It’s the 250th anniversary of America, and there’s a lot of attention paid to history in Boston, where we were at the epicenter of the American Revolution in 1776. But there’s also a lot going on in terms of looking at the future, and the next 250 years: what will we see, in America and around the world, with AI?
That’s part of the buzz around the America Innovates conference in San Francisco, which happened May 16-18, with the involvement of Forbes. I went and saw a lot of what happened there, including the exhibit at Fort Mason’s Gateway Pavilion, and how a lot of people attended just to see what’s on the horizon in these interesting times.
I also facilitated a panel on American innovation, with my friend and colleague Alvin Graylin, who has his own transnational experience, Cinnamon Sipper, co-founder of Godela, and Daniel McGill from Joby Aviation.
Notes on Advances
I first asked Graylin to talk about the respective strengths and weaknesses of approaches to AI in both America and China.
“The key thing right now for America is that it has been the source of innovation for technology space for the last several decades, and it’s doing that right, but what it’s doing wrong now is that it is doing everything that it can to destroy the source and engine of that, with essentially decimating the education system, and the immigration system that brings the best minds into this country, as well as investing in research,” he said. “What China is doing very well is that it is focused on diffusion, taking technology to anywhere in the world and making it accessible, and affordable, both internally as well as to the rest of the world. What it’s probably not doing well is maybe creating enough of a positive relationship with America to try to do more joint development and sharing of technology.”
I asked him how we preserve human agency and meaning in this era.
“We need to be able to put some constraints on ourselves in terms of how we use it, when we use it, and how much we rely on it,” Graylin said, using Greek mythology as metaphor. “And one very simple thing you could do, and also for your children, is: when you’re using AI, don’t go directly to AI for the answer. First, come up with your own answer, take that answer, feed it into AI and say, this is a question. This is what I answered. What am I missing? What could I do better? And when you use it that way, then AI actually enhances what you’re doing, and you’ll remember, you’ll learn more, versus: if you go directly to AI for the first answer, you’ll then delegate your agency.”
The Friendly Skies
I asked McGill about his contributions to VTOL or vertical takeoff and landing systems, and what he thinks these systems will do in the future.
“We really think it’s going to revolutionize the American society, if you will,” he said, referring back to the interstate highway system in Eisenhower’s time, and how that changed transportation. “You think about the housing crunch we’re in now, and the affordability crisis. This will certainly allow us to expand the parameters in the geographic footprint of society in general.”
I asked about the flying cars that we were all promised on the Jetsons, decades ago when we were kids, and from McGill’s response, I could tell that we’re finally pretty close.
“A call to action, a drum I personally continue to beat,” he said, “is looking to both the public and private sectors to invest in this vertical infrastructure. It’s going to take federal input, it’s going to have to be a real true public-private partnership to stand up the infrastructure required to support this new mobility mode.”
AI Learns Physics
I asked Sipper about the importance of training models on real-world physics.
“Today’s AI models that we commonly use are trained on text and language, and AI trained on text data can’t build the physical world,” she explained. “So, the problem that we’re focused on solving is: how do you accurately predict things like how heat transfers, how fluid flows, how structures break? And the reason that’s important is that’s what codifies physics, right? Some of the most challenging problems that humanity has to face, protein folding, turbulence modeling, and energy, will require AI models to understand and accurately model physical behaviors, and so we believe it’s an important piece to start from, in order to unlock a lot of the most challenging problems that we’re facing in society.”
I asked Sipper about her predictions for 2050. She had a different timeline, addressed by a slogan used at Godela: 12 years of engineering in 12 minutes.
“It means that for everyday people, what you experience is cheaper energy, faster cures, safer products, just being able to push progress much faster and much more efficiently, is where this lands,” Sipper said.
Exploding Value
My next question was to the whole panel, and it’s something I’ve been wondering about, in general.
People talk about the “AI bubble,” suggesting that a lot of this investment money is going to just go up in smoke, in a sort of global rug pull that leaves our world economy cratered. I don’t know, myself. I think there may be enormous applications that are going to, if you will, “tier out” of the AI sector currently experiencing high levels of investment. In other words, maybe the investors are NOT totally dumb.
Anyway, the panelists responded.
“I actually think that the focus on a trillion dollars is actually maybe misguided,” Graylin said. “Everybody wants to make the next trillion dollar company, but when we do that, we start to take focus off of what really matters, which is: what value are you bringing to society?
Here’s more of his economic perspective:
“Technology is going to be a deflationary force that actually makes things cheaper, so which actually may bring margin down over time, so you may actually create more value by creating a shrinking nominal economic market,” he added.
Sipper talked about possible outcomes from simulation and modeling.
“I think right now, a lot of engineering process is bottlenecked by the tools and developed around the tools that existed up until this point,” she said. “Traditional numerical simulators, I think, having AI with physics constraints baked into it, such that you can design in real time, continuously model, test, and understand how a system would perform if perturbed in one way or another, without actually even going to physically build the object, that is a reality that we believe will (drive) those great innovations.”
Graylin went back to some of the potential issues with the current rate of change.
“The country is not prepared for the world that we’re creating, and if we don’t get our policy makers to create the social safety net that will give us the economic and social soft landing, we’re going to create instability that is both national and international,” he said.
Sipper talked about personal strategy:
“Given how much technology has changed in the last year, I feel like personally, the trend is to place a lot of my time, my free time, in spaces that I don’t believe will be quickly disrupted by technology,” she said. “That means spending more time playing piano, and reading, and indulging in ideas. I think there is synergy to what we’re doing, because if you think about what it means when tools like Godela scale, it means enabling democratized access to highly technical tools.”
McGill, for his part, was enthusiastic about federal partnership and FAA involvement, and suggested that right now, a big part of the job is getting the message out to people.
“We’re trying to be very logical and strategic, partnering with some of the AV companies, and thinking about these multimodal hub opportunities,” he said. “You think about autonomous vehicles on the ground, and, like I said, our aircraft will be piloted, but having interconnection points in cities, which could connect you to resort destinations or other cities, airports, et cetera, and just being very thoughtful about how we kind of weave these multimodal modes together.”
All of this contributes to thinking about how we will spend the next 5 years, the next 20 years, and yes, the next 250 years. What would the founding fathers think about AI? How does traditional thinking impact our moves forward with these technologies? Think about it – and drop me a comment.
