine Forrest: Oh, yeah. “Chain of Thought, Monitorability: A New and Fragile Opportunity for AI Safety.” And you can get that, actually, on arXiv, which is that incredible repository from Cornell University. So, all right, we’ll stop there. And let’s then, go back now and do a little bit on some of these little items that we’ve sort of dropped in as topics of conversation. We can start with the OpenAI agent mode.
Anna Gressel: Yep. So back in January of this year, OpenAI announced Operator, which is an AI agent that could access the web and accomplish certain tasks. And that was available to pro users. So those are the folks who have paid like for the highest level of public access.
Katherine Forrest: Right. And some folks would recognize that as some of the deep research capabilities, but it could do more than that. And they were incredibly exciting. But now, as of just a couple of days ago,—we’re recording this on July 18th, and it won’t air for a week. So, you know, it’s a little bit…we’re going to talk about things two days ago that are going to actually be like eight or nine days ago. But on July 16th, ChatGPT announced its agent mode, which is actually able to do exactly some of the examples that Anna and I have been talking to audiences all over the country and the world about—what AI, agentic AI was going to be able to do. And we’ve often used travel and booking travel as,—you know, who has a landline anymore? And nobody has a landline. Except for me in Maine. And so that’s…what that phone was.
Anna Gressel: It’s usually me calling you on the landline, Katherine.
Katherine Forrest: There’s only two people. I won’t even mention the other person, but it’s another person from work. That’s it. It’s like, I know it’s one of the two people. But in any event, sorry about that interruption. So the examples that you and I have given so often have been about travel and booking travel because that’s like an accessible idea for people, which they often do on their computers. And what this agent does,—this ChatGPT agent mode, as they call it,—is you enter a prompt and you might say, “hey, let’s, you know, plan for me a weekend getaway,” just as a prompt. And it will go off and it’ll compare calendars, it can actually look for itineraries. It can actually then book it all for you.
Anna Gressel: Yeah, and the agent can like click around your computer using your browser and any other information you might use yourself every day.
Katherine Forrest: Yeah, like you might have a pre-populated method of payment which you can use.
Anna Gressel: Yeah, and then it can find really fun and interesting travel options for you and the best price that you’re looking for.
Katherine Forrest: Yeah, you know, this is actually really big news for the travel agency industry or the travel industry sort of writ large, but for so many industries as well.
Anna Gressel: Yeah, this has really big implications, not only for the travel industry, but really for any website that requires people to navigate around and look at different things and look at different options. And that’s because we really rely on human eyes to see advertising. So that’s very tied in with the advertising industry.
Katherine Forrest: Right, I mean, what we’ve got now are a bunch of websites. Let’s just take a travel website as a, for example. And when you go onto the website, you might see different offers for rental cars or different kinds of hotel options and things like that. And the AI is just not as impulsive as we are. And their eyeballs aren’t really eyeballs. And so the implications of this for not just the travel industry, but for any website that’s relying on advertising is really, I think, going to be interesting and important for us to watch, and then to watch what the reactions are and the changes are that are coming.
Anna Gressel: So, Katherine, you’re basically saying I should use an agent if I don’t want to get tempted to buy skincare products while I am browsing the internet.
Katherine Forrest: Wait, when you’re looking at travel, do they sell you skincare products?
Anna Gressel: Oh, yes.
Katherine Forrest: Because that could only mean that you are looking for skincare products like all the time.
Anna Gressel: It’s my favorite impulse buy. I mean, I think like my friends know this, but I will take skincare product tips from any one of our podcast listeners because I find it to be so fun. So…
Katherine Forrest: All right, well there you go.
Anna Gressel: I probably won’t use an agent to circumvent my own [impulse purchasing].
Katherine Forrest: But that’s an example...as we know, with the algorithmic advertising that occurs these days with websites, you really are looking at—the humans are looking at—some of the advertising that’s both on the side and the banner advertising. And the AI is going to be navigating around that. So we’ll have to see how that develops. And the agents really have a toolbox. This agent mode has a toolbox of different agentic tools that it can use. And it’s not just limited to the travel industry.
Anna Gressel: Right, so as an example, OpenAI gives on its website, for some of our financial services listeners, they might say, “compare three competitors and give me a slide deck.” And you can also ask it to create graphs of competitor earnings and compare earnings announcements or look at personnel turnover, give you a chart, stock price analysis, so much.
Katherine Forrest: Right, and, this is really going to be an interesting moment as other companies now race to launch and deploy even more agentic tools. I mean, there’s already a bunch of agentic tools out there, but they’ll be now—I think we’ll see a whole bunch of them. But let’s move on to topic two, which is the upcoming announcement from the White House,—at least as of the day of the recording of this—relating to their AI plan.
Anna Gressel: Yeah, there is not a whole lot we know about it except that President Trump will be laying it out in a speech and it will contain a series of executive orders that are being positioned as enabling the US to like “win the AI race.”
Katherine Forrest: Yeah, you gotta wonder what that means.
Anna Gressel: I think it could mean a lot of things. ’You know, it’s actually, it’s been really consistent with the way that the Trump administration has thought about AI, which is as an enabler of national competitiveness, but also as a way to mitigate geopolitical risk. And so there’s kind of a national security element to that. It’s a competitiveness element to that. But there’s also a workforce and kind of infrastructure element to that. So it may have to do with opening up energy and land opportunities for AI infrastructure development. And Trump was just in Pennsylvania talking about the importance of AI. So you can see this has a major kind of tie-in to some of these areas that have been traditionally invested in other energy-producing technologies and how that might work for AI. Potentially, I mean, who knows? We’ll see what it says.
Katherine Forrest: We’ll see what it says. But just a few days ago, it just sort of ties into what you were saying, the White House announced a $92 billion commitment to various kinds of energy initiatives that are tied directly to AI. And, you know, we don’t know what’s going to happen with this, but it’s going to be a whole plan. It sort of comes out of things that were announced that the White House, was going to be doing starting in January. And I predict,—and I think a lot of others are predicting, it’s not just me,—that there’s going to be a lot of loosening of restrictions on how and where American companies can engage in different kinds of infrastructure development and what kinds of impediments may be removed, I think, more than imposed on them.
Anna Gressel: Yeah, and there’s an expectation in the press that the announcement will also advocate for American leadership and standard setting. And don’t forget that, you know, for our listeners, some may be like “live and breathe standards,” and some may not. Standard setting is often where a lot of the details of requirements of products and services are hammered out in order to kind of create consistency in the market. But then there is often kind of a very strong importance placed on national leadership in standards bodies, and we’ve seen that over many, many years in different standards contexts.
Katherine Forrest: Right, it really is a big deal. Let me just give an example which has nothing to do with the AI world or even the high-tech world, but just to give people a real life example of why you need standards. When you have, say, an infrastructure,—let’s just take the train rails—back when trains and locomotives (as they called them at that time) first came out, different states and actually different companies, it was done by companies, would have different size tracks and different gauge locomotives. And so when they would hit another company’s moment of transfer, they actually sometimes would actually have to change trains because the gauge was different. And so there was then eventually an attempt to, and a successful attempt, to standardize the gauge of tracks and then therefore the gauge of the rail cars. And it allowed for America then to have cross-country infrastructure to have rail cars go across the country. And so if you analogize that to AI and to sort of agentic AIs,—we’ve been talking about it, needing with the baton handoffs between one agent and another,—having a standard as to how that’s done will be very important to cybersecurity, to tools working together, etc.
Anna Gressel: Yeah, and the same is true for AI embedded or embodied, as we might say, in any kind of physical infrastructure, critical infrastructure. So we’re going to start seeing standards come up, kind of at the intersection of the hardware and the software, which is going to be a completely fascinating set of developments. I mean, it’s going to be years off for those to be finalized, but the effort to understand and kind of map that terrain is already underway in Europe and at major standards bodies. So American participation in that, it’s understandable why that might be a priority.
Katherine Forrest: Right, right, right. Now let’s turn to our last, and it’s actually a pretty meaty topic, but our last topic for this morning, which is the G7 commitment that they announced in June of this year, 2025, to quantum.
Anna Gressel: Yeah, so the G7 announced a commitment to coordinating on the development and responsible use of quantum technologies.
Katherine Forrest: Right, they put out this joint statement in June, actually in the middle of June, that recognizes the potential impact of quantum computing on really all aspects of society once it’s commercialized. And there are still some impediments,—which we’ve talked about in some of our prior episodes—to that, but there’s a lot of progress being made. But there is a commitment to trying to talk about and work together on aspects of quantum computing that will impact national security, economic growth, and things of that nature.
Anna Gressel: Yeah, and as part of that announcement, the G7 discussed a working group, a joint working group, which has like a real focus on cybersecurity and international collaboration.
Katherine Forrest: Right, and part of the reason that this is coming up now is—we know from those prior episodes that we did on quantum, and there are two for our listeners who want to look back through our sort of, index of episodes,—that when you combine quantum computing and the power, the enormous power of quantum computing, with AI’s potential, and you put those things together, it can turbocharge. It’s a theory that will be realized. It can turbocharge AI progress in ways that we can’t even imagine and make advancements in AI, which we’re already predicting to be extraordinary, to be even more extraordinary.
Anna Gressel: Yeah, I think it’s so exciting. And you know, if you remember, with respect to the wider global stage, we mentioned before that the UN has declared 2025 the International Year of Quantum Science and Technology. And this is not a PR move. I think it’s important to recognize that it’s instead a recognition and statement about the importance of quantum science, which is central to everything from health and energy to climate and economic growth. And the UN’s goal is to really raise awareness of that and encourage collaboration across borders. And it’s really a call to action for scientists and policymakers to get involved and shape the future of this technology, which is going to be so, so critical and change so many things in different fields and different countries. And so they’re all looking ahead to that.
Katherine Forrest: Right, exactly, it’s a realization that we may be on, really, the cusp of revolutionizing a whole variety of fields. And, you know, there’s some real practical implications. You know, you can imagine a world where drug discovery takes days instead of decades, and because a quantum AI can simulate molecules that we have no chance presently of simulating using classical computers, we can really, really speed that drug development, or where financial institutions can model risks across thousands of assets in real time that might provide more predictability and stability. Of course, there are other risks and other concerns about some of that in terms of what if there are errors early on, etc., or malicious use. But for the moment, let’s focus on the positive.
Anna Gressel: Yeah, I mean, supply chain is also a super interesting area that could be right for disruption. It already actually has been right for disruption when it comes to AI and agents. But if you think about adding quantum AI into the mix, we could imagine like a global logistics network that could potentially reroute itself instantly around disruptions like a traffic jam, a hurricane, a pandemic. And the reason it could potentially do this is because quantum AI would be running millions of what-if scenarios in parallel, and that leverages those capabilities of quantum computing that we talked about in prior episodes. But, you know, if those scenarios can be run all together, you can get to that answer faster. And scientists could also use quantum AI to create new materials with properties we can’t even imagine today, designed kind of at the atomic level. So it’s a super interesting area to think about really supercharging what we already see as the promise of AI today, because quantum adds a whole new layer of being able to do those really sophisticated calculations almost instantaneously because we’re actually leveraging the properties of the quantum computers themselves.
Katherine Forrest: You know, still there are going to be some big challenges, as we’ve talked about in the prior episodes. We still have to have a company that’s actually able to create a commercial application for quantum computing. These things are still, and they always will be, incredibly sensitive. The issue is going to be how they can actually stabilize them. And so we don’t yet have that winning implementation of quantum computing, but the momentum is real, and the G7 announcement just shows you really how people are now paying attention to this.
Anna Gressel: Yeah. And so to sum it up, quantum AI is the marriage of some of the most powerful technologies of our time, which is quantum computing and AI. And it’s not just about like faster computers or smarter algorithms. It’s about changing the fundamental way that we’re solving problems or making decisions. And that is super exciting, and it’s likely to affect the way that we even think about reality itself.
Katherine Forrest: Reality itself. Wow, that’s—that’s some big, heavy stuff to end the episode on, right? But that’s what we’re going to do. All right, so there’s been a lot happening. We’re going to talk about some of these things again because we’re only able to touch on them, and we want to look at the White House plan when it comes out after July 23rd and talk to you all about it. But for today, that’s it. We’re all set, and thanks for joining us. I’m Katherine Forrest.
Anna Gressel: I’m Anna Gressel, and feel free to send us your favorite quantum computing joke or invite me to visit a quantum computer in my puffy vest. I am still waiting for my invitation to the biggest and coolest quantum center someone can come up with in our listening base. So subscribe, like, and we will see you next week.
