How to use AI to think like a superhuman | Jackie Barbieri | TEDxWarrenton

15.24k views1794 WordsCopy TextShare
TEDx Talks
Jackie Barbieri discusses the transformative potential of human-machine teaming (HMT), and outlines ...
Video Transcript:
I had the privilege of spending my career in National Security intelligence and in that job I teamed with some of the most elite minds and Advanced Technologies in the world to solve very hard problems and keep this country safe but in that context where insights needed quickly the unknowns are many and the problems truly complex human machine teams that achieve incremental gains in performance simply aren't good enough and in that world unlike the world of consumer apps not good enough isn't measured in Lost ad revenue or clicks it can be measured in casualties this is
why I'm on a mission to improve human machine teeming outcomes by 1,000 times over the status quo how many of you show of hands used a navigation app to get here today this is an everyday example of how we rely on machine intelligence to help us execute cognitive tasks better than we could on our own algorithms play a key role in our lives whether it's how we manage our personal finances how we see the news and learn about world events or even how we find love and when it works well we experience this 1 +
1 equals 3 effect in fact that's the Hallmark of good human machine teaming when I was an intelligence analyst I was utterly outmatched by the amount of information that I had to consume and analyze every day my first job in the field was to report on changes and Insurgent tactics how they were building bombs that were killing US soldiers abroad I had to go through mounds of information in different languages across different computer systems but man I love that job from the very first day and why because I felt like if I could figure it
out I could really make a difference but every day for that first month I left work with a splitting headache I was pushing through as much information as I could but I still wasn't moving fast enough what I didn't know then but very smart people doing cognitive science research have shown many times since is that information overload actually limits our ability to figure out what all of that information means so the technology I was teaming with to help me access more information faster than ever was actually making it harder for me to do sense making
so how did I get the headaches to stop I started to realize what I needed to pay the most attention to I realized that I needed to focus my attention and slowly but surely I went from being behind the ch and tactics to giving soldiers and decision makers the information they needed to anticipate threats and to be better protected this was the first clue in the formula for superhuman thinking I told you I'm not looking for a 1+ 1+ 3 Improvement I'm looking to get to a THX Improvement Jackie how do you do that by
unblocking and delivering a 10x Improvement on three common bottlenecks faced by human machine teams you're probably thinking 10 plus 10 plus 10 that's not a thousand and that's right that's actually what most people get wrong about human machine teaming they think about making incremental improvements at different steps of a task and just stringing that together and adding it together but Superior human machine teeming is is about learning how to leverage the relative weaknesses and strengths of people and algorithms in ways that compound so if you pick the right bottlenecks to unblock you can turn that
addition into multiplication and you can achieve 10 * 10 * 10 equals a th000 so what I learned in that first job was that if I could design human machine teams that have avoided information overload we could unlock that first 10x once you get on the other side of information overload I found that I encountered the next bottleneck and this is one I am sure all of you have experienced at one point or another the frustration of not being able to answer a question at the speed of thought how many times have you been researching
something for work or for school and you're deep in the text or you're watching a lecture and out of nowhere a question strikes you but you're pretty sure the answer to that question isn't in whatever you're watching or reading right now so you're faced with this decision man do I follow my curiosity or do I stay the course if I follow my curiosity I got a task switch it's going to affect my focus right and it might not pan out but I just might find that deeper insight that I really need that's going to make
a difference if I stay the course at least I know I'm going to make some progress but I'm gonna have to keep track of that question somehow maybe I'll just open a new tab in my browser how many tabs do you have open in the browser on your phone right now okay that's what I call curiosity debt how many of you when you're faced with that decision tend to stay the course how many of you tend to follow your curiosity so intelligence analysts are faced with this Crossroads as cognitive Crossroads every day but so is
every single human being learning something for the first time and the tension between those two choices only gets more acute when there's time pressure when the clock is ticking 10 Xing this or unblocking it would feel like it's opposite perfect flow you go from question to question to answer no friction no curiosity debt this is one of the things that I find so exciting and a little bit magical about AI based chat interfaces like chat GPT in a single conversation you can pivot from question to answer to question to answer to new topic back to
your train of thought no new tabs AI also points to the key to unlocking the third barrier to superhuman thinking this one is a little bit different than maybe anything you guys have heard about before it relates to being able to read and understand machine data as easily as we understand text Advanced sensor technology can see the world in superhuman ways whether it's measuring greenhouse gas emissions from a factory all the way from space or precisely locating a radio frequency signal thousands of miles away on the other side of the world the problem is it
takes significant expertise to read and understand that data and unlock its value and there are very few people that can do it well but even on a small scale it can have an outsized impact on October 2nd of this year in Spruce Pines North Carolina a team of First Responders got a call that a man needed to be medically evacuated by air less than a week earlier hurricane Helen had ripped through the spruce Pines area and the pilots had no idea if they were going to be able to find the man let alone a safe
place to land the helicopter luckily one of the of the First Responders had a friend his name was Kyle and Kyle had a superpower Kyle was an expert in interpreting a highly technical type of data collected by sensors that are called synthetic aperture Radars and he happened to work for a company that builds those sensors puts them on satellites in space and collects data all the time so Kyle got the call he got his hands on some data that had just been collected over North Carolina after the storm hit and he got to work Kyle
reviewed information that to most of us would kind of look like the crackle on the old television screens but in all that noise he found the only building still standing in that entire area just might be where that man was sheltered he also and more importantly found right nearby a safe landing zone for the helicopter so those First Responders they loaded up the helicopter with supplies they took off and they headed straight to that Landing Zone they wasted no time they wasted no fuel when they got there they found that many people were Sheltering at
that building and several not just one needed medical attention all of those people were medac that day and the people that were left behind were left with critical supplies that would help them survive just a few days longer Kyle's ability to read machine data had an outsized impact that day he enabled those First Responders not just to save one but many lives and it begs the question what if instead of Designing our AI teammates to understand us better as chatbots simulate we also designed Tech technology to help us understand how they see the world this
would expand our ability to sense and understand the world around us in superhuman ways these are the types of problems I work on with my team at my company whites space and not just understanding One sensor data type at a time but many it's a very exciting and promising area of research and development and every day we work on these problems every day we get a little closer to that THX but you have to keep in mind that this technology breaking through this third barrier can only be done effectively if you keep the other two
pillars in mind avoid information overload and enable people to answer questions at the speed of thought so every day I go to work and I'm a little bit more excited not just because thinking like a superhuman is going to make us more productive it absolutely will but because I believe it has the potential to make us better humans if we can speed up sense making by a thousand times AI might never threaten to replace human judgment in some of the most critical decisions that our leaders face in fact it'll make more time for them to
exercise that judgment and still act in time to matter we will be living in a world where preparing the information for key decisions and tradeoffs whether on the battlefield or in the boardroom happens as quickly and as easily as finding a route to a new destination we will be living in a world where exercising superhuman judgment becomes as commonplace as using an app and I don't know about you but that's the kind of world I want to live in and it's coming are you ready thank you thank you [Music]
Copyright © 2024. Made with ♥ in London by YTScribe.com