Four Battlegrounds: Power in the Age of Artificial Intelligence

Product Information

Leading companies such as Microsoft and Google are pursuing a very cosmopolitan strategy that involves US and Asian researchers cooperating toward shared goals. A generic anti-military stance among tech workers has ensured that those companies cooperate less with the US military. (It almost sounds as if Scharre is complaining about capitalists for subverting US imperialism.)

Well, it's a real challenge. One of the problems with AI systems is their intelligence can be very narrow, and so they can be very good at a task that they were trained to do, but if the environment or their context for use changes ever so slightly, their performance can drop off fairly dramatically. So one of the earlier versions of Alpha Go reportedly, if you slightly change the size of the board, its performance will drop off. It didn't have the ability to generalize its knowledge about Go even to just a differently sized board. That's a real problem when you think about, for example, a military environment where we don't get to control where we're going to operate or who we're going to fight against, the enemy's not going to give you a preview of their tactics and the enemy's adaptive.Paul Scharre, Four Battlegrounds: Power in the Age of Artificial Intelligence (New York: W. W. Norton & Company, 2023). China has things we simply don’t allow here, like facial recognition software everywhere. Do lax policies like that give Beijing a leg up? This book really covers a ton of amazing topics and really gets some great arguments and key points about the use of AI and the potential future we all face

Pilots take some heat for this in the military because automation in drones has been able to physically replace what pilots do, in terms of enabling remotely piloted aircraft. Over time, with more autonomous features, it’ll increasingly hand over the piloting of the plane itself so that pilots are in supervisory capacity, which is a good thing from a military standpoint.And so obviously any human would notice these things. The human would see the box moving and be like, "Oh, there's somebody under the box," but the AI wasn't trained on that. And so that is a real problem when we think about AI being used in competitive environments because it can be so easily manipulated and people are clever. Our adversaries are clever, and that's a limitation when we think about how we're going to use AI in the real world. It’s probably a relatively level playing field between the U.S. and China, and what’s going to matter more is whether companies or government institutions have the ability to refine the data that they have and apply it toward machine learning applications. Four Battlegrounds: Power in the Age of Artificial Intelligence" is an exceptional read that explores the intricacies of technology allocation as a political action. The author's ability to demystify the workings of AI for the average reader is commendable, making complex concepts accessible without compromising on detail. In particular, head-to-head gunshots are actually banned in training by human pilots because there's a high risk of collision if the pilot is trying to maneuver the aircraft when you're racing each other hundreds of miles an hour. And it's extremely difficult to do in any case and requires superhuman levels of precision, but none of that was a problem for the AI agent. It can achieve these split-second shots while also avoiding a collision. And the AI agent learned to do this all on its own — it wasn't trained to do that. The AI system that won was trained in a simulator and had over 30 years of simulated flying time and this was one of the things that it simply learned on its own from all of these years of simulated dogfights.

Pilots should be worried because AI is coming for their jobs. The reality is that, like in many areas, AI is going to offload tasks that people can do, and there’s a lot of advantages for that in the private sector or in the military. Reminder: The Phrase "No Evidence" Is A Red Flag For Bad Science Communication.) But he keeps those biases separate enough from his military analysis that I don't find those biases to be a reason for not reading the book. But that depends upon other countries agreeing to go along with the United States, particularly I guess the Japanese and the Dutch. The Dutch company ASML is on the cutting edge of lithography, the equipment used to cut these incredibly small transistors.So lets talk about how one succeeds in this AI space. You argue that there are really four main ingredients to success in AI. Walk me through them. What about concerns I often hear, Paul, that that's actually a double-edged sword, that Chinese talent is coming to the United States learning science here in the United States, getting access to cutting edge technology that then gets returned to China even if those researchers themselves don't relocate to China? Okay. So let's talk about computing power, or I guess the term of art is compute. It's a little awkward to talk about it in that way, but let's talk about compute. Who has the lead when it comes to compute?

We need to focus on the things that really matter. The two things that matter where the U.S. has a huge asymmetric advantage over China are hardware and talent, and we need to find ways to maintain those advantages and to double down on them when we can. So expanding high skilled immigration, making sure that we're bringing the best and brightest from around the world, and keeping U.S. companies in the lead in some of these critical choke points for semiconductor technology. We don't need to compete on everything with semiconductors with China. We don't need to worry about trying to ... Competing at every single kind of chip at every component. What we need to focus on is maintaining a hub of leading edge manufacturing here in the United States so that we can have U.S. companies at these critical choke points, so that we can control China's access to the most advanced hardware in the future. It is a real threat, and it's really essential that the U.S. manages this in a way that we continue to be an inviting place for talented scientists and engineers from China and other countries to come to the U.S. and to stay here to study, to start their lives, to found a business. That's really core to sustaining American competitiveness. At the end of the day, the U.S. as a country of 330 million people is always going to be restricted against a country of 1.4 billion if we restrict ourselves to homegrown talent. If we tap into the best and brightest of the 8 billion people around the world, the U.S. will have an advantage that China cannot match. Over time, regulation in some fashion of AI technology; probably much of which will be sector-specific. The regulations for AI in medical applications will fall within the sort of the broader paradigm of how we regulate tools and safety for medical devices, and the way that we regulate AI in finance applications will fall from how we regulate trading and other financial things. But I do think that AI doesn't get a pass on regulations and it's worth reflecting on the fact that we only have clean air and water, and safe highways, and safe air travel, and safe food to eat in America because of government regulation of industry.So the third of your four pillars of power in an AI world is talent. Where do we stand in the talent competition? Where is the talent? Where might it be going? But as I think about this, and you've touched on this in your last book, as you start to rely on systems that can in essence, outthink, out anticipate, out act human beings, you can potentially create a whole set of pitfalls. I think you have written about how AI can be simultaneously transformative and brittle, and it can do things that you don't. How do you think about those trade-offs?