Yesterday, I wrote a letter to a university leadership team describing how powerful an opportunity The Intelligence Tsunami (https://www.amazon.com/dp/B0CXH5DDRP) is for the university. Autonomous intelligence will impact every organization and individual in every discipline, providing a distinctive opportunity for cross-collaboration of faculty and students on campus and for engagement with industry.
I noted that I have the privilege today of being a judge in a senior design bioengineering symposium. I said, “Imagine, in a year or two, these student projects having the capabilities to operate autonomously to monitor an individual's health or to conduct surgery.” And then, light-heartedly, I came up with an illustration on the fly. “I reach for the Ben and Jerry's in the freezer, and the intelligent agent on my watch gently nudges me that my new grandson is coming in August, and I want to see him grow up.”
My human brain easily and efforlessly congered up that insight, but is very sophisticated insight for a machine to produce.
Writing the letter, I imagined the university leaders reading it, getting to this illustration, and chuckling. All of these folks understand personally leaving the ice cream in the freezer. I also communicated to the leadership that I am engaged on campus. Yann LeCun calls this a worldview. My worldview allowed me to visualize the letter’s readers, empathize with them, plan an action, anticipate their response, and decide and act on my expectations. There is a lot packed in that to code into an artificial intelligent agent.
That insight required connecting disparate data from my unique perspective. My watch needed to be able to perceive that I was reaching into the freezer for the Ben and Jerry’s, and to quickly understand what Ben and Jerry’s is. It didn’t need to match an image of every possible Ben and Jerry’s carton, but only abstract enough information to understand it is high-octane ice cream. Our brains don’t process all of the sensory data they receive. That would be overwhelming and debilitating. Our brains have a powerful capability of extracting the most relevant bits of that data and matching it to patterns so quickly we understand the world. Intelligent agents need to be able to abstract the most relevant data from images like that as well.
The intelligent agent in my watch needs to know I am pre-diabetic, so I need to lay off the ice cream. I already know that, but Chunky Monkey is so good. I may give the intelligent agent access to electronic medical records from my last annual physical exam and glucose readings my watch takes from the interstitial fluid in my skin.
The agent needs to know that I have a new grandson coming in August who I’d like to see grow into a young adult in the next twenty years. My agent could have nagged me about losing weight, which would have been irritating and probably wouldn’t have worked. My agent had a worldview of my situation, considered the nudges it could provide me, and framed this nudge in terms of wanting to see my new grandson grow up, which is a powerful, intrinsic motivator for me. My agent’s ability to connect two disparate facts, I am pre-diabetic, and I have a new grandson coming in August, is powerful.
This illustration that popped into my head while writing this letter wasn’t arbitrary. Several years ago, I explored opportunities to develop and market a smartphone app that allowed individuals to live a healthier lifestyle. I met with an HR director to explore an app that would remind employees in his organization about health factors like their weight. He got up and walked to his conference room’s window, which looked out over his campus. “See those folks walking around? Many of them are overweight. They know that. Telling them isn’t going to change their behavior.” On contact with the customer, that new product idea crashed and burned.
At about the same time, I also connected with another entrepreneur interested in developing a healthy lifestyle application. He took a different approach. He started with a survey to identify the individual's intrinsic motivations and crafted nudges inspired by those motivations. I also have three other grandsons, who are 9, 7, and 3. I took the entrepreneur’s survey and felt it was a tedious exercise as I worked through it. But then it produced an amazing result. The entrepreneur gave me messages that the app would provide me about living a healthy lifestyle, and most of them were related to watching my grandsons grow up. Man, was that ever spot on. I was impressed.
That experience was stored in my memory. I had no reason to believe it would ever be useful again. Yesterday that memory popped out as I wrote the letter. Our brains' ability to do that is fantastic.
Large language models have a similar capability. In another post (https://www.innoventure.com/post/the-most-amazing-thing-happened-you-should-know-about), I describe how I checked the accuracy of a chapter of my book with ChatGPT, and it connected an esoteric fact about an event forty years ago with a prompt I entered in 2024 to correct an error in the book. That was impressive as well.
Autonomously intelligence agetns are not yet here, but they are right around the corner. For all their amazing capabilities, current large language models don’t truly think; they pattern match at such an exceptional level that they fool us into believing they are fluent. My book describes that what’s coming are intelligent agents that have a worldview to be able to perceive, plan, decide, execute, and learn.
That's going to rock your world. You can develop and deploy autonomously intelligent agents to supercharge your productivity and to create enormous new value your customers. Get off the beach, ride the wave, or get washed away. The best way to predict the future is to create it.
Let's talk.
Let's inspire your team and your organization to excel.
John Warner
864-561-6609
