“AI is going to accelerate the military-industrial complex. It’s not just good for the economy or academics, but also for national security. That's going to be a very interesting fusion. There may also be multiple attempts by states to build out those AI hubs over a 20 to 30-year timeframe, so we might see more subsidies, education, advocacy and championing for folks to learn AI.” Jeremy Au
In this episode of BRAVE, Jeremy Au reflects on deep learning breakthroughs’ timeline, AI’s impact on military technology and the increased incentives for more AI engineer training, funding and integration, based on a ChinAI article written by Jeffrey Ding.
Jeremy Au: (00:50)
Last week I was walking around the reservoir, with a technical founder and CEO. He's an incredible engineer and we were discussing the impact of AI on the future. I also receive a newsletter that I finally got around to reading called ChinAI which I recommend, written by Jeffrey Ding. The article talks about, their view on AI and the military, especially in the context of the US and China arms race. What I really got away from the article was that, first of all, I think there's a lot of hype speculation about how AI is really going to change weapon systems, and it's going to be a very immediate effect. However, the point of view is that from an economist's perspective, artificial intelligence is very much similar to the release of electricity as a science and technology as well as a distribution platform.
And so their point of view is that, even though the breakthroughs in deep learning happened in the early 2010s, and of course, have become very popular in the early 2022s, the truth is that this extended trajectory of adoption could really last all the way into the 2050s or the 2060s. This was really important because I think even in the walk with the founder, we had this conversation about whether it was really too late for AI, right? Whether it was too late to build an AI startup and it felt like there's no economic moat and there's no future to it. And I was like, yeah, but you know, I think if AI was like a very quick model, for example, a business model like Uber for deliveries, or Uber for hotels, I think there's a business model that's very fast to spread, but also very fast to be copied. I think when it comes to artificial intelligence, there's a little bit of that awkward dynamic here where there's actually much more research to be done, much more implementation, and much more distribution needs to be done.
(02:43)
The second part talks about was very much about how there's a US dash-China gap and of course, everyone's talking about how as if China's going to crush it. But I think what was interesting was that the article argues that actually China does not have the same in-defense industrial base and is still catching up in terms of the civil-military fusion. And what that means is that it doesn't have startups in defense the same way as Palantir and Anduril. In fact, China didn't even allow non-governmental capital, and international defense industries until 2005. That's a big difference since Silicon Valley was actually very much subsidized, in partnership, and effectively accelerated the US military technologies. In fact, that's why it was all co-located with the military basis, and DARPA, for example, in California, especially with the academic dash, industrial dash, military funding that was really available to kickstart basic R&D as well as honestly distribution, manufacturing, and even the purchase orders needed to note as demand at the end of the tunnel for venture capitalists to do.
In fact, we've seen VCs like Lux Capital succeed because they made that bet to say that one day the US defense industry is going to really require startups that are thinking innovatively. And as a result, even though defense tech is a lot of hardware technically takes a long time to build. And of course, it's very lumpy in terms of capital and therefore very risky for startups historically as a hardware dash governmental buyer startup. This is still very much doable in the US especially when you are in the defense zones like Silicon Valley.
(04:18)
The last thing that inspired me, of course, was that the article talks about how AI technology, for example, is really being built primarily in the US and China. And so they believe that all these advantages are not necessarily going to accumulate, for example, in this, the weapon systems like manufacturing, but will also really accumulate in logistics, in encryption, in communication, in coordination. That got me thinking about the differential AI industry buildout in each country. Right? And so, for example, even though Russia, for example, is normally seen as a very strong world power, especially in arms, unfortunately, it has a very weak AI system because so many researchers have left the country, as well as, them not really investing, in, you know, artificial intelligence as well, compared to China and America. And so I think there's a point of view that is basically implying that Russia is going to lag behind over time as AI percolates into military systems. That's also, I think, really true actually, for not just us, China, and Russia, but also for the rest of the emerging world, right?
So you look at India, you look at Singapore, you look at Indonesia, Malaysia, Vietnam, Philippines. There's not a strong AI fundamental industrial base, right? And so there's going to be an interesting dynamic where the new weapon systems of the future are going to be driven by some sort of AI competencies. And some countries are going to be better positioned to push that through than others. And so I think there's an interesting implication about how Southeast Asia would absorb AI technology into the military.
(05:53)
The other thing that was actually quite interesting was that it was implied that the linkage primarily in the past has been between military and hardware engineers. And I think there's a lot of truth to that. But basically, it's saying that AI is a subset of obviously software engineers, which is a subset of all engineers and that's actually a very different type of talent base that has historically been encouraged, and incentivized to build out. So what this implies to me is that there may be increasing governmental subsidies for not just engineers, not just electrical engineers, not just for software engineers, but really for AI engineers because of that belief that this is going to accelerate the military-dash-industrial complex. And it's not just obviously good for the economy, not just good for academics, but also good for national security. That's going to be a very interesting fusion. And that means there may be multiple attempts by states to really build out those AI hubs over a 20 to 30-year timeframe. So, we might see more subsidies, might see more education, might see more advocacy and championing for folks to learn AI.
And what's most interesting about it is that behind it would be just not just obviously the economy. Therefore, it wouldn't necessarily just be subsidized by the Ministry of Education. You wouldn't just be subsidized by the Ministry of Trade and Commerce, for example, but also subsidized by the Ministry of Security, for example.
(07:31)
Alright, that's my first ever diary episode where I share what I thought of and learned of today. So thank you so much.