Bring physical AI from demo to deployment.

Edited for readability from the recording’s automatic captions.

Artificial intelligence is an umbrella, right? We’ve been flooded in the market and marketing, as to your point, generative AI, as we’ve seen with the front-end models and the chatbots, your ChatGPT, Claude, etcetera, etcetera.

However, those systems run on things, right? That’s just the application layer that you see on your phone, right? Layer seven, for all my folks out there. That’s not AI. AI is a data center, right? This was why I had to build one, to show it. So that means, like we talked about earlier, energy, silicon, infrastructure, etcetera, etcetera. However, these models were trained on text, some photos, some videos, but not a lot of interaction with the real world.

So the components that I’m bringing about is physical AI, right? Interaction with the physical world, like the robots you see downtown. And how you have to adapt to those little limitations and bottlenecks. And then there’s edge AI as well, which is basically running the AI closer to where the data is being run.

Salute to all the hyperscalers, especially our partners Microsoft and AWS, love you. But I feel like the cloud should be used to train the models, right? Not having your whole infrastructure there. The cloud has a place. You just have to be smart about it.

So if we keep being inundated with the chatbots, right, and ignore all the robots and manufacturing that’s going on in, say, China or South Korea, or even in this country, of all these conferences that I go to that, let’s face it, have a lack of diversity, right? You start to see the picture and how we need to start learning now, right?

I keep telling the professors all the time, when the house is on fire, you don’t have time to measure the buckets, right? And I’ll leave it at this, right? I talk about systems a lot. Systems thinking, engineering over tools. Tools change every six months. Systems don’t, right? And if you just learn how to prompt, you’re just renting relevance from whoever built the tool, right?

The durable jobs and the generational wealth in this wave are the infrastructure layer that we talked about. And I want our students positioned where value accrues, not just where demos are, right?