Generative AI in the Real World

What fun! I spoke at the “AI in the Wild: Beyond the Cloud” event at the Stanford Research Institute (SRI) last night (which recently acquired PARC) on behalf of the Post-Industrial Forum. Now that’s a sentence – rife with the history of computing – that I never thought I would write.

Basically, it was a discussion of what will happen when we build large-scale generative AI-based systems in the physical world. Besides myself, speakers included Lawrence Lee from SRI, Dr. Ivan Poupyrev from Archetype AI, Bruce Pittman from SpaceBilt, and Daniel Isaacs from the Digital Twin Consortium. After the presentations, Lawrence ably led a panel discussion of the same group.

As you know, most of the hype (and money) of the last year has been related to the infrastructure layer (GPUs, datacenters, and their massive power consumption) and the model layer (ChatGPT4, Llama, etc.).

But the presentation I gave focused on the application layer – applications that must utilize generative AI technologies to transform how people accomplish goals in the real world.

Pratexo’s efforts in this space relate to Project DeepCause, our work to perfect an expert system ‘builder’ that can augment the capabilities of an operational, process, or domain expert with a generative AI-based assistant.

Our ultimate goal is to reduce by over 90% the time and technical complexity involved in creating a fault-cause ‘graph’ for a complex machine or process that then forms the basis of intelligent systems (applications) running on the industrial edge. In this case, it is not just about using generative AI in an application, it is about using generative AI to help create the application in the first place!

Blaine presenting Pratexo Project DeepCause

A few other notable accomplishments that jumped out at me during the event:

  • The incredible work on bringing the massive compute power required by AI-enabled applications to space (via R. Bruce Pittman at SpaceBilt)
  • Super-cool work on the combination of AI with sensor technology to enable applications that make sense of the physical world (via Ivan Poupyrev, Dr. at Archetype AI)
  • And how most companies are finally understanding the value and concept of digital twins (via Dan Isaacs and the Digital Twin Consortium) and how these twins are now becoming increasingly intelligent via AI
Panel discussion led by Lawrence Lee of SRI

Thanks to Frode Odegard and the Post-Industrial Institute for putting this cool event on and inviting me to speak. It’s going to take a LOT of brainpower over the next few years (and some AI power as well) to figure out how to best harness these new technologies. I’m confident we can get there!