Transcripts For CSPAN2 Lorien Pratt Link 20240713 : vimarsan

CSPAN2 Lorien Pratt Link July 13, 2024

I think i like to think of it as efficient a pond. Maybe there are 200 fish the side of traditional human societies, everything is going okay for a few years. Then a storm happens in the river opens up connecting that fishpond to the ocean. New fish swim upstream, the environment is fundamentally different. We are in that situation today as a society, things are changing radically. Actions have outcome at a distance and that is what link is about, the actions that we take and the decisions we make as we consider the actions in the past through which those actions become reality. How can we be responsible for the outcomes of our actions if they are invisible, if we can see the outcome of our actions, a fish in a larger ocean, we simply didnt eat all the for this situation. I am lucky i went through the computer revolution. When i was a kid computers were big old things no one could use, my mom didnt know the difference between software and hardware. We have gone through a massive democratization of computing technology. We are in the same cusp today with ai and data. Ai, we dont have control over it. Data is overwhelming, at best we get a data visualization. I had the honor of interviewing hundreds of people as a Technology Analyst and asking them what you are frustrated about. Of technology could solve one problem for you what would it be . Over and over i heard a similar answer that looks like this. This is why i am pretty sure i know what i am doing. I have been building Machine Learning, applied systems a really long time over 30 years. I was funded by the human genome project in graduate school, built 100 million budget for the government, i have built thousands of Machine Learning models mostly supervised learning, back propagation, who would have known it would be with us so many years later. So i believe this background has given me an insight, something has been missing all these years. Something coming up from the technology instead of putting humans at the center of the occasion done. And i like that he called it intelligence augmentation. Ai, flip it upside down to the ia, putting humans at the center of the equation again. When i interviewed all these people i found what i called for a while the decision archetype. What is the decision . In action, a thought process that leads to an action. Action in a complex world flows through some stuff and i dont know what buying that scarf will do to the world. It is going to have some impact but i dont feel very motivated because i cant steal the the impact, it doesnt grab my primate brain in a way that makes me think i really need to buy that hybrid car. I cant see it and the data stat today isnt giving that to me. This is my dog. I am training him to be a service dog. I have had him his whole life. Is 11 months old and i had an awesome thing happened to me. A trainer teaching me to train the service dog teaching me about antecedent behavior consequence. My head exploded. That is what i heard from humans i have been interviewing, executives i have been talking to. Always talking and antecedent which is the context. In the kitchen i say sit, the behavior, dog sits, the consequence, he gets a cookie. This is a universal archetype, not just one way to think about how we might use ai and data. I will tell you how that fits in in a moment. Im certain this is the way to think about it because it has the lowest friction to how humans think. Lowest friction to how humans naturally think, and busy people live in complex environments. They dont have brainpower for optimization or inference methodology or any of those things. We have to meet them where they are and the fact that we are not has created a cultural barrier between people at the head of government, at the head of businesses and even be as they try to make decisions and use evidence and data and ai to help make sure those decisions have a Ripple Effect that is good so farmers im working with have to decide what crop to plant, down the road they dont know if it makes them productive or what happens when they have fewer migrant workers, the situation has changed. Businesses might decide where to acquire a company or what product a larger what price to change and as they think that through, much like my dog here there is an antecedent, theres a behavior, we will watch this product at this price in down the road somewhere, my dog is immediate, he is a dog. What makes it special is we can think through long chains of consequences but that is limited and we need computer help so again a decision is an imaginative process in our head as we think through the actions in some context that will lead to some results. If you remember nothing else remember this template. What is cool about decision intelligence which is what the book is about it is because we start with humans, you can take home and use immediately. That is my promise if you stick with the talk. How do we make decisions today . I am sorry to say i had a sense this was happening especially in a complex world but going back through millennia of Human Evolution we dont really think through the consequences of our decisions very deeply. We are more likely not to think through things rationally and instead use social signaling, look for someone who looks successful in our society, dominant or prestigious and copy the decisionmaking. It turns out that is very effective. Hugely successful for the human race, that separates us from many other species. We are great copiers and cultural evolution theories as we develop behaviors and patterns any individual cant understand but society like the unconscious process of genetic evolution we use cultural evolution to come up with these behaviors. We are programmed to look at some prestigious or dominant person and do what they do as opposed to thinking through the consequences. That worked great for a few millennia but the situation has changed. If there is a bad actor here or here and they tell us what to do and they are smart they can subvert our behaviors, influence us to make decisions that benefit them but not us if they are smart about the situation. Second, the context is rapidly changing. We need to develop new ways of coping with the big ocean that is fundamentally different from our pond because it keeps changing. There are new fish in the old ways of thinking through problems are no longer working. These have complex system dynamics, feedback affects, winner take all patterns where large companies, 90 of the benefits and massive inequality. Action at a distance we talked about, intangibles are important. Anyone who has worked with data, we tend to focus on the measure easily. Money, size, price. We tend to overlook reputation, happiness, more row. I never build a decision model that didnt have at least one feedback loop that involves something intangible. We must Start Talking to the sociologists, cultural evolutionists and all the other disciplines to understand those factors. The decision intelligence creates a roadmap for how to do that. The other thing i didnt phase the future is no longer like the past. When we assume the past and the future are the same, we didnt realize the situation has changed, office was used to be white and now they are black. I believe ai which im going to tell you about, decision intelligence can solve this problem. I grew up in appear go of technology optimism. We are sharing our code, the internet was going to democratize, we were going to collaborate. We had a dream and i dont think we have realized that dream. I think decision intelligence will help us go there. I think we created a number of links in the chain, learning collaboration, social media and one more link we need to start to make a big difference, to have a nonlinear impact. We will start to talk a little practically right now. How do we do di . We start with people. We dont say where is the data . We dont say we cant do this ai like thing without data. Data is great but there is a huge amount of Human Knowledge that has no data set whatsoever. We are good about knowing how actions lead to outcomes in your homework is to go home, ask a friend how they think about a complex decision. They will talk about actions or intermediate effect and that will lead to some outcomes and talk about context. I sit down with a diverse group, old, younger, gender, what outcomes are you trying to achieve . So Many Companies have big projects. And many organizations, what are you trying to achieve . The list of outcomes is different for each person. You dont need technology to get better. Just a brainstorming process where you think through what outcomes to achieve as a team, higher revenue, net revenue after two years, some kind of military advantage, a back lash its us with the reputation of our country. What outcomes are we trying to achieve . Make sure you ask that question. Second, brainstorm through the action. Many folks dont take the time to have an open brainstorming session where they are allowed bad ideas and funny ideas, actions that might achieve those outcomes, to the creative side of your brain, when the blood is in the creative side or analytical side of your brain which is over here maybe you dont have room for the creative side so separate those two. Spend time being analytical. These triangles are where ai fits in. Most models as we democratize ai, here is the pattern. A decision i am facing today, i saw greta on tv and she was so compelling. She said we have a Climate Crisis in the way we solve this is really simple. Stop worrying about analysis. At the very least pay for some trees, organizations all over the world will take your money and buy trees and those trees will grow so there will be more biomass and that will sequester some carbon and if enough people do this perhaps on its own, dont know she said trees alone would do it but it would make a big difference. I havent sent any money to a Tree Organization yet because i cant visualize how the money i might send leads through a chain of events to some outcome. If im going to use ai to benefit me i want a visceral interactive fun experience. This is what i think is the future of ai, looks like a video game and i hope we can do this in the basement. We can walk through these spaces. What are we doing in these spaces . We are experiencing with actions we might take and letting the computer help us understand the chain of eventss that sets in motion to lead to the consequences. Valuable at a personal level and highly valuable in an organizational level. Lets see if this works. I have been having a lot of fun, i really like it. Theres institute for the future ball, this shows you a physics Engine Running here. What is going on here . We are trying to make a decision how much money under the pay each year to sequester carbon. As i change this decision this data is telling me about the future that puts in motion. I spend money, change the decision, changes the number of trees i will purchase, heres the biomass, the carbon sequester, the total atomistic debited atmospheric carbon. I can see the chain of events. Linda ritchie is the expert in the background who has done some research and barry hayes has done some research. Except first has an opportunity to say how actions connect to outcomes. I can change my decisions and also who i can trust because i can see different people, different experts claim different things. I can click on their names and go to a site where i see them making their case, their model for how it leads to outcomes that can help us and that will be like wikipedia. A site that is cure related to understand the situation. It looks like a Business Intelligence dashboard, looks like stock we have been building for a long time. It is not. We are not looking at the data set here. We are looking at the future. Let me summarize so you can understand it. In the background we also have a physics engine generating implications of those choices. As i changed my investment, i can see how my decisions interact with the situation as they characterize it to impact the outcomes i care about. This is a universal pattern. This example is an example of something you do in your head 500 times a day and Large Organizations struggle with understanding the impact of todays decision tomorrow so i will give you a couple examples. These are Machine Learning models in these triangles. We might have built a machine, whether a Computer System has a current intrusion happening now. That is a common Machine Learning model. Machine learning is widespread with intrusion detection and gives us a score, 20 , 95 , pretty sure there is an intrusion right now. Here is the type of intrusion. This picture is typical of many decision model, it is mapping how people naturally think. I promise you if that is how you are thinking about things it is a lot better to have it on paper than to keep it in your mind and explain it when using invisible mechanisms, word and text which is inherently linear to communicate these decisions. We need a blueprint like this, a learning system that shows how Machine Learning fits in. I could send information to the police, investigate the intrusion, and impact some outcomes. If i call the police every time it might be pretty costly. Police only when necessary. This is a decision model i built with a bunch of farming experts as part of that big project i talked about, 23 phds and finishing up the proposal this week, a National Center for excellence in ai and agriculture. What is important about this, i didnt have to explain decision intelligence to anybody. I sat down with my diverse team of experts and said what is a typical farmer trying to achieve . They want to be profitable at the end of the year. Is that all . They wouldnt want him to take any action that will put him out of business in a few years so that is a second goal they need to balance against that. Making choices how they spray their top, spraying schedules, the choice of crop, this gives us a map to understand how the Machine Learning Technology Fits together. There is a model tier to tell us how it will impact the yield, another model, the contaminants based on the sprain and these are sensors that farmers might have on the field that they can use, we went as Early Warning as possible so that we can spray as little as possible to achieve our goal. It cost a lot in longterm viability and the climate and pollution point of view. It is kind of spaghetti but having it on paper is better than what is going on right now which is this invisible in peoples heads. It becomes an artifact and we talk about design thinking, we designed a decision, something that we can design. That is pretty radical but we could bring all those things, best practices to decisions. We can you a this thing and continuously improve it. It acts as a blueprint, this is their mental model but it connects them to the ai people so they know where they fit in. Before we did this diagram i dont think anybody knew how it fit together, you dont have to explain in words anymore. We have a map. We know how it fits together, where the ai will go in. What is going on with decision intelligence . It has been quite the slog. I am trying to talk, sometimes three people in the audience trying to tell people the story, to the extent this matters on its ai hype cycle. That i consider a big accomplishment. Ibm, google trained 20,000 people, she is awesome, google her and a bunch of companies started to identify, most in specific areas. These guys are hr curriculum labs, medical devices, these are all people who recognize that if they go beyond building a Machine Learning model and put it in a decision model surrounding the Machine Learning model, connect actions to outcomes, that will help them be more successful. It is really cool. I love that we are starting to get this. This is how kathy defines it, a discipline of turning information into better action at any scale. We have a different way of saying it but it is essentially the same thing. Di answers the question if i make this decision today which leads to this action, what will be the outcome . Lots, my causal decision diagram is mine, other people have other approaches. Some people dont even come from tech, sociologists, economists, and more. What is common to all of us as we have taken seriously the action to outcome as the wave of technology and science interact. We made it to hollywood. Who knows what show this is . It is the good place. I am not going to do the spoiler. This is episode 10. I recommend you take a look at it, he basically discovers di by the end of the series, the big reveal which is so cool. Who knows who this is . That is janet. Is she human . He is trying to understand these unintended consequences and janet was helping him. How cool is that . I dont know if they read my book. I hope they call me. It is so important to the future. With great power comes great responsibility. You have the ability to take actions as an individual and organization that have a giant impact. Im an optimist. I believe we are at the beginning of a solution renaissance where it is not just ai but all of these technologies will come together under a common blueprint, system dynamics, computational neuroscience, cultural revolution. These are viewed as soft fields in their little silos, deconstruction of 1000 years of being specialists. We are entering the age of synthesis. Those of you who are neogeneralists now what im talking about the we are the expert to focus on the outside of these boxes, doesnt matter if we know about Operation Research and we start to now using some version of di how to glue that to get her to crystallize solutions, we understand a solution to water impacts poverty. A solution to poverty impact the status of women. Those impacts hit all of us, they had governments, democracy, companies. We call these the Sustainable Development goals. I think they are all the same problem. I dont think we can solve them separately. I think the only way we can solve them is to have a new approach to understanding bouncing around to the whole world and ultimately lead to the best outcome. [applause] a share transporter. I am shocked that you finished under the time. One question i have is the case study that shows how this works in practice. The example i showed with farming is a good one. We also built an initial model for the government decision regarding confli

© 2025 Vimarsana