Fractal’s Post

Transcript

Are kind of at least the theoretical starting point is of course teams are more innovative. Of course they generate more ideas and more diverse ideas. It's like the question is, are they five times more innovative? Are they 10 times more innovative? Are they 100 times more of it? I think for us the interesting thing was how much better our teams, not are they better. And so for it to come back, especially in the early studies and go they're not better, they're actually worse. Yep, uh, what? You know, And then and then later. Also teams sometimes maybe performed a little bit better than non AI assisted teams. Nobody is like really outperforming, let alone, you know, multiples. It's all kind of middling results. And so I think for us, it was really fascinating because our assumption was AI's amazing. Of course it's going to amplify innovation outcomes. And the truth is it just didn't. And it just didn't for what we know to be relatively predictable cognitive bias and also relatively preventable cognitive bias. So if you think about what what afflicts problem solvers writ large? I assisted or non AI assisted is this kind of cognitive bias that that Abraham and Edith Luchins called the Einstellung effect back in the 1940s nineteen 42. And they basically demonstrated and Karl Duncker later validated it. So did researchers at Oxford even, you know, using eye tracking technology. Even way more recently, umm, some people call it cognitive fixation. Some people call it, you know, it goes by other names. But the point is. We, uh, Herbert Simon called it satisficing. He won a Nobel Prize for this phenomenon in the 50s. But the point is human beings have a tendency to settle for good enough as quickly as possible. And so we're trying to come up when we're trying to generate ideas. There's a reason that's, you know, I think 1 research study found that the average corporate brainstorm yields 2 ideas. That's two as in one single significant digit. You know, I tried to track down the source of that study and I couldn't find it. So perhaps, you know, that's just like footnote that. But the point is, there's a reason that people, you know that the creativity often doesn't happen in the conference room, and it's because there's a deep human longing just to. Answer a question and move on. And when we know that a volume of solutions is actually what yields breakthroughs, that's those things are at odds. Our desire to quickly solve and a willingness to generate a volume are really at odds. And whereas we would one would think that AI would liberate us from the shackles of this human cognitive bias, what we found is oftentimes it only amplified the underlying cognitive bias. And whereas perhaps say you have an hour to solve a problem. Maybe it takes a human, unassisted human team 30 minutes to get to the point that they feel like this is a pretty good idea now and then. And then maybe they refined, maybe they push. There's all sorts of, you know, different paths to take from there. What we found is with AI, all of a sudden teams are getting to pretty good in 5 minutes, but they don't spend the next 55 minutes pushing farther. They, they kind of go, we're going to grab some coffee. You know, we kind of solved it, right? And it just, and that was surprising to us to realize, wow, the tendency to settle for good. Now AI helps us get to good enough faster and we don't leverage it as a tool to amplify a possibility, it just amplifies our underlying bias. And I would also add that for some problems good enough is fine and you want to be able to do this quickly because the problem is not mission critical and AB plus answer an American academic speak is good enough. But in some situations where you have a mission, mission. Critical, uh, problem and you have to get the perfect answer or the right answer. Then good enough is not good enough. And so you have to really push forward to be able to figure out and to maximize for the right answer rather than to satisfy for a good enough answer. And as and as leaders trying to figure out this difference between is it is it is, do you need to maximize or can you satisfy this particular problem is very important because that will that inform. How you might approach, um, AI as a thought partner in problem solving and ideation, Yeah.