Everyone Is Asking The Wrong Question About AI | Ep. 422 with Rana Gujral CEO of Behavioral Signals

0:00We're officially in the era where hearing is no longer believing. Wall Street Journal ran a story about a mother who got a call from someone who sounded exactly like her daughter, panicked, crying, begging for help. It was just not her. This is Rana Gujral. Inc. magazine named him an entrepreneur to watch. He's the CEO of Behavioral Signals. And after years studying how machines read human behavior, he's landed on something uncomfortable. What you're about to hear might change the way you think about AI. The important question isn't how clever it makes you sound in the moment, it's whether... The terrifying part is that it...

0:46What is the biggest misconception that you think people have right now around AI? Yeah, I think the biggest misconception, honestly, is that the debate is about capability. Like, can it do this task? Can it do that task? Will it pass this benchmark? When will it beat humans at X? that is the entire public conversation and I think it's the wrong axis the real story isn't what AI can do it's what using AI does to us over time that's the shift I keep trying to get people to see because once the system sits inside your loop of attention and judgment the important question isn't how clever it makes you sound in the moment it's whether it's building your instincts or quietly just replacing them.

1:39So I'll give you the version I noticed in myself, right? So at some point, I stopped using AI the way I use, say, a calculator, like something you pick up and put down. I started using it in the way I use my own mind, like to draft thoughts, to test arguments, to frame decisions. That felt like productivity, and it is. But there's a second thing happening underneath that nobody talks about. Like, am I accumulating judgment or am I generating a convincing stream of outputs that feel like thinking without actually being it? And I think that distinction matters. Almost no one is asking it because we're all fixated on capability benchmarks.

2:23And the other big misconception related to that is people imagine harm from AI as something dramatic, like a rogue system, a job apocalypse, some rupture moment. That's going to happen. And what I actually worry about is something more dangerous. It's the drift. The most consequential changes AI brings don't arrive as ruptures. they arrive as defaults, like decision pipelines that surface a recommendation before you formed an opinion, like attention systems that learn your psychology with unsettling precision. And none of that is coercive. It's just convenient. And the convenience is more than enough to reshape a life or society without anyone noticing until it's done.

3:13So I think what I would correct one thing in the public conversation, it would be this. It's like, stop asking what AI can do. I mean, start asking what it's doing to shape the human thought. And I think that's the real frontier. Well, it seemed like we were so excited about AI, specifically generative AI, being able to replace so many tasks, being able to free us up. But a lot of the research is showing that many companies are not getting any better. So employees are not really adopting it at a corporate level. And they're not really finding that it's really great at replacing people. even. Many companies are hiring people back.

3:51What are you seeing in terms of this being our savior? I built Founder's Story from a$50 microphone. And the most important thing is I didn't do it alone. For years, I've been using Upwork to hire marketing, editing, branding, you name it. In fact, the editor who cut this very episode and the team behind all of Founder's Story branding found them on Upwork. The quality of people is top notch and paying people is simple. Upwork is a one-stop platform to find, hire, and pay expert freelancers across development, data, marketing, operations, and more. With Business Plus, you can access the top 1 % of talent on Upwork.

4:37And with AI-powered shortlisting, you'll get matched to the right freelancer in under six hours. no endless searching required it's free to sign up and posting a job is easy visit upwork.com right now and post your job for free that is upwork.com to connect with top talent ready to help your business grow that's upwork.com upwork.com yeah i mean the gap between the hype and what what's actually landing inside companies is very real and I think it's healthy that we're finally talking about it honestly. Here's what I see. The initial wave of enterprise AI adoption was driven by what I would say is a fantasy of substitution, right?

5:27Take a workflow, drop in a model, remove headcount, book the savings. And it turns out that almost... I built FounderStory from a$50 microphone. And the most important thing is I didn't do it alone. For years, I've been using Upwork to hire marketing, editing, branding, you name it. In fact, the editor who cut this very episode and the team behind all of Founder's Story branding found them on Upwork. The quality of people is top-notch and paying people is simple. Upwork is a one-stop platform to find, hire, and pay expert freelancers across development, data, marketing, operations, and more. With Business Plus, you can access the top 1 % of talent on Upwork.

6:15And with AI-powered shortlisting, you'll get matched to the right freelancer in under six hours. No endless searching required. It's free to sign up and posting a job is easy. Visit Upwork.com right now and post your job for free. That is Upwork.com to connect with top talent ready to help your business grow. That's U-P-W-O-R-K dot com Upwork dot com You know, that's almost never how work actually flows I mean, real work is full of exceptions Judgment calls, tacit contacts, relationships, etc And the model handles the middle 60 % nicely And then falls off a cliff on the edges And the edges are where the values live, right?

7:02So the company's deployed, measured, and found that productivity gains didn't materialize the way the deck promised. And some of them are now hiring back because they cut into the muscle, not fat. The second thing I point to is that most enterprises try to bolt AI onto processes that were designed for humans doing the whole task. That's like buying a Formula One engine and putting it in a mini WAN. So you have to redesign the vehicle. I mean, the companies I see actually getting value are the ones rebuilding the workflow around a collaboration, not just inserting a chatbot into an existing pipeline.

7:37And I think then there's the adoption piece, right? Which is, I find the most interesting, to be honest. I mean, employees aren't refusing to use these tools because their validity is, right? I mean, they're refusing because nobody's answered the basic question of what's in it for them. If I use this thing and I become twice as productive, does this mean that I get to do more meaningful work? Or does it mean I get more work piled on? What does it mean, you know, or does it actually mean I'm trading my replacement? You know, and until that question is answered, I mean, you're going to get quite resistance.

8:10And you're getting it, you're not necessarily seeing it on the dashboard yet. So, I mean, to your question, I mean, is AI our savior? I don't think so. It was never going to be. It is just a powerful capability that reshapes what's possible. but the returns show up when you rethink the work, invest in people and stop treating it as a cost-cutting shortcut. I mean, the ones that are doing that are winning and the ones that are chasing substitution, I think they're the ones that are writing the disappointed headlines. Well, I think it's really funny that we told people basically use this tool that we already knew will probably replace you at some point.

8:53And then we're shocked that they didn't want to use the tool. It's really funny. As we know, there's always a disconnect from the top executive level down, right? Something you said, though, that stuck with me is that you're worried about AI developing consciousness beyond human control. How real is that? I mean, I'd say, let's be precise about what we're actually talking about, right? So because consciousness is doing enormous work in that sentence, and most people using that word haven't defined it even for themselves. Here's what I think is happening, right? So we're seeing systems that are extraordinarily good at producing outputs that feel conscious.

9:34Stuff like fluent language, empathy, memory, personality. And the human brain is exquisitely tuned to attribute mind wherever it sees those signals. It's the same instinct that makes us name our cars, right? And when a model responds with warmth and nuance, our wiring says, is there something, someone in there? like that's a projection uh not really evident so the harder truth is that we don't really actually have a consensus on what generates consciousness in us um where the nerds neuroscientists are still trying to argue about whether it emerges from specific network dynamics or whether it's a global workspace phenomena whether the brain generates thoughts or receives it and if you can't explain the one working example we have that one working example i mean claiming about how to replicate it in silicon is at minimum, I think, just premature.

10:29So I don't, I mean, I want to be careful. I'm not in the camp that says machine consciousness is impossible. Dehan and others have made a reasonable case that if you replicate the right computational signatures, denying subjectivity to that system becomes just as arbitrary as denying it to other human. That they may come, but there's a big difference between, you know, may come and around the corner. And so what I, you know, I rather focus people on is something I call artificial general experience, you know, not whether the machine is conscious in some metaphysical sense, but whether it starts to have stakes, like, you know, preferences that shape behavior, something that functions like caring about outcomes.

11:15That's a threshold that actually matters for how we treat these systems. Quick break to talk about something I get asked about all the time on this show, Bitcoin. For years, I assumed it was complicated. Turns out it really doesn't have to be. If you've been curious about Bitcoin but haven't made the jump yet, Cash App makes it easy. You can set up automatic purchases with zero fees or buy larger amounts also with zero fees. Start small or go bigger. It's designed to be simple either way. For a limited time, new customers can get$10 added to their balance. Just use code BITCOIN10 when you sign up and don't forget this part.

11:56Send at least$5 to a friend in the first two weeks. Terms apply. Cash App is a financial services platform, not a bank. Banking services provided by Cash App's bank partner. Bitcoin services provided by Block Inc. Brand. For additional information, see the Bitcoin disclosures at cash.app slash legal slash podcast. Much more of a tractable question than the hard problem. What are you seeing in terms of voice deep fakes and being able to stop a lot of this? Because I think we're starting to already see people are calling you or they're calling companies acting like different people and people don't realize it.

12:36I've been called a few times and I'm wondering if this is a human. The funny thing is, every time someone calls me, I ask, are you human or are you AI? Yeah, I mean, I think you're touching on to something that does keep me up at night, honestly. I mean, we're officially in the era where hearing is no longer believing. You know, that phrase is dramatic, but it's just true now. I mean, you know, you've seen examples like, you know, earlier this year, Wall Street Journal ran a story about a mother who got a call from someone who sounded exactly like her daughter, panicked, crying, begging for help.

13:11It was just not her. I mean, it's synthetic voice generated by AI. And the terrifying part is that it isn't that it just happened once. I mean, the tools do it consistently. 11 Labs, Meta Voice, Open Voice, they're openly available. I mean, there's no barrier to entry anymore. I mean, all you need is like three seconds of your voice from a social media clip and you can clone. So on the technology side, on the defense side, most detection has traditionally leaned on what we call vocal biomarkers. So the little imperfections in real speech, stuff like micropauses, pitch variation, throat resonance, tiny jitters caused by breathing or emotion.

13:55And the assumption was synthetic voices couldn't fake those consistently. That assumption is dying fast. Models like HiFiGAN and Natural Speech 2 are patching those tells one by one. So, like, I mean, I actually compare this to the diamond industry. Like, lab-grown diamonds used to be spotted by their lack of natural inclusions. Because chemically, they're still a diamond. Now, manufacturers add synthetic inclusions during the growing process, and you can't tell anymore. I mean, as the same story with voice. So, you know, so I think what we've pushed in the last several years on our end is we pushed on behavioral mapping and temporal modeling.

14:38So instead of asking, does this audio clip look real? We ask, how does this specific person actually speak over time? You know, this is our technology at behavioral signals, right? And so, for example, focused on the cadence, their articulation, co-articulation patterns, the rhythms of how their porosity moves across sentences. I mean, those temporal signatures are much harder to fake, even when a window sounds perfect. But, you know, this is an arms race. Every detection gets studied by the generation side and vice versa. The real answer isn't purely technical. I mean, that's a cultural question at this point.

15:13We need to normalize verification. I mean, like, you know, maybe at some point we'll need a code word with your family, you know, call back to a known number, skepticism as default when emotional pressures are high. I built Founder's Story from a$50 microphone. And the most important thing is I didn't do it alone. For years, I've been using Upwork to hire marketing, editing, branding, you name it. In fact, the editor who cut this very episode and the team behind all of Founder's Story branding found them on Upwork. The quality of people is top notch and paying people is simple. Upwork is a one-stop platform to find, hire, and pay expert freelancers across development, data, marketing, operations, and more.

16:03With Business Plus, you can access the top 1 % of talent on Upwork. And with AI-powered shortlisting, you'll get matched to the right freelancer in under six hours. No endless searching required. It's free to sign up and posting a job is easy. Visit Upwork.com right now and post your job for free. That is Upwork.com to connect with top talent ready to help your business grow. That's U-P-W-O-R-K.com. Upwork.com. We kind of headed there. You have a great story. You grew up in India. You left after undergrad without anything ahead of you, essentially an unknown. You walked into this unknown. you took a big leap.

16:51What did you learn from that? You know, so much, right? I mean, I mean, back to your sort of previous question, this, my interactions with model, you know, and it's like, there's so many different things sort of that have come out in terms of sort of my interactions with my own interactions with AI, like, you know, the back and forth, the model surfaced, an angle that I hadn't considered not a solution, an angle or reframing. In my head, this is the only way I can describe it, like a lock turning and a door open. And I think to me, that's been the epiphany. I think on my own journey, as you sort of said, coming into this new culture, new country, the biggest thing that I've learned is that future is very unpredictable.

17:46You are naive, and I see a lot of young people do this. You're naive if you feel that, you know, success or success is a heavy word. Let's say the outcomes you want, let's go with that. The outcomes you want are driven by a formula. Like you do X, then you do Y, and then you do Z. That results in number 45. life. But that's what we have been culturally taught by our parents, by our society. For example, you have to go to school, you have to get good grades, then you have to get a job, then you have to progress at the job, you have to be a good employee, you have to get along with your colleagues, you have to be all of those things, and then you get that.

18:39The truth is that it doesn't quite work that way. I mean, life is very, very unpredictable. Now, you know, there's a debate about whether, you know, things are happening in a deterministic fashion or everything is all just pure random and how much of what happens you actually control yourself. But to me, like whether you're in this camp or that camp, like, you know, whether in the deterministic or undeterministic camp, whether you believe there's free will or no free will, it doesn't really matter either way. So if you really ponder on that questions carefully and sort of say, well, am I in control or can I affect outcomes or things are going to happen?

19:18The bottom line is it doesn't matter. And so just enjoy the journey. I mean, I think that's one thing that I've learned is that, you know, and that's the thing to it is whether you're doing things that are going to happen in an unpredictable form, Um, and, you know, uh, or it's already destined, uh, in both cases, uh, the, you know, just chill out. I mean, just take it easy, um, and go with the flow and, uh, be in the moment. Uh, and it took me like, uh, decades to get to that realization. I wish like, you know, I wish I had that realization many decades ago. And if I did, I would have made different choices or have done things differently or had lived those years differently.

20:10But I think that's the biggest learning on my end. What is something, when you think back, that if you had done differently, it would have made a positive impact on your life? Honestly, I mean, if I had started trusting my own instincts sooner, and I think I mean that in a very specific way. Like, you know, long stretch of my career, I treated intuition as a lesser cousin of analysis. If I couldn't defend a decision on spreadsheet, I'd second guess it. I'd gather more data, run one more model, get one more opinion. And there's a lot of times weeks later, I'd land exactly where my gut had pointed me on day one.

20:50But that's expensive, right? Not just in time, but in conviction. So every time you override your instinct and it turns out that you've been right, you teach yourself to distrust the same signal. you should be actually strengthening. And there's a thread in the book about this. I mean, humans have two modes of thought, like the fast intuitive one and the slow deliberative one. And we built this whole cultural bias towards the slow analytical reasoning as the serious mode. But the intuition isn't magic. It's accumulated experience compressed into a signal. It's your pattern recognition talking.

21:23And if you've been in an arena a while, that signal is worth listening to, even when you can't fully, uh, article it why, at least in the moment. So, I mean, the thing I tell my younger self is, you know, run the analysis, sure do that, right. But treat it as a check on your instinct, not a replacement. Um, and, uh, and if, if there are two are not lining up, uh, or the disagree, get curious. I mean, don't automatically side with the spreadsheet. Um, so I think that's the big learning. Maybe our intuition is our advanced human brain figuring things out in a very short, quick time. But it's hard for us to fathom that we could come up with that answer because subconsciously we're coming up with something that we didn't really think about.

22:06Now there is a quote, everything you want is on the other side of failure. What does that quote mean to you? Yeah, before I get to the quote, I want to just pick up on what you said, because you're right. Intuition, you know, isn't magic, it's compression. I mean, it's your brain running an enormous amount of pattern matching in the background faster than conscious thought can actually track. I mean, there's a concept called embodied cognition, the idea that thinking isn't just in your head. Your gut, your nervous system, your hormones, they're all processing signals constantly. And most of it is actually below awareness.

22:42That's why a gut feeling shows up as a physical sensation, actually, before you can articulate why. Your body knew before you did. it. So now the quote, like everything you want is on the other side of the failure. For me, that lands in a very personal space. I think we've been sold this with that success is a straight line. I mean, you work hard, you're smart, you win. But that's not how any of it actually works. I mean, every meaningful thing I've built, every insight I've had that mattered came out of something breaking first. I mean, a company that didn't work, a decision I got wrong, a moment where I was pretty sure I'd be embarrassed myself beyond recovery.

23:21And so what I've come to believe is that suffering, failure, the hard stuff, it carries information. It tells you what you value. It tells you what needs protection. It tells you where the edge of your understanding actually is. So if you try to numb it out, and I see a lot of people aiming to do that, I mean, if you optimize around it, you will lose the signal. I mean, you lose the thing that was going to make you remarkable. And I write about this in the book as well. It's like, you know, that my own suffering has helped me make who I am. And I suspect a lot of people feel the same way if they're actually honest about it.

24:00I mean, a life without contrast collapses into just numbness. I mean, you need the low points, not because pain is virtuous, but because that's where the shape of what you actually want become visible. So the quote to me means stop treating failure as a detour. It's not a detour. It is the road. That's the road. Reading your book, the argument that I got that your book has is machines need more than intelligence. They need experience and consequence. What's the difference though between a smart machine and one that understands? The cleanest way I can put it is this. A smart machine gives you the right answer.

24:40A machine that understands can tell you why that answer holds, where it breaks, and what would have to be true for it to be wrong. but so i'm i'm borrowing here from uh david deutch uh who has uh influenced a lot of my ideas and i built upon some of his ideas in the book uh who also also i think has the sharpest lens on this i built founder story from a 50 microphone and the most important thing is i didn't do it alone for years i've been using upwork to hire marketing editing branding you name it in fact the editor who cut this very episode and the team behind all of Founder's Story branding found them on Upwork.

25:25The quality of people is top notch and paying people is simple. Upwork is a one-stop platform to find, hire, and pay expert freelancers across development, data, marketing, operations, and more. With Business Plus, you can access the top 1 % of talent on Upwork And with AI-powered shortlisting, you'll get matched to the right freelancer in under six hours. No endless searching required. It's free to sign up and posting a job is easy. Visit Upwork.com right now and post your job for free. That is Upwork.com to connect with top talent ready to help your business grow. That's U-P-W-O-R-K.com. Upwork.com.

26:15he argues that knowledge doesn't grow through prediction. It grows through good explanations, like explanations that are hard to vary, that connect across domains, that survive criticism. And by that standard, most of what we call intelligent AI today is doing something quite different. It's doing extraordinarily sophisticated pattern matching. I mean, it's finding the shape of the answer in the training data. That's not nothing. I mean, it's quite miraculous, actually. But it's not understanding, right? I mean, so the test I use is take a chess engine that can crush any grandmaster alive. Does it understand chess?

26:54It doesn't know it's playing a game. It doesn't feel the weight of a bad move an hour later. It has no relationship to consequence. It just has competence. And competence without a relationship to consequence is exactly what worries me because we're now embedding those systems inside human decisions where the consequences are real. So the second piece, understanding. in the human sense, isn't just explanatory, right? It's accumulated. You become a different person after you've been wrong. I mean, after you've caused harm, after you've watched a patent play out for years. I mean, your judgments is scar tissue in the best sense.

27:31That's what it is, a judgment. And a machine that outputs brilliant answers is identical before and after every interaction. It hasn't learned anything in the way that matters. It's just been queried. So when I talk about artificial general experience, that's really what I'm pointing at. Like, can the system be transformed by what happened to it? Can it carry forward not just data, but the weight of having been through something? And until we build for that, we're going to keep confusing eloquence with wisdom. And those are very different things. You had walked into a company that was facing potential bankruptcy, and then you were able to have a huge turnaround and that company would then go on to have a successful IPO.

28:14What would you say is a secret unlock that you find when you walk into a company like that, that if you make that switch, it changes everything? I mean, I think I'll be honest, right? I mean, most of the stuff is usually worse than what actually is on the paper. you know like for example you know obviously the cash is an issue morale is an issue there's usually a fog over the whole place and everyone knew something is wrong or was wrong but nobody can actually name it out loud you know so I think what I've seen is that and I've seen this pattern now play out in multiple turnarounds is the secret isn't a strategic insight it isn't a clever pivot It isn't, you know, cutting your way to profitability.