In this episode of Innovators Inside Podcast, Vivienne Ming shares a sharp and practical perspective on what AI means for founders, operators, and innovation leaders right now. Rather than framing AI as a simple productivity tool, she argues that the real opportunity is much bigger: using AI to help people think better, explore uncertainty, and solve harder problems.

For leaders trying to build resilient companies in a fast-changing environment, this conversation offers a useful shift in mindset.

Here are the five key takeaways from their conversation:

1. The future belongs to people who can solve ill-posed problems

One of the most important ideas in the episode is Vivienne’s distinction between well-posed problems and ill-posed problems.

Well-posed problems are problems where the answer is already known, or where there is broad agreement on what the answer should be. These are the kinds of tasks AI is increasingly good at handling. If the job is routine, repeatable, and based on patterns that already exist, AI will keep getting better at it.

Ill-posed problems are different. These are the problems where the question itself is still unclear. They involve ambiguity, uncertainty, judgment, and exploration. According to Vivienne, that is where human value becomes even more important.

For founders and innovation leaders, this matters because real innovation rarely comes from repeating known answers. It comes from working through uncertainty and asking better questions before anyone else knows what the answer should be.

2. Most companies are using AI the wrong way

Vivienne challenges one of the most common narratives in the market today: that AI should mainly be used to make work easier and faster.

Her argument is that this is often the wrong goal.

If teams use AI only for efficiency, they may end up reducing the very capabilities they need most. In the episode, she calls this the efficiency lie. When AI handles more routine thinking, many organizations do not automatically create more space for creativity. Instead, they often just generate more routine work.

This is a critical point for operators and executives. AI strategy should not be limited to cost savings or output speed. Leaders should be asking a more important question: Is this use of AI making our people better thinkers, better decision-makers, and better problem-solvers?

The companies that win will not just automate work. They will build systems where AI strengthens human judgment.

3. Great leaders reward productive failure

One of the clearest leadership takeaways from the episode is that innovative organizations need to stop rewarding only safe answers.

Vivienne argues that in any truly intelligent company, people should be wrong a meaningful amount of the time, as long as they are being productively wrong. In other words, they should be exploring ideas that may fail, but that could lead to meaningful breakthroughs if they succeed.

That is a powerful challenge for leaders.

Many organizations say they want innovation, but their incentives reward predictability, caution, and polished certainty. When that happens, people stop exploring. They stay close to what feels safe. Over time, the company becomes efficient at repeating what it already knows instead of discovering what comes next.

For founders and managers, the lesson is clear: if you want innovation, you need a culture that supports experimentation, visible learning, and thoughtful risk-taking. Innovation does not happen when everyone is trying to avoid being wrong.

4.  The best use of AI is augmentation, not replacement

Another major theme in the episode is that the future is not human versus AI. It is human plus AI.

Vivienne makes the case that the most powerful model is not full automation, but collaboration. AI can help with pattern recognition, synthesis, and well-posed tasks. Humans bring context, uncertainty awareness, creativity, and the ability to navigate situations where the rules are still unclear.

That combination matters.

For innovation teams, the goal should be to design workflows where AI handles parts of the problem that are structured, while humans focus on interpretation, insight, and strategic direction. This is what makes augmentation different from replacement.

For founders and operators, this means AI should not just be deployed as a labor substitute. It should be built into the way teams learn, test, decide, and improve. The smartest organizations will use AI to increase the quality of human thinking, not to remove humans from the loop too early.

5. The real competitive edge is learning how to learn

The episode also pushes beyond business strategy into education, talent, and long-term capability building.

Vivienne argues that the most important skill in an AI-shaped future is not memorization. It is learning how to learn. That includes resilience, self-assessment, communication, curiosity, and the willingness to explore uncertainty instead of rushing toward easy answers.

This applies to both kids and adults, but it is just as relevant inside companies.

Teams that know how to learn will adapt faster than teams that only know how to execute yesterday’s playbook. Leaders who can create environments where people deepen judgment, expand understanding, and build better questions will be much better positioned for the future than leaders who focus only on short-term efficiency.

This may be the biggest takeaway of all: in a world where AI can generate answers quickly, the advantage shifts to the people who know how to think more deeply.

Final Thoughts

This episode of the AlchemistX Innovators Inside Podcast is a strong reminder that AI is not just a technology story. It is a leadership story, a learning story, and an innovation story.

Vivienne Ming’s core message is simple but challenging: the goal is not to let AI do more thinking for us. The goal is to use AI in ways that push us to become better thinkers ourselves.

For founders, operators, and innovation leaders, that is the real opportunity. The future will reward organizations that can explore uncertainty, build cultures that support productive failure, and combine AI with stronger human judgment.

If you are thinking seriously about AI strategy, innovation leadership, the future of work, and building better teams, this episode is worth your attention right now.

Have a question for a future guest? Email us at innovators@alchemistaccelerator.com to get in touch! 

Timestamps

00:00 🧠 Intro to Vivienne Ming

01:55 ⚡ Rapid fire and the nature of real innovation

05:37 🔍 Why facts are not enough anymore

08:16 🤖 AI, well-posed problems, and ill-posed problems

15:58 🚀 Why the future belongs to people who explore the unknown

20:26 🏢 How leaders build cultures that reward productive failure

21:51 🔥 The efficiency lie in AI

27:00 👥 AI augmentation vs automation

29:52 👶 How to prepare kids for an AI-shaped future

33:27 🗺️ Designing tools that make people better

35:56 📚 Why AI tutors should not give answers

37:14 🛠️ What leaders should do differently right now

43:36 🌍 Vivienne’s realistic view of the future of AI

47:55 📊 Using AI to uncover human behavior and hidden patterns

57:40 🧭 How people really make decisions

01:01:28 🧪 Hybrid intelligence and human + AI teams

01:07:10 💡 Why human + AI teams outperform

01:10:20 ✅ The real choice: stay shallow or go deep

Full Transcript

00:00:50:28 - 00:01:27:00
Layne Fawns
Doctor Vivian Ming is a theoretical neuroscientist, inventor, author, and serial entrepreneur. Often described as a professional mad scientist. She's founded and led five companies, including Soco Labs, an independent research institute exploring the future of human potential. Her work has been featured in the Financial Times, The Atlantic, The Guardian, quartz, and The New York Times. Vivian has developed AI systems to predict manic episodes in bipolar patients, and even reunite orphaned refugees with extended family.

00:01:27:02 - 00:01:52:24
Layne Fawns
She has served as a visiting scholar at UC Berkeley's Redwood Center for Theoretical Neuroscience, and today contributes as chief scientist at Possibility Sciences, chair of the Advisory Board at Neuro Tech Collider Lab, and visiting professor at UCL's Global Business School for health. Vivian, thank you so much for joining us here today on Innovators Inside. It's going to be such an exciting conversation.

00:01:52:26 - 00:01:54:28
Vivienne Ming
I'm looking forward to all of it.

00:01:55:00 - 00:02:10:09
Ian Bergman
Oh my gosh, I am too. This is this is going to be super fun. So we're going to we're going to get started with some warm up. We love playing games with our guests. And I think, Lane, why don't you take us away on our first little game?

00:02:10:15 - 00:02:33:01
Layne Fawns
Okay. Perfect. So this is what we like to call, our rapid fire section. So, I'll probably pick one or either of you, and I'll ask a question and try to just answer with the first thing that pops into your head. Usually these lead to some discussions or some debates. We've had really intense discussions about microwaves in the past, but, yeah.

00:02:33:02 - 00:02:41:02
Layne Fawns
So first question for Vivian. If you could have dinner with any innovator, past or present, who would they be and why?

00:02:41:04 - 00:03:03:13
Vivienne Ming
My fast answer is that's not the way my brain works. I'm never looking for the person to have dinner with. I don't have, you know, apart from maybe dinner with Aaron Burr to just figure out what the hell he was doing in Mexico way back when, that sort of thing. It's the ideas that get me excited and drive me forward.

00:03:03:16 - 00:03:20:06
Vivienne Ming
So whoever has the most transformative ideas and unfortunately, most of the famous innovators aren't really the innovators. They built the systems that allowed it to happen. So there's my really honest answer. I'd had dinner by myself, and I'd read science papers.

00:03:20:08 - 00:03:21:25
Layne Fawns
I love that. Oh, it's.

00:03:21:25 - 00:03:34:28
Ian Bergman
So good now. Now, maybe you could, you could sit quietly reading your paper across the table from Da Vinci, reading the same set of papers and coming up with crazy, you know, crazy batshit ideas that may or may not ever work.

00:03:35:01 - 00:04:05:05
Vivienne Ming
Yeah. You know, it's obviously he threw a lot of things on to, parchment and, you know, not every corkscrew, helicopter was actually going to fly. But he was there in the early days. You know, I look at people that were multi time successful. Clearly, Einstein, as many really hardcore physicist might point out, other people were thinking similar thoughts at the time.

00:04:05:07 - 00:04:33:19
Vivienne Ming
Maybe he wasn't the greatest mathematician of his age or anything along those lines, but three times. Photoelectric effect, for which he actually won the Nobel Prize. And, you know, relativity and general relativity, like when you see people being able to come up with something truly original multiple times. And that first one, the photoelectric effect, for which he is not famous, but justifiably should be, that is innovation.

00:04:33:21 - 00:04:54:26
Vivienne Ming
The Gilded Age thought they'd figured it out. Every. There's this weird thing about electrons. Everything else we figured out. Turns out we hadn't figured out anything. And he was out there changing that. Again, he wasn't the only one. But a mind like that sure would be really interesting. But I also appreciate, you know, I don't know, maybe he'd be tired.

00:04:54:28 - 00:04:59:21
Vivienne Ming
Maybe he eats with his mouth open. Who knows? I'm. I am not a socialite, sir.

00:04:59:24 - 00:05:23:09
Ian Bergman
I well, I love it and I, I what? Here's what I do love also the point that you made. You know, lightning doesn't just strike twice or three times, you know, by happenstance, but also, you know, the notion that this is an example of somebody. And there are a few examples of people who not only had the spark and the idea, but got it into the common lexicon, got it.

00:05:23:09 - 00:05:36:21
Ian Bergman
You know, they they built the thing, they conveyed the idea multiple times, which is not easy to do. I, I enjoy trying to watch and learn from those people because it is kind of magical.

00:05:36:23 - 00:06:01:25
Vivienne Ming
I mean, this is the thing that I have learned as an advisor to graduate students. And I often talk about this. I talk about it in my new book, Why Am I there? My grad students at UC Berkeley and elsewhere are brilliant. They're smarter than me. Yeah. And for 5 to 7 years, 24 seven, they study one thing.

00:06:01:27 - 00:06:23:25
Vivienne Ming
There are 5 or 6 other people in the world that might know this subject to the depth that they know it. For me, it's one hour a week at best. Not even that. Why am I involved? They know everything. They understand nothing. I'm not there to teach them facts. They could go look them up in a library. And of course, in the modern world, they could just ask.

00:06:23:28 - 00:06:49:10
Vivienne Ming
Pick your favorite film. I'm there to teach understanding. I'm there to teach how to explore the unknown. So it's not so much. Tell me about what's going on. And the thing you discovered. If I wanted to study that I'd been an engineer. How did you explore the unknown? What do you do when there's a problem in front of you?

00:06:49:13 - 00:07:13:14
Vivienne Ming
And all the world's facing in a different direction, and you have to push forward on that. And more importantly, to me, how do you teach that to other people so that they can explore the unknown and change the world? Like, that's the thing to me. That's why I jokingly call myself a professional mad scientist is my job is to do that.

00:07:13:15 - 00:07:30:15
Vivienne Ming
But in sort of traditionally ignored areas, you know, how do you reunite orphaned refugees with their extended family? Like, that's a question that that is transformative. If you could share how to solve a problem like that.

00:07:30:18 - 00:07:57:03
Ian Bergman
Well, and it strikes me is, certainly like this is it strikes me that this is actually kind of at the heart of some things that are fundamentally changing around education and critical thinking in today's era. Right. Like when you made the point, you know, setting aside that you're working with people who are one of a couple of experts in their domain in the entire world, right?

00:07:57:06 - 00:08:15:26
Ian Bergman
Any of us can go to pick your favorite search engine and get a superficial understanding of the facts. So what is that thing that pushes beyond facts, pushes to understanding, and maybe pushes to exploration to try and do something different?

00:08:15:28 - 00:08:37:18
Vivienne Ming
Well, I mean, this is what's been interesting in my career is. You know, I get introduced on stages as an AI expert. I am not I don't have a computer science degree. Thank God, it's not a science, and I'm a scientist. I'm also not.

00:08:37:20 - 00:08:38:26
Ian Bergman
Citing the words.

00:08:38:29 - 00:09:08:15
Vivienne Ming
Actual scientist. How in our job as scientists and the job of many scholars is how do you explore the unknown within your discipline? Well, my discipline is people. You know, I'm a cognitive scientist. I happen to be a computational or theoretical cognitive scientist. So I'm in this hoity toity world of can we start from first principles and figure out mathematical models, then explain why we make the choices we make?

00:09:08:17 - 00:09:39:27
Vivienne Ming
Or, how it is that an endocrinologist makes judgments about diabetes treatments or why it is a kid, a brilliant kid from a, you know, a poor neighborhood, has a full ride scholarship to go to university and doesn't go. It seems so irrational if you don't understand how people work. So in reality, what I do is I study intelligence, natural intelligence, artificial intelligence, largely.

00:09:39:27 - 00:10:06:23
Vivienne Ming
I use the latter to study the former. And that's given me this sort of unique insight about, well, if Clod or Gemini or GPT has all the answers and I hear I don't mean like one narrow error area, it has all of the answers for free in my pocket or effectively I had a dollar for a million tokens.

00:10:06:26 - 00:10:37:23
Vivienne Ming
Then why am I there? What am I doing? And in my book, I talk about this idea of well posed an ill posed problems, a well posed problem is a problem where we already know the answer. You know, question answer question answer. That's how we train students in schools. That's how we train AI. Sometimes the question and answer is, here's 10,000 words.

00:10:37:25 - 00:11:05:25
Vivienne Ming
And the answer is what's the next word? But it all boils down to question answer. So we live in a world where now there's this thing that's better than human beings at Well-Posed problems. It's not perfect, but then neither are we. The question isn't, is AI perfect? It's in what domains does it outperform us? And the more well posed a problem is, the more there's no reason to have a human doing the answer.

00:11:05:28 - 00:11:40:09
Vivienne Ming
Except if you don't know how to factorize a polynomial. If you don't like the kind of nerds I hang around with, use orthogonality in common conversations about word. You know, I think the chocolate éclair is really orthogonal to, the trifle as a dessert. Then you're not. You haven't had understanding these problems change the way you think, and the way you think is important because of the other class of problems, ill posed problems.

00:11:40:12 - 00:12:09:19
Vivienne Ming
We don't even know what the question is. Forget the answer. Every problem of interest to me, even the ones that feel like we ought to understand it better than we do, is an ill posed problem. And AI is terrible at it. Turns out we're terrible also. But we're the only game in town. And so when I study humans and machines, solving problems and even cooler my current research solving problems together, that's clearly where they bring their strengths.

00:12:09:19 - 00:12:35:13
Vivienne Ming
The AI brings the well-posed, the human brings the ill posed, which they only got because they worked on well-posed stuff. And that dynamic does something amazing how you make a claim and maybe later will back it up. That right now is the smartest thing on the planet. This cyborg collective of humans and machines, dynamically interacting is where problem solving in the future is going.

00:12:35:15 - 00:12:57:26
Ian Bergman
I don't actually think that's a controversial claim, although I do want to dig into it in a bit. But. But before we do, I want to ask you something that might be one of these really stupid questions. I don't know if it's well posed or not, but how you know that I actually. How do you know that a question is well posed?

00:12:57:28 - 00:13:00:23
Ian Bergman
How do you know that there is a known unknown answer?

00:13:00:26 - 00:13:27:12
Vivienne Ming
I mean, obviously in a sense, you know, it's well posed, maybe only in the sense that it's written down in a textbook somewhere, or was in a training data set. So, as with all things, I mean, this is the truth. Once you really begin to study the philosophy of sciences or just be a philosopher and say the problem of induction, we don't know anything.

00:13:27:15 - 00:14:00:11
Vivienne Ming
That's the sad truth about science. To, quote a, podcast of its own. You know, we're just attempting to be less wrong over time. Doesn't matter how brilliant my ideas are, they will always be wrong. The question is how much utility I get out of them. So from an epistemological sense, wow, we can crack this open and really get into the fuzziness of all this stuff and statistical model free learning and model based and all this complexity.

00:14:00:13 - 00:14:27:08
Vivienne Ming
But essentially, I'm just saying if any other person with a similar education would give the same answer as you, then that's probably a well-posed problem. And that may have been, you know, if that was an esoteric answer to a question that most people don't understand, that probably was the key to unlock a great job for long swaths of recent human history.

00:14:27:10 - 00:14:56:07
Vivienne Ming
And it's just not clear that it will be any more to to, you know, a lot of what we need people to do and have always needed people to do isn't simply review this contract. Look at this x ray. Write this code. It's given the problem solving. What's the right idea for how to code this problem? Given this x ray and this unique patient?

00:14:56:10 - 00:15:24:06
Vivienne Ming
What should we do and why might they be an outlier? And it just hasn't been profitable to think that way in the past, because it just got a lot better by memorizing all of these well-posed answers to those problems. And just the minute a patient walked in the door or, legal client just sum them up, they look like they're will template B and you just pull out and you write it out.

00:15:24:11 - 00:15:56:26
Ian Bergman
If if we if it. Excuse me, I want one more question, Charlie. If if we accept and you don't have to, but if we accept the quote unquote truth that innovation, inevitably leads to behavioral change, some kind of change, like, and as almost a part of the definition of innovation, then it strikes me that the only way to do that is to ask poorly posed questions and answer and get answers that are unexpected and novel.

00:15:57:02 - 00:15:58:24
Ian Bergman
Is that is that fair?

00:15:58:26 - 00:16:33:15
Vivienne Ming
I'm going to take it a step further, 100%. The only way to move forward. If by the very definition that a well posed problem is one we already know the answer to, or at least we have collectively agreed to the answer to, then everything of interest is ill posed. Which means in a world with increasing, you know, I, infiltration, then a lot of that well posed work is gone, and the ill posed is what's left.

00:16:33:18 - 00:16:53:15
Vivienne Ming
And that could sound terrifying if I'm telling everyone your job in the future is being an innovator. Whatever your job was yesterday, that's the one thing you know won't be a job tomorrow. Because if that was economically valuable, then now something else is going to do it. They can learn faster than you and and doing it greater economic value.

00:16:53:16 - 00:17:19:00
Vivienne Ming
So you need to explore the unknown. The wonderful thing about that story is the known as finite, the unknown as infinite. There will always be unknown, and there will always be ill posed problems to explore. The. Here's where it gets more complicated, though. It isn't just about asking these because again, problem with ill posed problems is we don't even know what the question is.

00:17:19:03 - 00:17:32:08
Vivienne Ming
How are we setting up the cultural system, the incentives? How are we building businesses or schools or any of it to make people want to explore the unknown?

00:17:32:11 - 00:17:37:09
Ian Bergman
And that's a tough one. I mean, that doesn't happen, right? I mean, like in a.

00:17:37:09 - 00:18:01:10
Vivienne Ming
Study that my wife and I published together, this was years ago using long before LMS came along, we did a natural language analysis of students in an, discussion forum so they could talk about anything. This was free form text, using some old school NLP. And we analyzed these discussions that the only thing they had to do was show up and they got full credit.

00:18:01:13 - 00:18:21:15
Vivienne Ming
We found it week one. We knew what grade they would get in the course. And what's interesting is what was not what distinguished the failing students from the passing students. You know, it was pretty clear the failing students didn't show up at all. They showed up and talked about the weather. They talked to their friends. They didn't talk about the actual subjects.

00:18:21:17 - 00:18:46:24
Vivienne Ming
What was interesting is the difference between the B and the A's students. The B students gave right answers regularly on demand. If this was a pop quiz, they got full credit and the A's students didn't because the A's students would translate and explore what these were university students in the paper that we published, but we looked at other groups as well.

00:18:46:26 - 00:19:16:22
Vivienne Ming
So you're an MBA student in an economics course and you're learning about collective bargaining. Well, the B student parents, everything they heard in class, the A student says, you know, we talked a lot about auto workers, but let's let's think about what this means for nurses. And they're wrong because nurses are different. But that exploration, our AI, our machine learning system picked up that that predicted not just a better grade in the course.

00:19:16:24 - 00:19:47:13
Vivienne Ming
It predicted that they would actually graduate sooner from university and get a better paying job for their first job out of school. So this huge thing. But keep in mind, again, there was no formal incentive. You got same credit for being right as wrong. And yet the vast majority of students that passed the course wouldn't be wrong in public, even as they saw the elite students engaging in this behavior for a thousand reasons.

00:19:47:20 - 00:20:13:13
Vivienne Ming
We train kids and adults to not take risks, to not be wrong, to not explore. And we could go on and on with this, but here's a my provocative takeaway from research I did on collective intelligence in the optimally intelligent community, including your company. The majority of people should be wrong the majority of the time. If they're not, they're not exploring enough.

00:20:13:15 - 00:20:25:25
Vivienne Ming
And to be clear, I mean productively wrong. But that's what has to be happening. If you haven't built in the culture and the incentives to make that happen, I guarantee you people are hurting around safe answers.

00:20:25:27 - 00:20:33:26
Layne Fawns
In your opinion, what does that culture like? Building that culture and and implementing incentives look like?

00:20:33:28 - 00:20:57:23
Vivienne Ming
So again, as a miner plug, I talk about this a bit in my, upcoming book, which is out in March. And so we, I actually put in some like this is not my traditional style. I like just exploring big ideas. I'm a sci fi nerd. That's what got me into science in the first place. So telling people, here's the five things you need to do is not my style.

00:20:57:26 - 00:21:26:01
Vivienne Ming
But I literally have chapters titled How to Robot Proof Your Kids, How to Robot Proof Yourself, and How to Robot Proof Your Company. So I felt like I kind of needed something. So rather than rules, they were recipes. Here's a recipe. For example, for your kids, the recipe is the nemesis prompt. Don't ask Gemini to write your essay for you.

00:21:26:03 - 00:21:50:26
Vivienne Ming
Tell Gemini that it is your worst enemy. It has gone to your entire career and found every mistake you've ever made, and pointed it out to the world. Instead of writing your essay, you then say, Gemini, my enemy. Here is the essay I just wrote. Tell me everything I got wrong and how I can make it better.

00:21:50:29 - 00:22:19:23
Vivienne Ming
Oh, I don't let I do your work for it. Don't. Here is my golden rule for our entire discussion. I call it the efficiency lie. Don't let I make your life easier. Use AI to make your life harder in the ways that make you better. If you're a leader in an organization, particularly one dependent on innovation. And I'm going to argue that's going to be most organizations moving forward.

00:22:19:25 - 00:22:41:18
Vivienne Ming
Then you need to really rethink this. Why hire a bunch of brilliant people and then tell them what to do? It'd be like me telling my grad students what to do. They know their problem better than me. I understand the problem better. I'm there to feel out the uncertainty. I'm there to role model. What do you do when you fail?

00:22:41:21 - 00:23:05:14
Vivienne Ming
So the kind of directives I give is essentially one point out the power of role modeling. Highly incentivized, culturally incentivize productive failure. When people go out and do hard work to explore an idea that's probably wrong. But if it's not, it would change everything. And sure enough, it turns out to be wrong. That is valuable. It's valuable to your company, and it's valuable to everyone else.

00:23:05:16 - 00:23:22:15
Vivienne Ming
Reward that. I don't know whether that's monetarily or not doesn't really interest me as much, but I have things to say about it as a neuroscientist. But create systems that will reward productive failure and very few organizations engage in.

00:23:22:21 - 00:23:54:18
Layne Fawns
Well, I think I'm almost hearing like a paradox here, though, because it's inherently in human nature to to to utilize something that makes your life easier. And right now, that's the way that people are using AI. Yeah. And and but what you're suggesting is that we should kind of fundamentally go against that nature. And, and so that's why I understand the need for incentive.

00:23:54:18 - 00:24:12:08
Layne Fawns
But what is the future of AI and the relationship between humans? And I look like if, like you said, from your study with those students that most of us are going to lean toward the way of just using it for what's easier, but not to challenge ourselves.

00:24:12:10 - 00:24:46:23
Vivienne Ming
So it's interesting, you said, and earlier, no one would disagree with me that the future of innovation is this dynamic. Here's where every major AI company is disagreeing with me. They're selling AI as an efficiency tour. Let it do all the boring work so you can do the amazing creative work. All right, not to scare away any particular potential book buyers, but I think the one equation I actually have in my book, if my publisher allowed me to keep it in, they took they literally took out all of the dirty words.

00:24:46:25 - 00:25:00:04
Vivienne Ming
I was shocked, they let me keep in all of the the dumb jokes and sci fi references, but no dirty words, apparently. And that that meaningfully reduced the, the actual.

00:25:00:04 - 00:25:02:27
Ian Bergman
The word count. That's about an 8% drop in the word.

00:25:03:00 - 00:25:35:19
Vivienne Ming
Yeah. But, in this book I built it what's called an elasticity of substitution model. So this is really traditional, you know, if you are recent Nobel Prize winner. So Moghalu, has this really notable paper with another famous economist, labor economist, David Otter, and they have this paper looking at what happens, when AI and automation, sort of enter into a skill based market.

00:25:35:21 - 00:25:59:06
Vivienne Ming
Notice I said nothing about skills, high skill, low skill. What does I care about any of that? Nothing. Low skill jobs are probably better protected because it's expensive. I don't want to build a robot to pick strawberries. I I'm barely pay. I'm paying much less than the living wage. Note no legal resident of the United States is willing to do that job.

00:25:59:09 - 00:26:22:13
Vivienne Ming
So, the only reason anyone's ever going to build a robot to pick strawberries is because we don't allow anyone to come from other companies countries to do it anymore, so those jobs are better protected. Dishwashing robot? Why would I ever do that? But I is attacking the professional. I gave an interview to the Financial Times in 2015.

00:26:22:13 - 00:26:56:06
Vivienne Ming
Maybe, and the title was the professional middle class is about to Get blindsided. So this is not a new idea for me here. It's not about skill level. It's about this dimension of how well posed and ill posed a problem is. It's about creativity versus routine. And here's the thing. When we build this into the elasticity of substitution model, if the AI does the routine work, you don't get more creative labor for humans, you get more routine work.

00:26:56:08 - 00:26:59:14
Vivienne Ming
And that may sound paradoxical, but let me put it in.

00:26:59:17 - 00:26:59:23
Ian Bergman
This.

00:26:59:23 - 00:27:45:15
Vivienne Ming
Paradox terms. If an AI bot reads and writes your emails for you, shock of all shocks you get more emails, not fewer. Right? And yet that is the dominant use case. Being pitched by so many organizations today is cognitive automation rather than cognitive augmentation. So when I say the future of innovation is humans and AI interacting together, I mean, both of them working strictly on the creative side of the problem with the AI doing integrative work on the well-posed and the humans with our unique qualities, such as a new term, perhaps for your audience.

00:27:45:18 - 00:28:21:19
Vivienne Ming
Meta uncertainty. Humans are sensitive to our own levels of uncertainty. Donald Rumsfeld is wrong, and he's right. There is unknown unknowns, but human beings are better and worse. There's real heterogeneity in the population about our sensitivity to that. So when we go out of the training set of even a massive model like Gemini three, the AI starts falling apart fast and making things up or staying in the safe place of existing knowledge.

00:28:21:21 - 00:28:52:17
Vivienne Ming
Humans, paradoxically, because we have these narrow lives, because we don't get to see everything, we spend more of our innovation time out in the long tail, particularly those of us for whom it really comes naturally. We need to change education to make this meta uncertainty, to make foundation skills that allow for meta learning, learning how to learn the actual rest of education.

00:28:52:19 - 00:29:17:12
Vivienne Ming
Learn real things along the way, learn what the Treaty of Versailles was about, or all of them. Learn about how to factorize a polynomial while you're learning that. Learn how to be resilient. Learn how to. If you're young enough, work hard on task to expand working memory, numeracy, and literacy. If you're older, work on communication skills. Work on self-assessment.

00:29:17:16 - 00:29:47:29
Vivienne Ming
You can run through this long list of human factors. Then most actually predict not just innovation. It predict how long you'll live, how happy you'll be. A sense of purpose is an astonishingly good predictor of an amazing and long life, including wealth and income levels. Like these are the things we should be training into kids. And as leaders of organizations and communities, these are the things we should be supporting.

00:29:47:29 - 00:29:51:20
Vivienne Ming
Rewarding role modeling inside of organizations.

00:29:51:22 - 00:30:17:28
Ian Bergman
Well, okay, so there are so many places I want to go here, but I as a parent with young children, I want to talk kids for a second because, you know, I think like many parents, I am terrified about whether or not I'm properly preparing the kids for a wildly uncertain future. And as you said, and as our partner doctor, my Kim has said multiple times and innovation talks, a lot of kids are born.

00:30:18:02 - 00:30:54:12
Ian Bergman
All kids, all kids are born with not just curiosity, but a willingness to be wrong, a willingness to ask stupid questions, a willingness to do risky things. And we as a society kind of beat it out of them. Right? So that's a classic innovation problem. But here's what I'm terrified of now, after hearing you talk, it's not just that we beat out risk taking and introduced the concept of embarrassment, it's that we are exposing kids to tools that reduce friction in their lives and therefore reduce a sense of purpose.

00:30:54:14 - 00:30:56:04
Ian Bergman
Oh my God, what the hell do I do?

00:30:56:06 - 00:31:24:15
Vivienne Ming
Let me make this pointed. And I'll start with a tool for mostly for adults, and we can easily generalize in our heads. Every year, the Department of Engineering at UC Berkeley makes the terrible decision to allow me to give a talk to undergrads and I, give a talk about how to solve unsolvable problems to engineering entrepreneurship.

00:31:24:15 - 00:32:09:18
Vivienne Ming
Students. Well, one of the things I give them is this following challenge, and it starts with something I shared on social media. I love social media with a passion that dims the sun, but, I do occasionally use it to broadcast research and, years ago, I shared this claim. Just a claim at the time. There will be a statistically meaningful and and genuinely worldwide meaningful increase in early onset dementia across the one, and it will be causally related to the use of automated navigation systems like Google Maps and others.

00:32:09:20 - 00:32:37:00
Vivienne Ming
I'm not saying this as Google is an evil organization. I'm just saying this is a perfect example. I travel all around the world. I use GPS to be like a native. I like to walk. I hate being in cars trains, but walk is at the top. I'm wandering through the old town of some ancient city. You know, in the southern Europe.

00:32:37:02 - 00:33:00:18
Vivienne Ming
How quickly? I have no idea which way I'm going. So it makes me better when I'm using it. But it makes me worse when I turn it off again. And sure enough, a few years afterwards, the research started to come in. People that relied more heavily on navigation systems were showing worse memory performance and evidence of early, dementia.

00:33:00:18 - 00:33:26:16
Vivienne Ming
And sure enough, even people that just are born into messy, complex cities maybe, Bratislava was one of them, but there was a city. Or maybe it was Prague versus Chicago. Total grit. Turns out people that grew up in these complex cities actually do better with complex navigational patterns. Yeah. Which is almost certainly good for their long term cognitive health.

00:33:26:19 - 00:33:48:14
Vivienne Ming
Well, we don't have to explore anymore. So my challenge for my students is how would you redesign Waze or Google Maps or Apple or any of these? So not only am I better when I'm using it, I'm better than where I started. When I turned it off again. And I get a wave of sometimes clever ideas, sometimes not.

00:33:48:14 - 00:33:50:05
Vivienne Ming
But that's great.

00:33:50:07 - 00:33:51:10
Ian Bergman
What's a good one is that you've.

00:33:51:10 - 00:34:11:22
Vivienne Ming
Got they have to pitch ideas that are wrong to, you know, I hate to break it to you, but there really are bad, questions to ask in a class. But if you never ask them, you're never going to learn. What's a good question? So, so I let them pitch. Then I like to offer the following, because I actually, I actively use this myself.

00:34:11:24 - 00:34:30:12
Vivienne Ming
Doesn't matter where I am, even here in Berkeley. I don't know if there's an accident on Shattuck Avenue that's going to snarl up traffic. So, of course, I turn on Google Maps and take a look. Then, having seen where I am and where I want to go, I try to beat it there by taking a different route.

00:34:30:14 - 00:34:53:05
Vivienne Ming
I think to myself, what do I know about Berkeley and Oakland? That the masses of people using Google Maps do not? Maybe there's a left turn. It wants me to do an unprotected left turn that today of all days will be a nightmare because I know, blunt the bump. So what do I now? I'm thinking for the simple task of going across town.

00:34:53:12 - 00:35:23:18
Vivienne Ming
I'm actively thinking about it, and I'm not engaging in what we call, you know, scientists call automaticity, where, you know, when you whenever you've arrived at a definite destination and you don't even remember driving there. That's terrifying. But that's a natural thing our brains do. Because once you've codified something, why bother thinking hard about it anymore? Well, you should think hard about it because it literally keeps you healthy cognitively across your lifetime.

00:35:23:20 - 00:35:56:18
Vivienne Ming
So to any of you thinking about building technological systems for our lives, particularly the lives of kids, is it not only making them better when they're using it, but are they better than where they started when they turn it off again? A study of endocrinologists in Portugal showed that they are dramatically worse in doing colonoscopies. Now, when you turn off the AI assistants, and a wide range of studies, this is like the golden rule of AI in education.

00:35:56:25 - 00:36:25:26
Vivienne Ming
If an AI tutor ever gives students the answers, they never learn anything which has been replicated without a lens. But, I mean, people have been studying AI in education for 50 years. It's maybe the longest, that applied to I like the the oldest apply it. I feel there are. And I'll tell you, Sam Altman does not seem to know anything about this field, because that's his pitch to the US Congress was imagine every kid has their own AI tutor.

00:36:25:28 - 00:36:31:00
Vivienne Ming
Yeah, I can imagine that because I read that science fiction book, The Diamond Age.

00:36:31:03 - 00:36:41:11
Ian Bergman
Yeah, that is that is Neil Stephenson. Yeah. I who want by the side tangent. Have you, anthem.

00:36:41:13 - 00:36:44:10
Vivienne Ming
I haven't read that one. Yeah. But yes.

00:36:44:12 - 00:37:13:27
Ian Bergman
That one, that one was a mind twist for me. Okay. So look, you're busy. Absolutely terrifying me. But. So I have a question. Is your message getting out there? Are there organizations like I want? Are there organizations that are consciously thinking about the idea that, you know, what is it? You know, the the sand makes the pearl friction and thought lead to better outcomes later versus just giving the answer.

00:37:14:03 - 00:37:34:12
Ian Bergman
And maybe more tactically, for folks that are listening who are in positions of leadership, business, personal, whatever, who are as scared as I am right now. Beyond internalizing these concepts, what can they do? What can they model for their organizations?

00:37:34:15 - 00:38:09:17
Vivienne Ming
Yeah. So listen, there are things that we should fully automate away. Despite what I said earlier, I grew up in the Salinas Valley. You know, John Steinbeck wrote books about the places that I grew up. Human beings shouldn't have to spend 16 hours a day in a field just to send money back to their families. But if we do automate all that away, come up with something real for them to do, rebuild every bridge in the United States and hire anyone who wants the dignity of work to help build them.

00:38:09:19 - 00:38:37:24
Vivienne Ming
So that's one thing, is let's stop lying to ourselves. There are a huge the majority of people on this planet are not ready for the change that I just described. This isn't, an erasure of human potential. It's a simple reality that there's a lot of variability in the world, and we did not support them early in their lives when they really needed to make changes.

00:38:37:26 - 00:38:58:22
Vivienne Ming
So let's find something truly productive for everyone we're going to leave behind. I will tell you, this moment in history, whatever side of the political divide in the United States you happen to lie on to my mind was born in the 1990s, when everyone was promised that globalization would make the world better. Yeah, and it did if all you cared about was the stock market.

00:38:58:24 - 00:39:19:14
Vivienne Ming
But it didn't. If you were a coal miner and everyone was promising they were going to retrain you to be a software developer, turns out you cannot take a 30 year old coal miner and suddenly get a job at Google. That that was a fantasy from day one, that it takes a long 30 years to build a human brain.

00:39:19:16 - 00:39:37:04
Vivienne Ming
It only takes me six months to build a new generation of AI. That's bad math for humanity if we don't start early and make a change. So there's that blanket statement. We need to think of everyone and make certain they're part of this. Otherwise we violate the social contract. I'm scared of what comes next. If we do that.

00:39:37:06 - 00:40:06:22
Vivienne Ming
Having said that, now, how do we move forward here, and see change? I do see organizations lean into it, but frankly, there's terror. So much management decision making is based on the same fear we're talking about for the individual employees. What do you want to do? What will my board reward? What would my CEO reward? It is it's own kind of terror managing uncertainty.

00:40:06:25 - 00:40:32:11
Vivienne Ming
But I'm telling you, if you are a manager at whatever level, that should be your job. Your job is to manage the uncertainty of the context so that the people getting the work done can manage the uncertainty of the problem. And it's easy to say and brutally hard to do. I'm because this is the thing. I think this again, fair to be honest about this is hard.

00:40:32:13 - 00:40:55:14
Vivienne Ming
Everything I'm saying is hard. It's kind of in line with what I am saying, which is stop being shallow and go deep. Stop spending 200 milliseconds of cognitive energy, deciding whether a video is interesting and swiping on it and take some time to think, why don't I like that video? Or why do I? What does it mean about me?

00:40:55:16 - 00:41:20:15
Vivienne Ming
It doesn't have to be every video, but every now and then go deep. I won't take you down the detour. But do we have amazing research showing that kids can spend as much time on social media as their peers and look great? And the reason is, if you look at their metadata, they go deep. They aren't just swiping and pissing away their lives.

00:41:20:18 - 00:41:46:08
Vivienne Ming
They're thinking about the stuff that they're looking at. So, you know, it's a privilege for us to be worried about our kids because for many of us who might be listening to this podcast, your kids are already getting role modeling about what changing the world looks like and the value of, you know, working hard. They may well have some additional genetic endowments.

00:41:46:08 - 00:42:13:20
Vivienne Ming
They are definitely getting role models behavior about how to deal with this world. So here's my favorite clothes for this little concept. The best I tutor never gives students the answers. It's Socrates. What an interesting thing you used to say. The answer is X. You know these people said the answer is X. These other people said it was Y.

00:42:13:23 - 00:42:47:02
Vivienne Ming
Here's some really interesting things you might want to read about them, and some questions you might want to think about. So if it's X then what does that suggest about Z. And it only gives context. Amazing enough in a series of experiments, first with old school AI tutors and now with LM found exactly this sort of I tutor is the only one where pretest post-test students outperformed students that had no AI at all.

00:42:47:04 - 00:43:09:16
Vivienne Ming
So we can dream, we can indulge in the imagination disease. I can imagine a perfect student that uses it the way Vivian's talking about. Therefore we're done. We solve the problem. No, you solve the problem for one tenth of 1% of the world population. The other 99.9 need something different, something that will give them the scaffolding. And I'm here to tell you.

00:43:09:18 - 00:43:13:26
Vivienne Ming
So do your employees. You know, similar percentages.

00:43:13:28 - 00:43:36:12
Layne Fawns
So Ian and I can tell we're slightly terrified, but it sounds like you're actually, like, quite hopeful, like you're out there sort of fighting the good fight and doing the work to promote this and get the messaging out there. How likely do you like, do you see a positive future, and how likely are we to listen to this and put it into action?

00:43:36:14 - 00:44:09:21
Vivienne Ming
I am an AI realist. I have equal scorn for the utopian s and dystopian s. AI isn't a magic wand. I am not Himani Granger. You sprinkle GPUs and TPUs on a problem and they do not go away. This idea of AI superintelligence suddenly, magically solving all problems, even, frankly, as a cognitive scientist, the idea that current a genetic AI are reasoning models is, to me an absurd abuse of the word reason.

00:44:09:23 - 00:44:39:21
Vivienne Ming
And that is very little similarity to human explicit reasoning, which it is implying we can get. Again, nerdy about model free learning and model based, but so that's, for, the nerds that really want to go deep on this. But what it can do is amazing. So while and here's another point of disdain I have for the utopias and the dystopias, they seem to share one fundamental characteristic which is disdain for humanity.

00:44:39:24 - 00:45:07:02
Vivienne Ming
Humans can't be trusted with this stuff. No human can, or only I can defines whether you're on one side of this or the other. I truly believe because of my life, I, I forgive me anyone who's ever heard me share this story before, but it's worth pointing out, especially if you're terrified right now as a parent. When I was little, I was supposed to win the Nobel Prize.

00:45:07:02 - 00:45:29:16
Vivienne Ming
My dad didn't get a chance to win. He was born a sharecropper. Nowhere. Kansas, like Dorothy, has never heard of where my dad was from. The Vietnam War sort of changed the trajectory of his life. He was brilliant, but he got a full scholarship everywhere, and he can go. I mean, he went to Ku, tutored Wilt Chamberlain in Bath and not in basketball and chemistry.

00:45:29:19 - 00:45:58:07
Vivienne Ming
And, but, you know, Vietnam change the trajectory of his life. He was a flight medic. And became a doctor in California. So then I grew up the child of a doctor in coastal California. All the potential he had and all of the opportunity, he didn't. And I ended up homeless. I spent a significant chunk of the 1990s not knowing where my next meal was coming from.

00:45:58:09 - 00:46:20:27
Vivienne Ming
$0.49 is the price of a box of Rice-A-Roni back then, because it meant two quarters and I got to eat, well, all self-inflicted. Then in 1999, I got my life back together and I went back to the same school I flunked out of. And I did my whole undergraduate degree in a single year, and I got perfect scores in every course.

00:46:21:00 - 00:46:44:17
Vivienne Ming
And what I got introduced to was this place called the Machine Perception Lab with the eyes of Paul Ekman, who passed away recently, and Terry Sinofsky, the student of John Heartfield, that Nobel Prize winner doing with $5 million from the CIA to build AI models, to analyze people's faces, to tell if they were lying or not. I'm a sci fi nerd.

00:46:44:17 - 00:46:58:09
Vivienne Ming
That's what I did when I was homeless, is I read science fiction and fantasy books that kept me alive. Now I had the chance not simply to study people, but to build science fiction technologies to study.

00:46:58:12 - 00:46:59:12
Ian Bergman
Yeah.

00:46:59:14 - 00:47:23:15
Vivienne Ming
And I went on to continue to do that, and, graduate school and have CIA funded lie detector sounds ethically complex to you. Note that I got to take what I learned in doing this and build the system to reunite orphaned refugees with their extended family members. I got to take it for Google Glass behind me here, and build a system for autistic kids to learn how to read facial expressions.

00:47:23:18 - 00:47:54:27
Vivienne Ming
Based on what I learned in that open science project. That's why I'm bullish on what I can do. I fundamentally believe everybody on this planet, in their own way, can do what I do. I have my unique strengths. Everybody has a unique strengths. Turns out, importantly, everyone has their own unique weaknesses as well. That turns out to be a big predictor of life outcomes is coming to terms with that and figuring out ways of dealing with it.

00:47:54:29 - 00:48:13:18
Vivienne Ming
That's what I did to those hard years in the 90s. And and then I got a chance to grow up. When I see what I can do, I have an itch to scratch. And I'm the chief scientist of a company called Guild. And I've got data on 122 million working professionals. Why do women get paid less than men?

00:48:13:21 - 00:48:33:13
Vivienne Ming
Every labor economist in the world, including some of those Nobel Prize winners I mentioned earlier, have papers about this, and they all come to the same. Finding women were fewer hours than men. So if you're a more progressive economist, we need better family leave policies. If you're a conservative economist, you say, look, it's not my fault I didn't do anything wrong.

00:48:33:15 - 00:48:53:26
Vivienne Ming
When I look at that and I say, well, there are women literally from Venus, why are rational people making seemingly irrational choices? They worked hard to get that job. And now suddenly, out of nowhere, they're choosing not to work as hard as they all wanted to suddenly raise families. And they'd never thought about that before. Come on.

00:48:53:29 - 00:49:27:01
Vivienne Ming
So I built a bunch of A's. They crawled through the websites of 60,000 companies and all the data we collected. And two days later, I had novel insights that no one had ever published before about why women choose to work fewer hours than men on the job. Two imagine when I did this work in 2015, the Army of Research assistants, the masses of data, the years it would have taken to do that research.

00:49:27:03 - 00:50:13:29
Vivienne Ming
Two days. But again, I didn't build an AI that did the hard work for me. I built it. Noel Elmes back then. I had to build it all by hand. I built it, ran it, explored the results, reconfigured the models, ran it again, reconfigured this dynamic of human and machine working creatively together. I seen what people can do when cyborgs, this fundamental integration of human machine truly becomes achievable rather than it does the boring stuff or the scary, deep professionalization of, well, once it's doing the boring stuff, why do we even need someone with a fancy degree doing the job at all?

00:50:14:02 - 00:50:18:24
Vivienne Ming
Why don't we just hire a high school grad and have the Jiffy Lube colonoscopy?

00:50:18:27 - 00:50:37:23
Ian Bergman
That's a that's an image we're not going to get out of our head. But but okay, so I have a big question. And maybe a smaller but important question. The first the first one, what did your research tell you and where do people go and read about it in terms of the the answer to the ill posed question?

00:50:37:25 - 00:50:39:19
Ian Bergman
So rational actors.

00:50:39:24 - 00:50:52:25
Vivienne Ming
I haven't mentioned any book titles yet behind me. If you can make it out, there's posters, the one in yellow, robot proof when machines have all the answers, build better people.

00:50:52:27 - 00:50:55:04
Ian Bergman
March 2026.

00:50:55:06 - 00:51:19:14
Vivienne Ming
Yes, that is coming out in a few months. I think anyone that has heard this is, I think, legally obligated to not only buy copies for themselves, but to preorder for all of their Christmas list. Feel free to throw in Hanukkah list. I mean, honestly, isn't this better than socks? So, then the next book after that will be one of two because they're already all written.

00:51:19:14 - 00:51:42:07
Vivienne Ming
Because to me, I'm not really writing books. I'm just telling stories out of my research. So really the that the gist is I tell the story of doing the research. Some of these is literally the story of my son being diagnosed with type one diabetes and me hacking all of his medical equipment and breaking all of US regulations and building the first day AI for diabetes.

00:51:42:14 - 00:51:55:00
Vivienne Ming
Some of it is the story of really grounded, topic specific research. So the next two books will figure out the order probably depends on the appetite of America. Is the tax on being different?

00:51:55:02 - 00:51:55:13
Ian Bergman


00:51:55:20 - 00:52:26:09
Vivienne Ming
Which will how's this research. That I just referred to about gender wage gap and small sacrifices. The subtitle of that book is The Science, economics and Story of Purpose, is bookended by two amazing experiments I did inside massive organizations. I'm crazy enough to consider helping giant multinationals to be a philanthropic venture. As long as I get to tell everyone what I got, what I did.

00:52:26:12 - 00:52:49:29
Vivienne Ming
And so that one really looks at the worst, about what makes people people and the best of what makes people people and how, in fact, we are exactly that same person all of the time, and is the context that makes us that terrible person to that amazing person. But I'm a hard numbers scientist, so it's what would a mad scientist say about the concept of purpose?

00:52:50:02 - 00:53:28:13
Vivienne Ming
So the answer, the biggest answer came from facial recognition when it crawled out across the data bases of the companies of all these websites, the number of female faces on the leadership team was the single biggest predictor of the single biggest covariate of gender wage gap, controlling for programs, controlling for, all sorts of different variables. If there was a female, CEO or CFO, younger women inside the organization put in more hours.

00:53:28:15 - 00:54:08:05
Vivienne Ming
And it you could measure it causally when a new, when a new CEO came in, regardless of change in policy, young woman put in more hours into their job. They work hard. But the question is, are they going to work 70 hours hard on their job because they love it? Or are they going to work 50 hours hard on their job and put 20 hours into the church, or into a social mission because they looked up and thought, I have no evidence that working hard is going to pay off.

00:54:08:07 - 00:54:39:09
Vivienne Ming
I know intellectually, so this is me. The cognitive neuroscientist. When I look at human behavior and I look at how our knowledge, our explicit knowledge about the world and our implicit beliefs born of our lived experience, when those two diverge, our behavior invariably tracks with beliefs, not with knowledge. So when you hear these brilliant students passing up scholarships saying, yeah, I get it, I understand they do.

00:54:39:09 - 00:55:04:11
Vivienne Ming
They understand it's not a marketing problem. The problem is they have no lived experience, that going off to another planet and hanging out at MIT is actually going to change their lives. And that drives their choices so much more than their brilliance does. And you see it in young women inside organizations, and you see it across so many different domains.

00:55:04:16 - 00:55:33:21
Vivienne Ming
But here's a, friend, forgive me. I'm on my soapbox now, here's a broad claim. Most of the complex problems I work on aren't technical problems. They are simply we are mistaken about how people make choices in their lives. And it's hiding often simple technical solutions for us, because we just can't imagine a brilliant kid would pass up a full scholarship.

00:55:33:24 - 00:56:03:16
Vivienne Ming
We can't imagine a hard working young woman wouldn't want to continue to put in hours, even if there was no obvious evidence. By the way, I'm not talking about whether you've ever been caught up in the office before, which I've never seen in a labor economics paper. All of these domains in which I work often boil down to this imaginary world in which we try to solve problems and the real world that desperately needs our help.

00:56:03:18 - 00:56:23:27
Ian Bergman
And I mean, that is utterly fascinating. And I think we could probably spend another seven hours on the topic of how people make decisions and how they perceive the world. But, you know, it is really interesting because I can connect it to moments in my life when people around me and in my family have made very unexpected decisions.

00:56:23:27 - 00:56:51:00
Ian Bergman
And there's one that I thought I'd share because I, you know, I think this is interesting and this is, an extended family member, who grew up and I, you know, frankly, tougher economic circumstances was not exposed there. It was taken from her parents, things like that. And it's interesting. And at some point when I was much younger, I had a well-paying white collar job and I said, look, like you're in your teenage years.

00:56:51:02 - 00:57:13:15
Ian Bergman
I think one of the most important things anyone can do is go explore. Here's a free plane ticket, here's free cash to, you know, live the hostel life for a month. That was 15 years ago. And that offer has still not been accepted. And it is always blowing my mind. And I love this person and. But.

00:57:13:18 - 00:57:35:21
Ian Bergman
And what I what I did not come to realize until I'd say the last few years. But even this puts crystallization on. It is like, you know, this person had no lived experience. To understand why I believed that that was such an amazing opportunity and why I would change their life. And it is really interesting. So like, I guess I don't even know where I'm going with this.

00:57:35:21 - 00:57:40:08
Ian Bergman
I, I'm telling the story because I think it makes a lot of sense. But I guess.

00:57:40:11 - 00:58:08:04
Vivienne Ming
I mean, here's a fascinating parallel. There was this paper that I don't remember exactly how it came to me. I follow, Tyler Cowan's blog. He's sort of this notable libertarian economist, partially because I don't agree with him a lot. But I think he's a reasonable person. I don't need everyone. It is true. I am always right, but I'm a big enough person to allow other people to be wrong.

00:58:08:06 - 00:58:19:26
Vivienne Ming
So I'm happy to allow Tyler to be wrong in my feed regularly. So I think he must have shared this paper. That was an argument in favor of payday lenders and the experiment. That's why.

00:58:19:27 - 00:58:21:03
Ian Bergman
Alton.

00:58:21:06 - 00:58:47:15
Vivienne Ming
The experiment that they ran was, a check on the rationality of actual people that regularly use payday lenders. And sure enough, in their experiment, these people were very accurate about whether they would need it again in the future, their ability to repay again. You ask them all of these explicit knowledge questions. They were right on if that was the whole story.

00:58:47:18 - 00:59:12:22
Vivienne Ming
They have a reasonable argument. I might still quibble with the value of payday lenders for a variety of reasons, but at least your argument that people should be allowed to make their own choices is supported. In a footnote in the appendix, they noted that they they motivated the participation of the people in the study by offering them a $200 Amazon gift card.

00:59:12:25 - 00:59:42:13
Vivienne Ming
Now, these are people regularly going to a payday lender to make ends meet 200 free dollars. They mentioned that I think their statement was 60% of the gift cards were not redeemed. That's the headline. Like that should be the abstract of the study. It should. That is everything they just said. Unfortunately, 50% of gift cards were not received means our entire study is bunk.

00:59:42:16 - 00:59:46:27
Ian Bergman
I'm looking at lane space here and Lane as a stunned as I am.

00:59:47:00 - 01:00:20:25
Vivienne Ming
Yeah. So, and we can look on the other end. Here's another of my favorite crazy findings. You look at fund managers and how they shift investment money around and what they found. I love this because it's so absurd that in Europe, European based fund managers, say, a Norwegian moves their money out of companies headquartered in Eurovision competition countries.

01:00:20:28 - 01:00:34:15
Vivienne Ming
It's not a huge effect, but is statistically detectable. Tell me again how people managing the trillions of dollars in our financial system are always 100% rational, because they got real money on the.

01:00:34:21 - 01:00:35:18
Ian Bergman
Nobody's rational.

01:00:35:24 - 01:01:06:11
Vivienne Ming
Nobody's rational. We are human. In fact, I'm going to make an argument. All cognitive systems are biased. You cannot build an unbiased AI. You cannot build an unbiased human or cockroach or rat or monkey. All systems are biased because my definition of intelligence is any system that can make decisions under uncertainty. If you already knew the answer, then you know we might call that robotic process automation.

01:01:06:11 - 01:01:15:04
Vivienne Ming
It's no longer an intelligent system. If you're simply a robot on a factory line doing the exact same thing every time, it requires bias.

01:01:15:07 - 01:01:15:23
Ian Bergman
But do.

01:01:15:24 - 01:01:17:25
Vivienne Ming
Mathematically.

01:01:17:28 - 01:01:28:02
Ian Bergman
Do machines augment human rationality in this cyborg model? In this model where machines help build better people? Or is that kind of just a false question?

01:01:28:05 - 01:01:53:23
Vivienne Ming
So let me offer two studies. One mine, someone else's. I'll do the other one first. They're both what I'm going to call evidences of hybrid collective intelligence or high hybrid intelligence. These are the cyborg collectives I referenced earlier. I can think of four existing papers plus mine, which hasn't come out, but it's time to come out with the book.

01:01:53:25 - 01:02:19:09
Vivienne Ming
Two of those papers are in medical research. I'm happy to share the link so you can post them in the show notes. One is in product. Development. And I'm blanking on the last one, but I'll come back to it. Mine is in human beings competing against prediction markets. Individual humans, or teams of humans competing against pi.

01:02:19:12 - 01:02:21:26
Ian Bergman
In a world of poly market, this is quite a thing.

01:02:21:28 - 01:02:50:18
Vivienne Ming
Yeah. So in, the, the, I think the best case in the study, medical diagnostics, this paper that I believe was in PNAS, but I could be wrong. And I will dig it up for show notes. What they found was when they put together teams of humans and teams of machines, that they made complementary errors.

01:02:50:21 - 01:03:23:10
Vivienne Ming
That's a phrase right out of the title of the paper. So that the the process was humans. All the diagnostic data comes in. Humans make notes, AI collects it, it presents an initial observation. Humans riff off of it, take it to new places, AI collects it, makes new observations. That system outperforms everything else. It's the smartest thing on the planet today.

01:03:23:12 - 01:03:54:05
Vivienne Ming
That's a big argument, isn't, you know, NYSE the smartest thing on the planet or Poly Market. Well, here's where my research comes in. It's still ongoing, so I, I, I can't make the absolute claim until we've finished all the analysis and it comes out. But the trend so far is, we bring people into a room, we give them an hour and they are given 30 issues off of Poly Market.

01:03:54:08 - 01:04:17:12
Vivienne Ming
And you know, that have some resolution criteria. So we know what we can just track Poly Market, see whether our people were accurate or not. So if people buy themselves AI's by themselves, people interacting with AI in the special condition, people interacting with a specially designed, a genetic model, humans are not great. Like, these are smart, but naive people.

01:04:17:12 - 01:04:35:24
Vivienne Ming
Of course they aren't going to outperform Poly Market on these issues. The AI is are better as a rule, as a sort of generic rule. I tend to perform better than even than maybe teams of humans. They only have an hour and they have to go to 30 items. So the machines are just better at that kind of time pressure.

01:04:35:27 - 01:05:04:17
Vivienne Ming
That the best of the AIS, plus the human humans appear to be at least competitive with Poly Market for judgments made in a single hour's worth of time, with no pre study, and best when the outcomes are unexpected. The things that are the most interesting. But here's the most fascinating of our findings. It's not the generation of the LM.

01:05:04:17 - 01:05:32:16
Vivienne Ming
The matters you could use. You know, GPT five or Gemini three. I use Gemini three a lot in my work since a lot of it is very nerdy bio stuff, and I think Google probably has a better handle there. It that versus an open source model, the, Lambda model or even a condensed model doesn't matter that much is a detectable difference, but it doesn't matter.

01:05:32:19 - 01:05:55:21
Vivienne Ming
It's the human capital on the team that matters if and effectively it matters in the way that interests me as a nerdy scientist. It's a dynamical system. When you have enough human capital on the team, there's like a phase transition. They don't just take what the AI says, they interact dynamically with it, and they don't just.

01:05:55:22 - 01:05:57:15
Ian Bergman
Validate the concept of prediction.

01:05:57:15 - 01:06:02:21
Vivienne Ming
Markets. Effort. Does it invalidate or validate? What does.

01:06:02:21 - 01:06:13:09
Ian Bergman
Invalidate? Does it doesn't this kind of reinforce the basic thesis of the prediction markets which are trying to aggregate the maximum amount of human capital into this kind of machine assisted decision making?

01:06:13:12 - 01:06:35:27
Vivienne Ming
Absolutely. I just think it's really interesting it being able to see the collective intelligence of the predictive prediction market, because they're not getting to look at all they're getting, is the questions, what role will the price of oil be above or below this threshold? Will Russia accept a treaty by this date? All these sorts of different questions.

01:06:36:00 - 01:06:51:29
Vivienne Ming
And when those questions resolve, our smart humans. Plus I look in an hour's time, look very competitive with the full scope of politics as.

01:06:52:01 - 01:06:56:04
Ian Bergman
That is utterly fascinating. But you have to give interesting.

01:06:56:04 - 01:06:59:15
Vivienne Ming
Implications about what was I.

01:06:59:17 - 01:07:09:15
Ian Bergman
I was going to say, but you have to give the hour's time. Like your key point is you have to give the Irish time, the back and forth, the interaction and the, you know, the engagement with the ill posed question, I guess.

01:07:09:18 - 01:07:10:00
Vivienne Ming
Yeah.

01:07:10:04 - 01:07:10:24
Ian Bergman
Is that so?

01:07:10:26 - 01:07:42:12
Vivienne Ming
There's some interesting findings that the study, there's this fascinating sort of, Procter and Gamble study is another one of these domains. The headline funding finding there was, one person with an AI can outperform a whole team. It was like, a hackathon inside, Procter and Gamble. And they bring in engineers and market people, and they work together like strangers, and they have to come up with products and so forth.

01:07:42:15 - 01:08:15:06
Vivienne Ming
And it is true, a person with an AI more particularly an engineer with an AI to replace a marketing pair, did, as well as an engineer and a marketing person. So that was almost the entire effect was a lift of the engineer to come up with a better product pitch, essentially. But what that should have been the headline, if you dug into the findings from the paper, was teams of people working with AIS, not first order.

01:08:15:09 - 01:08:59:17
Vivienne Ming
Did judges judge their idea to be better or worse? Human judges are wildly swayed by how slick the marketing company is. LMS are great at that. Did the idea get selected for development? Did it win an award at the pitch competition? The human plus AI teams were three times as likely as any other combination to get selected to go on and meet the world in additional research by other groups, humans code developing arguments or writing essays with AI support more persuasive arguments, and learned more, and other people learned more from their essays.

01:08:59:20 - 01:09:24:21
Vivienne Ming
Cyborgs. This is when I was going off to study in grad school. I told people I want to build cyborgs, and they justifiably scooted away from me for fear that the crazy would rub off and none of us would get in. It's actually a field called Neuroprosthetics, but I like the word cyborgs more. Sure. I want us all to be the Borg, but in a good way.

01:09:24:23 - 01:09:26:15
Layne Fawns
Not a Skynet way.

01:09:26:17 - 01:09:37:29
Vivienne Ming
Yeah. And, Okay, so maybe I don't want it. I want to be the Borg, but I want the the the cybernetics to actually augment what makes us different.

01:09:38:02 - 01:09:39:20
Ian Bergman
Yeah, individuality. You know.

01:09:39:24 - 01:09:53:26
Vivienne Ming
I alluded to this in my study. We had that special case of a specially designed, a genetic AI. First, it didn't give answers. Second, its job was to push the humans apart.

01:09:53:29 - 01:10:20:08
Vivienne Ming
Push them to create their own unique insights. Don't allow two people to talk together and agree with each other. Make them think about it on their own before they bring their ideas back. It it was trained to do a kind of ideation integrating take ideation integration phase and that includes itself. Don't let the humans just take your answers and run with it.

01:10:20:11 - 01:10:50:05
Vivienne Ming
And even in lower human capital team that AI they gave less information and therefore should be worse. They are also had that elevated performance. So we can build AIS that make people better if we do that. I'm bullish on the future of humanity. I'm not a futurist. I run a philanthropy called the Human Trust. And people bring me challenging problems and I help them for free.

01:10:50:07 - 01:11:13:25
Vivienne Ming
That's my day job. But the reason I am able to do this is because of the dumb luck that I got introduced to this field in 1999, when nothing in my life was going right and everything transformed. So what I want to say, rather than this is how the future is going to be, is this is a choice we have.

01:11:13:27 - 01:11:31:09
Vivienne Ming
We can stay shallow or we can go deep. We can make this choice collectively as a society. You can make it individually, as companies. I'm just saying the shallow is a dead end. And as mentioned already, I'm always right about everything.

01:11:31:11 - 01:12:01:03
Layne Fawns
Vivian, I could I want to keep this going for forever. But I think that that last sentiment was it was the answer to my question for for the final question of this talk, which was, what's the big piece, the big takeaway for our listeners here? And I think you hit the nail right on the head. If folks want to follow you, I know that you don't like social media with With the Passion of the sun, but, if if they wanted to follow you, connect with you, preorder.

01:12:01:03 - 01:12:22:22
Vivienne Ming
If your self-appointed day job is to help people for free, you got to kind of be available. So if you go to Soco start org associates.org, you can sign up for my free newsletter. If you're a fool, you can sign up for my paid newsletter. And the first one I just share research I think is interesting.

01:12:22:22 - 01:12:37:00
Vivienne Ming
And the second one I talk about what I would do with that research. You'll learn about my book. You'll find out what you can follow me on social media, where I will invariably share my science fiction and fantasy book picks.

01:12:37:02 - 01:12:59:02
Ian Bergman
Amazing. Well, maybe. And, we'll get that in the show notes. I am sure you've got new followers. You've got new followers. Here in the hosts of this podcast. And we really appreciate you coming on. Innovators inside. I want to say thank you for your time. It was a fun and fascinating conversation, and I want to just wish you the best.

01:12:59:02 - 01:13:01:27
Ian Bergman
As we head into the weekend and into the end of the year.

01:13:02:00 - 01:13:06:11
Vivienne Ming
It was a blast, and I didn't even need to tell any dirty jokes made by my publishers. Right?

01:13:06:13 - 01:13:11:14
Ian Bergman
I know that's part two. That's part two. We're going to bring it back. And the whole episode, Dirty Jokes.

01:13:11:15 - 01:13:15:08
Vivienne Ming
Well, I have the second twisted, version of the show.

01:13:15:10 - 01:13:16:07
Layne Fawns
Yeah, I love that.

01:13:16:07 - 01:13:18:03
Ian Bergman
Wait. Can't wait.

01:13:18:05 - 01:13:22:07
Vivienne Ming
Nothing but I necrophilia. Hahahahaha.

01:13:22:09 - 01:13:26:04
Ian Bergman
And that's our outro. Thanks so much Vivienne. It was an absolute pleasure.