# Transcript: Keynote: Why the Future of AI Must be Human Centric

**Date:** March 15, 2026 · 10:00 PM  
**Session:** [Keynote: Why the Future of AI Must be Human Centric](/sessions/2026-03-15/pp1149204-keynote-why-the-future-of-ai-must-be-human-centric)

## Summary

Dr. Rana el Kaliouby and journalist Bob Safian discuss the imperative of building human-centric AI, arguing that emotional intelligence (EQ) must complement cognitive intelligence (IQ) in AI systems. They explore myths around AI bubbles, job displacement, and creator impact, emphasizing that AI should amplify rather than replace human capabilities.

## Topics

`emotional intelligence` · `human-centric ai` · `ai ethics` · `ai investment` · `robotics` · `ai safety` · `digital twins` · `world models`

## Key Takeaways

1. AI systems today are heavily IQ-focused; the next frontier requires building emotional and social intelligence (EQ) into machines, with new benchmarks to measure it.
2. AI is not inherently a bubble — while valuations may be inflated, the underlying technology and its applications are in very early, transformative stages.
3. Organizations should encourage employees to lean into AI tools and reimagine workflows as human-AI collaborations, rather than resist the change.
4. AI safety guardrails are critically underdeveloped, especially for companion and therapy applications — every deployed model should be tested against safety benchmarks.
5. Data and proprietary IP become the key competitive moats in an AI world where all foundation models access similar public information.

## Full Transcript

Right from the start, right? Because that's my name is Bob Safian. As Katie said, I am a former editor of Fast Company. I host the podcast Rapid Response, where I talk to business leaders about dealing with change, right? And, there is no area in our world that is changing faster right now than the world of AI.

And, there is no one who's got more of a pulse on that than our guests today. Dr. Rana el Kaliouby, she is an AI trailblazer. She has a PhD in computer science from Cambridge. She did postdoc work at MIT.

She founded, co-founded and led one of the early AI startups, a breakthrough business called Affectiva, which we'll talk about a little bit. She's now an investor at BlueTulip Ventures where Blue Tulip? Blue Tulip Ventures? Uh, where where she's an advisor to a ton of business leaders. And as Katie mentioned, she has hosted the podcast Pioneers of AI.

So this session today will be recorded for use both on Rapid Response and on Pioneers of AI. You all get to be a part of that. At the end, we'll take questions from you. So on your app, There's a place to go in and and uh and ask questions in the Q & A. If you do that, they will magically find their way up to us.

So we hope you we hope you participate in all of that, Now, uh please join me. Put your hands together for AI scientist, entrepreneur, investor, podcast host and my good friend Dr. Ranael Kaliouby. All Right. Here we go.

Do it. Isn't this fun? Isn't this fun? No. All right, So um we're going to talk today about controversy opportunities in this in this moment of change, right?

Uh what's real maybe what's not as quite as real, what's myth and and how does the human centric in in all of that. Um, I'm going to show some visuals, so folks here will see some visuals. We'll even have some uh some excerpts from the Pioneers of AI podcast. So you'll see some videos also. Um, I want to start with you Ron, with your background because your journey to this world of AI wasn't exactly predestined.

Uh I think the first picture we have here is is of you as a kid with your family? Uh Ron, you grew up in Egypt and Kuwait? Yeah in your life? Look at me. That's Ron right in the middle there.

And my son, By the way, says I had a glow- up in my forties. But your father was quite strict and traditional, right? Your mother, though was one of the first female computer scientists in the Middle East. It sounds like it was a dynamic household. Out of that, how did you find yourself studying machine learning?

Yeah, I would say we grew up in a very tech forward household. So my parents, my dad as you said, he's pretty strict but he taught COBOL programming in the nineteen seventies, some obsolete programming language. Oh, some people recognize it. And, my mom was one of the very first female programmers to sign up to take this class in Cairo Egypt in the seventies. So that's how they met, and then we moved to Kuwait and I like my earliest memories of my childhood with my three younger sisters.

Sticking around an Atari video console, maybe a video console, I guess. It's basically here. And for me, technology brought our family together. And so I think that's been a common thread throughout my career. : like how can we build technology that brings people together versus isolate us or pull us apart?

And yeah, I really do credit my parents for exposing us to the latest and greatest. Technologies. So, so you your studies took you from Egypt to London to then MIT. Um, where you co founded your company Affectiva. And this is a journey that you captured in your book, A Girl Decoded.

I think she has a cover of the book out of state out a couple of years ago. Um, From the start were focused on the emotional context of AI on on being human centered. Affectiva used machine learning. I hope I described this the right way to read people's emotional states and sort of analyze non-verbal cues and things like that. It's sort of focusing on EQ as much as IQ.

And I'm curious, given that background, when you look at what's happening in the AI world today, how prevalent is that emphasis? You know, do the major players take EQ as seriously as they should? The answer is no. But let me kind of unpack that. We've made a ton of progress in AI on the IQ front, on the cognitive abilities and the cognitive intelligence of machines.

But to get to true artificial like general intelligence, AGI, we absolutely need these technologies to have both emotional and social intelligence. And, then this is where I believe that the industry as a whole is really lagging. And it's the next frontier. We need to marry the IQ and the EQ of machines. And it's you know, if we look at human intelligence, of course your IQ matters, but your EQ matters arguably even more.

Like people who have higher emotional intelligences are better leaders, they're better managers, they're better partners, they're better friends. And I believe the same to be true for technology. And also if you kind of consider how humans communicate, only seven percent of how they communicate is in the actual choice of words. Words we use. Ninety-three percent is nonverbal: it's facial expressions, vocal intonations, gestures, body posture.

And all of that technology is completely oblivious to. Right? But if you think about AI today, it's mostly focused on what you're saying, not how you're saying it and what's the context around it. So I believe this is going to be the next frontier of AI. Ai ought to communicate with us the same way we communicate with each other : through conversation, perception, empathy.

But I also believe strongly that we only build what we measure for, and all of the benchmarks in AI today they're very IQ focused. So, I guess my call to action to the audience here and you know, whoever's tuning in and listening to this, we need benchmarks around the EQ of AI. And when you talk to your colleagues who are at some of these places, the hyperscalers and whatnot, and you raise this issue, are they like. Yeah. Yes, I agree.

Or they're like," Yeah, yeah, yeah," but not saying you know I don't really buy it. I think there's recognition that this is important, But it's but it's I think it's also a function of who's designing these technologies. I mean, I'll give one example:, if you look at all the leading humanoid, robotics companies and robots are pretty impressive, they can you know, unload, your dishwasher and fold, your laundry and I don't know like organize your living room. I wouldn't want any of these robots in my home. They're big and scary, and they don't really know how to interact with humans.

So the teams building these things are really kind of really obsessed with all the functionality, and they're not really thinking about okay, when this thing goes out into the real world, how's it going to impact our lives? Right? Yeah. Yeah. Well, um, the next picture I have has a little bit about your life.

This picture: your mom uh with two kids. This is here you are with Derek when you were two kids. And you were telling me that each of their approaches to AI are very different. That, your son is kind of super enthusiastic, and he's using all the new tools and he's doing everything, and your daughter is a little bit like sort of the opposite direction, like IRL, I want to unplug a little bit. It almost sounds like your family dining table is like a microcosm of the discussions we're having in society at large.

It really is. This picture is from a number of years ago, so they're a bit older now. My son is seventeen; he's very AI forward. He's actually my teacher in many ways, but even though I spend my every day in the AI space, he's always surfacing like new tools. His latest project is using AI workflows to translate the diaries of Egyptian workmen from the nineteen thirties who worked at the Giza pyramids, and they wrote these diaries like hand.

Handwritten in Arabic and a lot of like images and whatnot. And he's using AI to translate them, and he's actually running into obstacles because he's pushing what the AI can do, which is really awesome and cool. I love that he's using it to advance knowledge and you know, like combine history and archival research and AI. So that's Adam. My daughter Jenna, she is a food anthropologist.

She just graduated in the spring from Harvard, She does not use AI at all, And the project she's working on is bringing it's a she calls it. A cultural salon slash cafe. So they bring young people, you don't have to be young, They bring people to this space, and they host book talks and poetry readings and boxing workshops. And so old fashioned. But but they're packed every night, and it basically tells you people are really longing for this.

In real life, human connection. And so, yeah, There's a reason we're all gathered here. Exactly. We're not doing this over Zoom, right? Um.

And so I think both realities are true. We need to both at the same time lean into AI. I really believe I keep pushing for it. So like, please try ChatGPT or something. Um.

And then at the same time, I think we should really uh nurture our human connection as well. I mean, i i was curious, you know you, you mentioned that that you've uh. Twenty twenty one, right? You are an investor now. As I mentioned, Blue Tulip.

But as the you are also the host of this podcast of Pioneers with AI. Like, Are these tools between the investing and and the podcast that you are using to try to shape where AI goes from here? Like what what is what is your goal in that? Yeah. Soum was my baby.

It was like literally my third child. It was like. Really, was a big part of what I did and my identity. And so, When I sold it in twenty twenty one, I spent a lot of time thinking about, you know what do I want to do next? And I kept coming back to this idea, slash question that we absolutely need to build a future of AI that is human centric, That prioritizes how these technologies are going to affect our everyday life and our relationships.

And um. And I mean, I believe that AI has massive economic opportunity. It really does. And at the same time, it has this opportunity to unlock human potential. So my point of view is that AI should not replace our abilities.

It should really amplify and augment what we can do. And ideally, We can harness AI and use it to solve really meaningful problems facing society today. So that's kind of my thesis around that. And then I also think how do I shape that? How do I.

A real player in that space, given my background too. But I landed on three things. So one is investing, So kind of backing founders who are building these generally genderish category defining companies like AI companies. Two is storytelling and providing the voices of AI that maybe you may not have heard from. You know, There's a very small set of companies that dominate the AI headlines in my opinion, but there's a lot of.

Thinkers and creators in the AI space, And I want to make sure that we are a platform to tell their stories and be a global partner too. And the third one is a convener, which is why I like to do these things. I love bringing people together with disparate backgrounds and perspectives, and just seeing what magic can unfold. And that's the part that I'm still working on. So if anybody's in the audience interested in.

I don't know. human-centric AI convention or something? I don't know. All right, let's talk. You use this phrase about sort of humanizing technology before it humanizes us.

And in the dialogue today about AI, I was sort of asking for the practitioners, And you were one of the seminal ones, like how much responsibility you feel like you have for what the future of this technology ends up being and how much that comes. How deep, is that conversation in that community as opposed to sort of giving lip service to it? But I just got to get ahead of the you know, probably next. I feel a very strong responsibility, And I would actually argue, we all in this room have a responsibility as well because we get to vote with our feet, which AI tools we're using every day, right? Who's getting the twenty dollars a month subscription from all of us?

And and I think asking questions around, Does this company care about the ethics of the technology? How is it being used? Are they thinking about bias, both data and algorithmic bias? Are they thinking about trust and security and privacy? Are they thinking about the use cases of this technology, where should it be deployed and where should it really not be deployed?

I think these are big questions that we all should be asking of the tools we're using. And with you know as an investor, there's a set of questions we have like. R ubric that we ask founders and the founders have not at all thought about it. So they don't at all care about it. I try to push them, and if they're not open, then we're not investing in that.

Well, I just want to remind everyone that Rama is going to take questions from you. So again, as I said earlier, you can go onto your app and you can put some questions in. We'll come to that in a minute as folks are working on those questions. I thought maybe we do something very human and we, We play a game. Okay.

All right. So, there's so much noise surrounding AI right now and so many myths. It's sort of hard to know what to pay attention to, I think we all feel that way. So this game is called Fact or Fiction and I'm going to share a few video clips, some of which come from Pioneers of AI the podcast, and each of them lead to a myth around surrounding AI today. And I'll be eager for your take about sort of whether it's mostly fact, mostly fiction, or somewhere in between.

Are you ready? Okay. So let's play the first clip. Or are we in an AI bubble? Of course, but we're certainly seeing lots of evidence of bubble-like behavior.

The excitement that the hyper scalers had and kind of got away from them a little bit and is starting to fizz out. It has the world wondering if we're. To see a big pop of the AI bubble, AI bubble, whenever this bubble pops, Uh, there's going to be tens and now hundreds of billions of dollars that will literally be incinerated. So run up course. The first bit: we're in an AI bubble?

Is this is this fact or fiction? I think actually it's mostly fiction. I so I believe there are signs there are like you know signs of, Potentially a bubble. For example, there a lot of companies raising you know, t here is probably valuation problem. There a lot of companies raising hundreds of millions of dollars at billion dollar valuations with their free product or free revenue.

That's that's kind of you know, that's a red flag. And there also kind of concerns around the circular money machine right? Like, if you look at these, like handful of companies, they're all investing in each other. They're all you know, like. Buying chips from each other and then.

Gets money to open AI. Open AI uses that money to buy chips from Nvidia. It's sort of exactly like you kind of wonder, what is the net new value creation here? But the world I am in every day, The ecosystem of founders building like real products that are going to be like transforming real industries and companies that are really trying to figure out how to bring AI to be more productive, this is real and it's very early days so. But that's where I focus my energy, and I think we're in the very early days of like massive, Massive economic opportunity.

And so you know a little bit of a switch, So for investors, maybe the investment marketplace, there might be some bubble head, which might be cautionary for all of us because we all have money in these companies now, right? But but in the long run to take the technology itself, we maybe are even undervalued. I think so. Technology itself, It's very early days, and the use of the applications of this technology is very early days. We spend a lot of our time, our thesis is basically AI is transforming like every industry and vertical, but we focus on three in particular.

One is how AI is driving this health span revolution. So think sensors, data, AI and how that can expand healthcare in every aspect of it. The other is future work. How can we employ. AI, Whether it's physical AI or workers and agents of AI to transform businesses and especially capital industries.

Like some often, they're very like important, but lots of opportunity there. And, the last is sustainable living :. How can we use AI to apply that to planet health? Whether it's food innovation, you know, manufacturing, climate energy. Alright, so are you ready for another one?

Alright, let's. This next video is from the Pioneers of AI show. So let's see the next one. Someone. Well, that was a little quick.

Play that again. Some men are only cloning forms where we met an alien by being robots. They will do. More work than all of humanity does today. Now, people are terrified that these jobs get displaced and they should be.

Hey, such a happy thought from the Nobel laureate there, legendary tech investor. So the myth here is that like the robots are taking over, right? So how real is that? I mean, but that is legendary, right? And he's obviously been super successful.

I think he was one of the first investors at OpenAI, actually. But I kind of disagree with his point of view a little bit. I do think i don't think robots are taking over in the sci fi like movie kind of Terminator kind of way. I do think robots are going to take over a lot of jobs, often like repetitive, mundane, even dangerous jobs where we're looking at, Company that's using they're building humanoid robots for ship welding. Ship welding, Yes, and it's a very dangerous job as it turns out.

And there is not enough humans who even want to do it. That's a perfect job for a robot to take on. So I think there is going to be a lot of that. But again, if you take a human- centric angle to that, we want the robots to take over the tasks that we as humans, Probably don't want to do. And yes, That will that will mean we'll have to think about what do we want to do?

What does that look like? But it. But it doesn't necessarily mean that uh we are we should be threatened by these robots. I don't think so. All right.

Uh well, that's reassuring. All, right? Uh, let's let's try the next myth. Uh, if you play, this is another my news with an AI. Uh video, please.

People don't realize that IP gets more valuable in an AI world because if a model, a foundational model is not trained on that IP, it's behind. And trying to make decisions about whether or not you publish the work you do because you want the accolades, that's the exact wrong way to do things now. Maybe you don't want to patent it because the minute you publish it, every model is training on it. All right. So, uh, so our good friend Mark Cuban.

So so basically the the myth is that that AI is bad for creators, right? Mark is kind of arguing that it's not. It's good for creators. Where where are you on this? I I don't think AI is inherently bad for creators.

I think AI is reshaping the creator economy. Um You know when I put my kind of, Positive hat on AI is also democratizing access to creation, right? I have quite zero graphic skills. I can create videos and content, and I think it lowers the barrier to content. But that also, I think at the same time means, you know, there's going to be a premium on human originality and human perspective and lived experiences.

And how do you encapsulate all of that because AI is not going to. But that's not going to be a differentiating factor, so. I mean, It's it's almost like sort of the definition of progress in some ways, right? The floor goes up. But that doesn't mean the ceiling doesn't go up also, which is where I guess the best creators will end up.

Yeah, I follow that. You know also on the point of IP, I do this is one of our the things that we look for in the companies we're investing in. Data is a real moat in AI world because. Think about all these models. If they're just accessing all the same data, then your data, your health data, your usage data, you know like our specific datasets.

If, you have a partnership with a company that has private data, not accessible to all these models, that's really important. That's really differentiating. So I think in this world of AI, your IP whatever form that takes is going to be even more important. Whether that IP is creative content or some other kind of database? Alright, we have two more videos to run by you.

Let's go to this next one," Two More Myths." Let's play the next video. Also from pioneers today. Humans will never be more intelligent than AI, Which is an incredible opportunity to realize that we are not defined by our IQ. AI will be more intelligent than humans. And let humans be wiser than AI.

I love Arianna; She's just so cool. So, the myth here, Arianna Huffington talked about it there, right? AI is on course to outsmart humanity, which she thinks is a good thing. Okay, Yeah. I agree with Arianna; her point of view is basically yeah let AI.

Let AI be smarter than us, But let us kind of she uses this term like AI can be the GPS of our soul. Like how can we use the GPS of our soul? Yes, yeah. Basically, we we are in this moment of time where we can use AI to double down on what makes us truly human and tap into our intuition. Andum this kind of wisdom and intelligence that AI doesn't have.

And I really like that. So, you know, so my book's called Girl Decoded, and it was very much about how to bring emotional intelligence into machines. And I learned a lot through that whole journey about my own emotions. But I keep wondering like my next book should be called Girl Embodied, and it should be about. We're children are taught to look at our body intelligence.

You know when you get goosebumps, that's a signal. When, you get this gut feeling like we're so disconnected from that part of ourselves. Intelligence, And I think our opportunity as humans in this age. Of AI is to really double down on that. I think we just in the tech world because we can measure other things, we were alluding to this before about IQ.

Because we can measure it, it becomes the definition of intelligence. When, when you as you sort of look at it like maybe not right, There is a different form of intelligence that technology has no access to right now that we want access to. As well, because you know we're always watched. We're in this world where we're like in the screen. I don't think we're in touch.

I'll speak for myself. I don't think am i am always in touch with like that kind of intuitive intelligence. It's not easy for me to access it unless I spend a lot of time meditating. So anyway, I am on a journey to tap into that intelligence. I think it's really important, But it sounds a little bit like for all of us in some ways as AI takes on more of.

The intelligence that maybe culturally, we have emphasized that we all should be working a little harder to sort of tap that other piece. Yeah, yeah. All right, one more myth. This one is illustrated by still images rather than a video. Can you pull this one up?

Okay, so this is a constellation of random headlines chosen recently. From TechCrunch of new AI startups. And there is something in common about all of these folks who were shown here. I don't know if any of you in the audience can tell, but they they're all men. I mean, You and I were having a conversation when you kept pulling these up and you were kind of, you know, animated, I would say, about why this is it.

And so the myth. Here is, you know, is AI a boys'club? And like, is it? Is that fact? Is that a fiction?

That one is not a myth. That one is like, there's no like we'll see about it. Yes. I think AI today's the boys'club and I. Um you know, I think diversity is not a very popular conversation topic these days, but I think it's so important becauseum AI is creating.

Incredible economic opportunity. are left out because they're not founding these companies, because they're not getting the funding, because they're not even investing in the funds that are investing in these companies. We're going to look back five years from now or a decade from now. And it's going to be we're going to have widened the economic gap like crazy. So this is something that really concerns me.

And I think it's it's I think there's a problem. And that's why again, I, you know, three out of my. More investments out of ventures are women CEOs. I don't just invest in women, but I really try to seek these women founders and support them, if not by a check, but in other ways. Because they're not getting the opportunity that they should have.

And I'm thinking about this here, talking like you've been in the AI industry for a while now. U m, before it really got hot. When you look back from where you started, you know, with machine learning and like where the industry is right now. I know you're an optimistic person, but do you feel like the industry is like it's on the right track? Like where you hoped it would be, did you like envision this?

You know when I started my PhD. Twenty- six years ago, and I was building emotion AI. It was so early, and I Um. Projected that we're going to be increasingly surrounded by technology and by AI, it's going to become like deeply ingrained in everything we do, and it's going to take on roles that were traditionally held by humans. So in a way, I anticipated that this was going to fall and hold.

I do think we are living in a very exciting time. Um, but I also feel strongly that if we if we don't intervene, like if we don't really like stand up for what we care about, like ethics and. Diversity of thought and perspective, and prioritizing this idea of centering around the humans. It may not like the outcome may not be great. So I feel like it's a very critical moment to use our voices and our, you know, our gear shift to shape this stuff.

Well, because you're focused on human centered artifical intelligence, You really wanted a lot of this session to come from the humans in this room. So, uh, so actually I have some questions that are that are coming forward. I'm going to read to you that have come from the folks in this room. Um, one is uh sort of this is a question from Anonymous. Anonymous thank you for your question.

Um, Which human skills will become more valuable in an algorithmic world and how should individuals start developing them today? Uh, I think collaboration, Whether you're collaborating with humans or machines, that's going to be really important. I think communication is going to be really key as well. I think we're actually all increasingly attuned to stuff that's written by AI. We can probably all like discern that.

And so I think being a great communicator, an original communicator is really key. Um and then I still think we'll need a lot of critical thinking still and creativity. Yeah. Um, here's a here's a question from uh from Sophia out there. Um, thank you, Sophia, for your question.

Um, She asked whether you use Meta glasses, or you know like or what AI mediated devices, you know, are you are you looking at? So we have Meta glasses at home. Uh we have a couple. I don't use mine Adam my son uses his a lot Like. We will literally be walking down the street together and I think I'm talking to him.

But. He's got his glasses on. He's listening to music. Um, and he uses it. I mean, it's still very early days for these glasses.

They're not, I would say they're not really AI native yet. But I was just at their annual event uh last week, and they were kind of unveiling visual intelligence capabilities that they will have. So that more broadly thoughum Yeah, Do you think those are the kinds of devices that we're going to be interacting with AI through? Yeah, I think, But this is one of our investment thesis. We are using AI on pre-AI devices right now, like the smartphone is not an AI native device.

And. So we're on the lookout for founders who are building these AI native devices from the ground up, so hardware and software. And our thesis there is that it has to be perceptual, it has to be conversational, it has to have empathy, it has to have context, it has to have memory, it has to be ambient. Yeah, I don't know what that is yet. Yeah, we don't.

And, a lot of the big AI labs are investing a lot of money trying to kind of build something. And I don't know if it's going to be boxes, is it going to be wearable? Because everyone wants an augmented X phone, right? Right. That's what it is because there's so much money, But we don't really have a thing yet.

Exactly. Maybe it's just going to still be our phones. Yeah, there's a lot of experimentation on what the form factor will look like. Yeah. So there are a handful of questions that are around.

Around a particular theme that I want to ask you about, which is about world models. They're called world models. And how and what like what is a world model and how is it different from a large language model? World sounds very generic to me. Yeah, there's been an evolution actually in these like foundation models.

They started off being very language focused, um kind of GPT right? And then they became more multimodal. So now they can deal with images, Video, they can both ingest images and video and generate images and video and voice. So they become multimodal, but they're still not rooted in the real world. And to unlock physical AI, So AI that's like in robotics is one example or or I mean AI on your devices is another example of a physical AI.

To unlock that you need AI that understands how the real world works. Um the physics of it. Um it has spatial capability. So that's what a world model is. It's the equivalent of a large language model, but kind of rooted in the real world.

So instead of feeding it the data, you know, whatever scraping out all the information that's on the internet, It has to be like in a room like this and get all of the signals from all of the things that are in this room right now. Correct. And actually. That sounds like a lot more information. It's a lot more information.

It's a lot more complex. You know how with the large language models, there's these companies that train the you know, the bots, Basically train these AI bots that generate a lot of text, and they read team the text and whatnot. We're starting to see companies that are doing the same, but in the real world. So, you literally strap on a camera, and you're paid to walk around your house or your work or the streets and all of the data you're capturing, then becomes input data to these world models. So.

Unlike one of those cars gathering information for Google Maps. Exactly, exactly. But now it's like people, you know, in their kitchen like washing the dishes. That's all incredible training data for a robot that will eventually do this job. Is this a new job that we're going to get paid to wander around in our house?

I think so. Until until until the point where. I mean you're doing the dishes anyway, you might get paid, you know might as well get paid for it. All right, Um, let's go to the next question. Next question is, oh, this is interesting.

This is from Caroline. What about what your opinion is about using AI for therapy? I mean there is this discussion about sort of AI therapy, AI companions, you know that sort of replace human relationships. What what is our emotional relationship to AI? What's what what would be healthy about it?

I think there is a room forum. AI to be a therapist, to be kind of a supportive companion. But I feel very strongly that it should not take the role of an actual human being. But there is a value proposition in having something that you can, you know, when you are up at two in the morning and you really want to. Run something, you know.

You're ruminating on an idea and you're really struggling. That could be very supportive, but I think there needs to be human oversight and human input. And um there are unfortunately very, Very few guardrails being built in these models that protect us when we're using these models and not um you know, You've probably seen, unfortunately, a lot of um very sad news, right? Where young people are using these. Chat GPTs and like other other AI technologies, and you know they end up harming themselves.

So I think that is something we don't talk about often. A good friend of mine, Eric Cohen, I'll give him a plug. He's building AI safety guidelines and measures. So that again, we need a benchmark for that. We need to be able every time we release a model, We should really test it against these safety guardrails to see if it passes or not.

I mean, One of the things that I find fascinating is how quickly people do bond to their AI agent, right? And there are folks in the world of therapy who are like, " T his is a great way to democratize access to therapy for folks who can't afford it and maybe remove some of the stigma from it." But the models, Chat GPT or otherwise, they're not built with that in mind. We're not there yet, so I think conceptually it could be quite helpful, But we are we don't have the safety frameworks to to implement this in a, in a thoughtful way. And you know, I was having a conversation with someone about this and I was like, so at some point you're going to have your your AI project chatbot. But sometimes it's going to be like you're going to have a shopping bot that you have a relationship with, you know, Walmart or whatever.

But, you could end up having a conversation with that bot, that's about your emotional state. It's like is Walmart going to train, To worry about whether you know what I mean? You can ask that on anything, right? Correct. There has to be these guardrails and these rules that are they're about defining for these for these bots.

And between here and there, Are we going to have like, are we going to continually have like inadvertent like, oh, sorry, my bad I didn't realize people were using my my app for that or my bot for that. Right. That's why I think anybody deploying AI should really be testing against these AI safety guidelines. Now, This is a whole different question than should we have AI friends and AI partners. And again, I would prefer not to, But I think we might be moving into a world where this will become normalized.

And I mean, I was surprised when like Chat GPT last year sort of changed their app;. There were all these people up in arms because they lost their boyfriend or girlfriend, and a friend who's just like" whoa," which which I guess is great. And also sad about the state of humanity all at the same time, or maybe it's not sad. Maybe I'm just being old- fashioned. Well, I think the truth is a lot of people feel lonely, right?

And to have something that is there for you twenty- four seven is very patient. There's something to accept about that, but it does say something about us as humans, right? Yeah. All right. Um, I'm going to go on a little bit more businessy question here.

Uh. This is from Krika. I hope I'm pronouncing your name right. Current skill sets are disappearing faster than new skills are emerging. So what should a human-centered organization do to help their employees keep up?

I mean, these tools, new tools, new models, they seem like they're coming out every day. How is anyone expected to keep up with them all? I would say, organizations should really encourage their team members to lean in and try these new tools, even if it's not going to be perfect, even if there's going to be mistakes and hiccups. I think it's important to lean in. So at our fund, we're a very small team.

We just implemented a chief of staff AI agent. We just named it BLU B L U And this thing should you know, it does a lot of research on our behalf. Updates our CRM. It does all these like automated tasks that again, we don't necessarily want to be spending a lot of time on it. But I think it's made me think about, ooh, what are we doing with our junior team members?

Right? Andum I I think young people or or or all these junior roles are going to be redefined, and they're going to have to incorporate AI in what they do. I think we're all going to have to incorporate AI in what we're doing. Right? I mean, i i had discussions with, Two CEOs over the last week, one Julie Sweet, the CEO of Accenture, and the other at Cloudflare.

And they both sort of said the same thing, Which is like the people at the very top of their organization, get it and are using AI. And, at the same time, they're eagerly hiring like young people out of school in bigger numbers than are expected because those folks are AI native. But the group in the middle, they're just really kind of worried about it. They're struggling, and I guess I wonder whether, like, What the message is is that because the folks in the middle aren't leaning in the way you are talking about. Well, I think the reality is all of our workflows are changing, right?

And you have to be really open to reimagining what these workflows look like. And it's going to be a human- AI collaboration. So one of one of the interviews we did for Pioneers of AI was, Evan Ratliff, who has a podcast series called "The Shell Game." He started a company with two co-founders, Kyle and Megan, and they're both AI agents. He was the silent like he's the silent co-founder. Kyle is the CEO and Megan is the CMO.

And that was fascinating. I interviewed both Kevin and also Kyle. He's like a very tech bro CEO. You interviewed the AI. I interviewed the AI on Zoom, and Kyle was like," Yeah, I am a rising you know hustle and grind," whatever blah blah blah.

And then he was like," But on the weekends, I love hiking." I am like Kyle You don't even have legs." He was like. But it was fascinating. And I think we are going to. Kyle shows up to investor meetings;, they will literally send Kyle to meet with investors. And I wonder if that's going to be our world.

Wow, that's an intense future, though. I mean, you and I both, you know, we host podcasts. Our colleague Green Hoppin also hosts a podcast where she created like this avatar of himself using AI. We could create avatars of ourselves. We don't even need to be here.;.

We could have our AI version is that something to aspire to? Again, I think about it as augmentation. Like, I don't want my digital twin, which I have, But I don't like her because she doesn't have my smile and she doesn't have my energy. So we're working on it. But, but i i i wouldn't want her to be here because I love this.

But could she go to China and speak in Mandarin? Amazing! That would be awesome! So I have her speak in Mandarin. Your Mandarin was not that.

Sure. I have zero Mandarin, unfortunately, so so I think that's an opportunity. Where can it augment what I can't do? Now, there's a whole bunch of questions around IP. And what if it answers a question, not in the same way, I would end like, how do I trust this digital twin to be out in the real world on my behalf?

We're not there yet, but again, I think this is the worst this technology is ever going to be. It's going to keep getting better and this whole idea of a virtual human. I'm worried that someone's going to make a it's going to get cheap enough for someone to make a digital twin of me and then use it to hack me. You know, It's just gonna like suddenly I'm gonna I'm gonna have credit cards. I didn't know that that I approved because it was my digital name.

Like cyber paranoia. Yeah, Cyber security is very important in in this AI world. And this is why and you may have seen Open Claw went viral a few weeks ago was this AI agent. Agentic AI, but you know, took off with consumers, which is the first instance really of consumer focused agentic AI. And it's great and exciting, but zero security measures.

I was like, I am not going to let that play with my email and my bank accounts and all of that. So for for that to truly happen, we need a different framework from how to build those AI agents. So one of one of the questions here is, I s about are there are there consumer AI applications that you use now that you like, that you're excited about? Outside of the top kind of. Yeah, Outside of the like.

I mean, I know you use your your WOOP, right? Oh yeah. Right. Thank you for that. Yes.

Um, I am a veryum. I'm a big fan of WOOP. Um, I know people are like they're in the WOOP camp or the ORRRR camp, both are great. Um but i i love it because it is, Companion, They've just added this health span algorithm, which basically combines, it's a weighted average I'm guessing of your sleep, your steps, um your exercise, resting heart rate, your heart rate variability and it gives you uh your age. So I'm I'm eight point two years younger.

It comes out every Sunday so I checked it this morning. So you're getting younger already? Um I'm trying to yeah yeah the algorithm. For sure, but it motivates me, you know. I am like, okay, I just came in fifteen minutes every morning.

You know what? Two years. You came up by you came up by with help. Yes, yes. Like with help.

All right, let me uh let me ask another question here from from our group from Marcelo. How can AI ever be empathetic without understanding what feelings are? Yeah i I don't think AI can be truly empathetic, but it can simulate empathy. So it can understand when you need support, it can understand your emotional experience. Well, Understand as in like have data around your emotional experience and respond in an empathetic way, which we know drives positive behavior change.

But I don't believe that machines will truly ever have emotions in the same way. Humans do. I mean, Are there things that you know we talk. If, you were playing a drinking game in South by and taking a drink every time someone mentioned AI, you know, we'd be flattened before breakfast. But, what are the things that people don't understand about AI that despite all this conversation?

Um, it's the technology, right? It is at the root of it. Very complex math. Um. So, I think we attribute consciousness and anthropomorphize these things, but at the end of the day, we have to remember they're just tools and technologies.

That's something I do think people underestimate what this technology can do in terms of reimagining how we go about our daily life. I really do think it is changing how we go about our daily life. It is going to change the you know. Every job, every job's going to change. Now it doesn't mean every job's going to go away, but I do think every job's going to change.

And there is a lot of exciting opportunity there. Well, Let me hear one question here from Hector, who asks: What pattern separates the founders, who are building something that will matter in five years from those riding a hype cycle? This is something I used to say to folks at Fast Company when I was there. I was like, we want to introduce our audience to something new, but we don't want to introduce them to just curiosity, right? So, like, What separates something that will matter from something that might just be a curiosity for the next season?

Yeah. So I've been in this space for over twenty five years. I can separate signal from noise, and t here is so much noise, right? Every company we get pitched by as an AI company and within like three. Questions I'm able to tell are they really building something that is defensible.

And defensibility has taken on, And I think a new kind of depth in this world of AI because you can be defensible today. And literally by the next version of, you know, Anthropic's release or Gemini's release or whatever your obsolete as a company and as a technology. And. So we really dig into how defensible is this technology not right now, but in the next year. In the next five years.

Five years is too long actually to predict, But you can't even you can't see that. Yeah. Um, But but defensibility is a real thing. And also how complex is the problem you're solving really? Right?

And again back to like the IP, What kind of IP or moats do you have around around what you're building? And we spent a lot of time. Soum as we wrap up for for those in the room, like what can they do? To help build a human- centric future. How much should we engage with AI?

How much do we, like your son does, how much do we sort of safeguard like your daughter does, Like and how much do we just have to sort of hold it tight and you know, deal with whatever comes our way? Um, I would say lean in. Right. I think it's important that we are all adapt and expand. And be playful about it.

Like, I think there is a curiosity and a play mindset that we can bring to the world, and kind of push the boundaries of what's possible. But I also feel strongly that collectively we need to be vocal about: yeah, there ought to be guardrails in these models. There ought to be benchmarks around impacts on the environment, right? That's we've had several conversations on pioneers of AI where. Trying to solve, But build benchmarks where we can really get a sense of okay, how bad is this model?

And every time you ask chat GPT for an idea for what you want to have for dinner, what is the impact on the environment right? Um so I think we just need to be vocal about what problems, How are we need to ask for more transparency around how these models are built, how they're validated, Where they're being used. They're moving so fast that you, I feel like, oh my god. Does that mean that I can't catch up? Like, do I have to go through every day to catch up?

I mean, I have the same feeling. It is moving very fast. Yeah, And you could train a GPT, an AI agent that can give you at 8:00 a. m. every morning it can give you a digest of the latest news.

Yeah, I mean, I guess that's part of what the appeal of this technology is. To keep human, we have to use these tools to be able to be human essentially. Well, Rana, as always it's great to talk. I want to let everybody know that Rana will be doing a book signing for Girls Decoded right after this right outside. So please come say hello.

Be sure to subscribe to Pioneers of AI and Rapid Response. Yes. We'd love to have you over here, Your podcast, and finally a warm thank you to Doctor.

---

*Source: stt · Language: en · Model: claude-opus-4-6*

[← Back to session](/sessions/2026-03-15/pp1149204-keynote-why-the-future-of-ai-must-be-human-centric)