On this episode of the AI For All Podcast, Michelle Zhou, Co-Founder and CEO of Juji, joins Ryan Chacon and Neil Sahota to discuss cognitive AI. They talk about the benefits of cognitive AI, artificial empathy, robots vs chatbots, psychographics in AI, why enterprises should adopt cognitive AI, no-code AI, AI as an assistant, and the future of AI.

About Michelle Zhou

Dr. Michelle Zhou is a co-founder and CEO of Juji. Prior to starting Juji, Michelle led the User Systems and Experience Research (USER) group at IBM Research – Almaden and then the IBM Watson Group. Michelle's expertise is in the interdisciplinary area of intelligent user interaction (IUI), including conversational AI systems and personality analytics. She is an inventor of the IBM Watson Personality Insights and has led the research and development of at least a dozen products in her areas of expertise. Michelle has published over 100 peer-reviewed, refereed scientific articles and 45+ patents. Michelle is the Editor-in-Chief of ACM Transactions on Interactive Intelligent Systems (TiiS) and an Associate Editor of ACM Transactions on Intelligent Systems and Technology (TIST). She received a PhD in Computer Science from Columbia University and is an ACM Distinguished Scientist.

Interested in connecting with Michelle? Reach out on LinkedIn!

About Juji

Juji is an AI company located in Silicon Valley specializing in building cognitive conversational AI technologies and solutions that enable the creation and adoption of empathic and empathetic AI agents. Their goal is to democratize AI and enable every organization, with or without AI or IT resources, to rapidly generate, customize, and operate AI beings that help scale out their high-touch, high-stakes services with a human touch.

Key Questions and Topics from This Episode:

(01:15) Introduction to Michelle Zhou and Juji
(01:48) What is cognitive AI?
(03:00) Benefits of cognitive AI
(05:03) Artificial empathy
(16:28) Robots vs chatbots
(19:26) Psychographics in AI
(27:13) Why should enterprises adopt cognitive AI?
(31:31) No-code AI
(35:19) AI as an assistant
(37:13) Future of AI
(40:13) Learn more about Juji

Transcript

- [Ryan] Hello everyone. Welcome to another episode of the AI For All Podcast. I'm Ryan Chacon and with me today is my cohost, Neil Sahota, the AI Advisor to the UN and the Godfather of AI for Good. Neil, great to have you here.

- [Neil] Hey Ryan, I think we got for our audience another great show they can't refuse.

- [Ryan] I like that. That's funny. We also have our producer Nikolai who is going to jump in with some questions throughout the conversation as well.

- [Nikolai] Hello.

- [Ryan] Today's episode, talking about cognitive AI assistance. While typical AI assistance can automate routine tasks like we've talked about in the past, there is a new generation of AI assistance powered by cognitive intelligence that can have better empathy, personalization, and psychological insight.

Stuff we're going to dive into here, and I'm very excited to talk about. So to discuss this, we have Michelle Zhou, the CEO and co-founder of Juji, a company that is specializing in building cognitive conversational AI technologies and solutions that is really helping organizations create and adopt empathetic AI agents. Their goal is to really democratize AI, enable these organizations to rapidly generate, customize, and operate AI chatbots to help them scale their high touch, high stake services with a human touch, which is very interesting just in the way that they have explained themselves.

Michelle, it is great to have you on the podcast.

- [Michelle] Thank you, Ryan. Thank you for having me.

- [Ryan] So let's start this conversation with cognitive AI. What makes cognitive AI different than other AI, let's say?

- [Michelle] So, cognitive AI, it's actually a branch of AI. What it means is the focus is to power machines with human cognitive skills. For example, active listening, it's one of the human effective communication skills to make the human communications more empathetic. More effective. And another human skill, especially for some very kind of how do you say the people who are very good at it is that we call the people reading skills.

So if you can think about the psychologist, counselors, and some maybe even psychiatrists, so they are very good at reading people, really understanding what people are thinking about, what their characteristics, what their unique qualities, right? So this is a type of skills we are focusing on to teach machines with those advanced human cognitive skills.

- [Ryan] How would you explain the benefits of that cognitive intelligence for machines to have that are going to be interacting with humans potentially?

- [Michelle] So thinking about this way, most of us, right, we couldn't possibly afford to have our own career counselors. So how many of us can spend hundreds, thousands of dollars to hire a personal career counselor? Very few of them have our, for example, financial advisors, right? And a few of them will have our own potentially personal health caretakers.

So if the machines are empowered with, are powered with this human advanced cognitive skills, they could really understand us as a very unique individual. Use that kind of insights, it can help us make decisions. So that's why we call it a cognitive AI assistant. So assistants, which means it is, they're never meant to replace humans, but instead, they can actually augment human wisdom to help individuals at scale.

So thinking about it right now, actually, things already happening, thinking about the university. So the some of the students, maybe the first generation college goers in their whole family, a first generation immigrants. So they may not know which learning program, which major might be the best for them based on their unique characteristics.

But they cannot really afford to hire a career counselor or college counselor. But the AI, interacting with the cognitive AI, AI can really understand their passions, their interests, their strengths, and by recommending programs that are best suitable for their unique talents and skills. That's one example in education.

You can think about many examples in healthcare, in finance, in talent, in career development.

- [Ryan] I'd love it if you could talk about how or I guess break down the role this is potentially playing in that health side of things. I think it's really interesting when we think about, and I think about personally when I interact with my insurance company or my doctor in some way, and maybe I'm not able to talk to my doctor, but I'm able to talk to a chatbot, and I not only is privacy and stuff important but for them understanding me, them listening to me, them being relatively empathetic in my, in that conversation is important, but how do we, what is the role that this, that the, this technology is able to play in those types of interactions?

- [Michelle] Very good question again. So I have a word actually to describe how this happened and what requires to make it happen, right? So I called it almost like AI ERA, E R A, right? E, which means empathy. Empathy, which means it is, doesn't matter who you talk to. Let's say you talk to a person, your best friend, or your family members or your colleagues.

So people you feel like they care about you. Normally means that is the really understand you, right? They can empathize with what you are in a situation you are in. So that's level of empathy the machines must have and must demonstrate that level of empathy during this human AI interaction. That's called E.

Responsible. Responsible means that is because AI is not perfect. So during the interaction, AI may be interrupted by users for various reasons, right? Or maybe the users do not understand what AI is suggesting, explaining. So there's always interruptions, what we call the exceptions. But the AI needs to be like a very good worker, needs to be persistent, needs to be responsible to finish the task it started with.

That's what the R is, responsible. And of course, because cognitive AI really understands individuals, we don't want the cognitive AI to take advantage of people. Because I knew somebody, if the AI knew somebody who can be easily addicted to the playing games, then I keep selling you the games. That's not good.

So responsible means responsible for the tasks at hand. Also responsible for the person's welfare. So A means accessible. So in AI application area, so many people actually sometimes forgot about it. In order to create AI solution or AI travel assistant, it's really very difficult. It's not, it's much more than meets the eye.

So in that case, it is that we wanted to power the domain experts to have the control of AI, right, to be on the safe side and to also have the knowledge to teach AI the right domain knowledge. Which means that we call it accessible. Which means that you want the AI to be empathetic and to be responsible and to be accessible, to really realize the vision we're just discussing about.

- [Neil] The field of artificial empathy has been around for a while and some people like myself actually argue that we've seen that AI can actually be more empathetic than another human being. And, but the challenge I commonly see, especially with business, is that they find that hard to believe, right?

Like, how can a machine that doesn't feel any emotions really be empathetic or do it at times better than a person can? When you experience something like that, what's your response to that question?

- [Michelle] I'm smiling because I got this question quite a bit, right? People always said if the machines doesn't even have the emotion, how could the machine exhibit empathy, right? So to my, actually especially for my research, to my experience, that's completely opposite. Having empathy, feeling the emotion versus demonstrating it, they could be very two different things.

Actually, machine can imitate humans to demonstrate empathy without having empathy. Actually, it's a better thing. You know why? Humans they, for humans to have lots of empathy, they normally need to be feel, right? They feel what, I feel what you feel, I think what you think, then I've, I show my empathy, right?

That's actually, it's a lots of burden on humans. That's why psychological studies show that the first line workers like a firefighters, the police men, doctors, especially ER doctors, the people who have lots of emotional actually attachment to it, they are the people who are first to be burned out because there's a double-edged sword.

If you can have a, kind of from a human point of view, lots of emotions inside you, and you of course you demonstrate empathy, but they're easy to burn out. But machines don't have that, how to say, baggage, right? Because the machines don't feel anything. But if you teach machines, how do you showcase empathy, that's very possible. That's what I think it's a machine's actually advantage versus disadvantage. And I just gave you an example. And I think about this way, right? So we have been collaborating with the learning institutes and healthcare facilities as well. In one of the learning institute, we use the, they use the AI assistants on Juji to help their students do teaming, to figure out who are the students better together to be a more effective project team. And some students during this conversation with the AI, they expressed a lot of anger, really anger, because maybe the past, the teaming experience was so bad, they felt like it is just wasting of time, right?

The AI doesn't have this same feeling per se, but AI can show the empathy and not just that, so these students dump a lot of anger on the chatbot, and chatbot AI still can remain very calm and polite. But humans, you can't do that because humans take in the emotions and express it. But the machines, because they don't have the feeling side actually can just showcase the feeling side.

So I, so this is something humans really can't do. So you separate your intaking from your kind of expression.

- [Neil] It's interesting you bring up the emotional aspect. One thing that we found like with artificial empathy is that the AI is much better at reading the emotional state of another human being because it has laser like focus on you, right? It's not thinking about what it wants to say back. It's not thinking about what am I going to do for dinner or where the kids are.

It's totally focused on you, the words you're picking, your body language. So it's picking up on a lot of different especially nonverbal clues that we as people might miss. And as a result, it, because of that, it understands our emotional state to a degree better, it can respond and interact as a result better than we often can, which I just, I think a lot of, I find it fascinating.

A lot of people find it frightening, actually, that it diminishes a bit of our place in the universe because people feel like emotions are the realm of human beings, which actually scientifically is not true. Most animals and even plants have emotional stating. But it's an interesting challenge.

If someone, if you have a prospective client or customer to push back on you on that, how do you try and assuage them or persuade them?

- [Michelle] You mentioned another interesting one, which is interesting. You said that the machines the kind showcase the empathy better is because it actually concentrates on you, right? It's true. Focus on you because the machine doesn't think other many things at the same time because it was made to do one thing at a time, which is interesting.

And another side, but don't also forget it is this. Many studies, including our studies, show that humans are much more open with machines than with human beings. This is a reason for it, right? Because of the, in psychology, people understand there's social desirability biases. So it doesn't matter, humans always want to showcase their strongest side, their best side.

They don't want to show weakness, right? So that's why when you see the machines, like I experienced this one first hand, when I was hiring intern, and we asked the intern to interact with machines first and interact with me as a human, and I was looking at what they're interacting with machines. They were very honest.

For example, I remember I asked the same question as a machine asked. We just asked, okay, we have three tasks in the summer. And which of the tasks do you think you are best at? In the machine, in the human to machine conversation, very few intern candidates choose all the three. They always choose one.

Even they said, jokingly to the machine, they said, Oh, please don't, do not ever let me work on task two because I'm not good at it. But you know what? When I interviewed with some of them, everybody told me they can do all three. Nobody ever admitted it. They couldn't do two or three or pick one. They said, actually, I can do all the three, right?

So if you compare even the same for the same person side by side, they're much more open. This is for a reason to it. First one is a human reason. They want to showcase their best side, their strongest side to another human being. But when they interact with machine, they care less. Another side it is, they don't think a machine will judge them.

They always think, so as I said, they always think machine is almost like a child. It's a machine, right? It's a, they tend to be much more honest. Which is, that's why I said in my ERA acronym, I stress the responsibility because of the reason too. Because the machines can get a much more authentic, much more open responses from the human beings.

You don't want the machines to use such information against the humans, right? To basically abuse that trust. So that's one, okay, that's one I'm worried about every day. We need to figure out the way of to how we can how the AI can be used to help people responsibly versus to be abused.

- [Nikolai] It's kind of interesting because right now we have a lot of interaction with chatbots. I'm curious, like, how you think things like trust and responsibility might change if we have an embodied AI. So like a robot that was empathetic as well. What are your thoughts on robotics and how that might change the dynamic?

- [Michelle] There are studies explicitly at testing the human form embodied agents versus just a virtual agents, right? This is without the physical form. They found out that it is depending on the topics. I remember. If I don't remember it wrong, I remember given the topics, what you're talking about, what the tasks, right?

This number one. If the tasks are relevant to the if for example, it's a physical tasks, maybe they're much more willing to trust that the robotics one because of the physical activities involved. But in other type of things, if not no physical activities involved, there might be a no difference or sometimes even more distrust in the physical forms because the physical forms are not so good.

For example, embodied agents, they are not very high quality, right? So then people actually tended to, I remember that even people tended to not trust the model because they felt like it's just a fake, right? So that's the one. But however, your question does elude to another one which is something we have to be careful about because this natural, like because of generative AI, this large language models, the chat and interaction becomes the more and more real sounding, even though it fills with the wrong information, right?

We could, we call it a hallucination, right? So false information. But if people starts to just to trust that without the checking the facts, that part, I'm worried about it because if they don't, how'd you say, be more of a vigilant to check out the answers. That's a part of it.

For example. It's very scary if somebody goes there to ask for medical advice, but the information, I saw there's an article, somebody did the study, and the large language models will cite references which don't ever exist. And if somebody said it is, okay, you might want to look at, look up this one, this is how the, maybe for example, Alzheimer's disease is normally being treated.

And then if you don't check the reference really, they just gave you a bunch of things, you said, oh, I can go try it out. That's scary, right?

- [Ryan] If I'm listening to this, and I'm not as technical or maybe not on the engineering side, I might wonder how some of this stuff is even possible. And two of the things, I read two things on your website that I thought was really interesting. Three actually, but we already talked about one. One is around empathy, but the other two were psychographic insights and kind of that authentic connection that these chatbots and these chat agents are able to create with humans.

Talk a little bit about what that means and how that's even done. Cause I know a lot of us have interacted with chatbots and it's very black and white kind of the interaction, but in order to achieve a lot of the things we're saying, in order to build that kind of relationship, in order to have these new age chatbots do the things we're talking about them being able to do, those things are super important, but they're also, empathy is, I think, a thing that people can relate to really well, but when you talk about psychographic insights and building authentic connection from a machine to a human, that might be foreign to a lot of people understanding, like, how that's even possible. So maybe if you could just, at a high level, explain what those two are and why they're so important for a company that's potentially going to bring in a tool like this.

- [Michelle] I think some of you probably have tests, have used the personality tests. Have you ever done that before? So it's a set of the queries, right? Ask you to evaluate different statements on the scale of one to five. So that's why we call it a very traditional assessment to assess your personality, to assess your unique qualities and characteristics.

But however, this kind of a method, it's very subjective. So for example, if you say, I like to work with other people, I put a five, Neil put five, or maybe Ryan and Nikolai also put five. But our four fives maybe mean many different things. They mean completely different things, right? So it's very subjective.

Another one is also very unnatural. We call it artificial. For example, some of them, because remember, those tests were designed by scientists within a lab. So it's a, so for example, one of the ways to test whether you have a aesthetic interest, it will say, I like ballet on a scale of one to five.

How much do you agree with it? So let's say somebody who is actually very artistic, very has a very high level aesthetic interest that has never heard of a ballet. It's very possible, right? But it doesn't mean this person's not a creative or aesthetically, have a aesthetic interest. So what we have done here it is to, because my, thanks to my co-founder as well, and he's a computational psychologist, so instead of asking people to take those very subjective tests, we are analyzing people's chat behavior, like we're doing right now. So the behavior of a chat behavior, we're focusing on mostly text. And based on the chat behavior, the machine on the fly automatically infer your personality traits. For example, what personality traits encompasses, actually personality traits normally encompasses three types of characteristics of the individual.

First one, what are your interests and passions? For example, some people like to paint, some people like physical sports. Second part of it is, what are your skills? What are you good at? Some people are very good at painting. Some people are very good at singing. So that's your talents and the skills. The last part, it's also one of the most important part parts, it is how you're handling life's challenges. Some people handle life's challenges, which means like a social challenges. They love to interact with the people, and some people want to keep to themselves. Other people remain calm under pressure, but other people get very worried, right? Very, have a highly level of, high level of anxiety under pressure.

So you can see, so this is very three broad aspects of the characteristics. Juji's cognitive AI now can infer by examining, analyzing people's what I call it communication behavior. That's why, so you say that why it's possible, so we made this very important one. It is we made this no code. So we said it is if you as long as you can use PowerPoint and Excel spreadsheet, you can come on our platform, create a very custom AI assistant to help you engage your audience.

For example, schools are coming here to use our platform to create a assistant to engage with their students. And healthcare organizations to use this to engage with their members or patients, right? So that's so based on this conversation on the fly, the AI, it's already can analyze the user's behavior and infer the user's characteristics and use that to make suggestions to guide the user behavior.

And of course, the information alone can be used by human beings in this process, right? So you come on, think about it in a school, you have a human advisors, human counselors, you can use the information and help better help students.

- [Neil] No, I think that makes a lot of sense because you always hear that like language is a fingerprint. It's very hard to mask. There's some things we can do to try and mask our body language or things like that. But when we talk, it's really hard to be picky with our words because it tends to come naturally to us.

So deconstructing that is actually a really good insight into the person. So it sounds like you're using some level of neuro linguistics as part of your AI platform. Is that?

- [Michelle] So the part we used for this personality inference, we called it's basically called a computational psychology, right? So you want to identify the evidence, we could also evidence based psychology, computational psychology, which means that is based on the communication evidence. We are trying to figure out what evidence will, they call it, you can think about this as being a theory called a latent trade theory, which means it is what evidence can help you infer a person's unique characteristics. For example, one of the unique characteristics would be simply people can understand very well it is extroversion versus introversion, right?

Introvert, extroverted. And there are certain behavior can really showcase that person's extrovert-ness or introvert and similarly for other dimension as well. For example, people's openness to new experience, right? Some people are very open and some people are not. Of course, they're not going to say that, but within a certain communication behavior, we can actually detect that fairly accurately.

So we just, this year we have a five university independent study to show that Juji AI inferred those personality traits can predict real world behavior. So that's very, that's a huge one. And also compared with the traditional assessments, which I mentioned earlier, and either on par with the performance or better than the performance of the traditional assessments can do.

So this is a huge encouragement to us, which means that is AI really can do this.

- [Ryan] You've mentioned a good bit about the when we talk about chatbots and stuff, I'm curious if you could talk a little bit about, in addition to what you've already mentioned, when we're thinking about, or if I'm listening to this as a company that has chatbots and uses chatbots in our daily life to interact with customers or maybe interact with staff, you name it, what is the real benefit that they should really be focusing in on or understanding that all this stuff we're talking about, bringing AI into these chatbots experience is going to help them enable or enable them to do with those interactions. Obviously, we've talked about kind of use cases in the healthcare space, the school space, things like that. But if we're looking more just enterprise across the board, what are some of those other use cases and some of those other benefits that should be considered when it comes to people wondering, okay, should I really be bringing AI into my experience for my users?

- [Michelle] So I would say it's really about three things, right? So first thing it is deeper engagement. So every brand really wants to engage with their audience, doesn't matter they are customers, they are partners, or they are employees, right? So the deeper engagement really help to connect, to establish that the connection that's required for the organization to function better doesn't matter to affect your workforce, right?

So this is for deeper engagement. If your AI can be empathetic and then it's really deepened that engagement. Number two, it's from a deeper insights. Right now, like you see the customer service chat, but like some of our, it's not just about, they're not empathetic. They really stay at the level, at the surface level.

So I knew that, and you asked me today, and he is actually struggled with this. For example, I'm just give example on you. I know you're not struggling with any learning programs. So they say he's doing a learning programs in law, let's say right, in maybe in AI law area. And then so the normal typical chatbot or AI assistant would just say, okay, Neil is struggling with X, but it doesn't go deeper to understand the why Neil is struggling with that.

It might be, this is not a good area for Neil to get in. Maybe Neil could partner with somebody because Neil is a great social person. If he has a great partner, he can actually overcome this obstacle. But in other times, but in another way, maybe Neil's best expertise is not in AI law, maybe it's in AI marketing.

So do you see that level of insights are often not extracted, are not definitely let alone to be used because they don't have the insights, right? So it's a deeper engagement and deeper insights. And then last part of it is I would say it's a deeper action, deeper guidance, deeper help to right?

Because if you have that insights, then either the humans of the organization or the machines can help you better, right? Those that maybe you can change your measure in a way, or maybe you can partner with you somebody because you're, because of your social power, somebody is willing to help you, right?

So that's so it's basically the really the deepen on the engagement, on the insights as well as the help side, the guidance and help side. I think it benefits all the three areas.

- [Ryan] Yeah, absolutely. No, I appreciate you breaking that down because I know a lot of people who will potentially listen to this are going to wonder how this applies to them or why this would be something they strongly consider. And it's, the access to the data, the better insights, the relationship they build, the understanding, all of those things matter in addition to just providing better support or better resources for their end users.

So that's fantastic. I appreciate you talking about that. Neil, unless you have something you wanted to add there, I wanted to jump in, I wanted to go back to something that was mentioned a second ago about no code and your all approach there, if I could shift topics just for a second.

But you mentioned no code and being able to enable people to use a tool like this for without having to be an engineer in any capacity. So how does, how have you seen the I guess no code AI kind of just evolve and take shape with the interactions you've had with customers and things like that because I know it could be daunting to think about how do I bring an AI into my business.

But if they're, if you're not technical or an engineer, there are ways to do that. And I think it's important to shed light on that from an organization perspective.

- [Michelle] So you know what? Before Juji's existence, there are already no code AI tools, right? But in those no code AI tools, normally it's for personal productivity gain, right? For example, I'm going to use this AI to help me. I'll write a paragraph or better or maybe help me analyze the image, right? That's all AI as well.

But what we're talking about it is actually we took the, we shoot the video. We wanted to give people an idea, how can you bring AI into your workforce? Not just the IT people can use it. It's really enable, empowers everyone in your organization to adopt AI. So let me just give an example of that. So because I said it is, we have been working with the education and healthcare organizations, right?

Thinking about the health care organization, let's say there's a instructor, a coach, right? The coach, let's say the coach is about leadership coach. This coach doesn't know anything about coding, doesn't know anything about AI. But this coach could really use an AI assistant. For example, before the coaches, before the students get on boarded, they started their coaching sessions, the coach might want to know who are those students, what they're like, what are the expectations, what are their potential challenges in my coaching sessions, right?

That would be so great. But this coach is only one person. I have 50 students. I can't possibly talk to 50 hours already. So how cool it is that you have an AI assistant that does that work prior to your coaching session. That's great. But you know what? I'm just a coach. I don't know how to write Python. I don't even know what API it is.

So I just, but I know how to use PowerPoint, how I know how to use Excel spreadsheet. So I can use maybe the tool like GG to create my AI assistant, to tell my assistant, literally like my intern, to say, you know what? Today I'm going to send you to talk to 50 students of mine who will be talking to me in the next six months.

You do this, right? You go ask them what who they are, what their expectation is, and also give me their individual unique profile so I can use that better help them. That's what we mean by no code. It's really. You are going to, like you in a, in a workforce, like an employee, create your own custom assistant, and then use that to help your work not to replace you.

So it's never meant to replace anybody because you are the final decision maker, or your AI assistant can do everything you don't want to do, or you don't have time to do. So that's an idea. Thinking about in the healthcare will be exactly similar scenario. If you are healthcare coordinator, you have a many patients are coming in.

You want the assistant that does very similar things. I want to know the personality, temperament of each patient, the incoming, right? So I can know how to deal with them, how to better comfort them. Healthcare coordinator doesn't know Python, doesn't know JavaScript.

- [Ryan] It's been an interesting conversation. We've had some other guests on previous episodes talking about these AI assistants and the approach of them not being viewed as something that replaces but something that helps and supports, but the adoption and understanding how to bring that into a business as either an employee or a company to benefit their employees is something that I think was worth talking about and shedding light on because like you said, you don't have to be an engineer to bring this stuff in to the business.

You can benefit from these technologies. As long as you, once you understand the value they provide, you can adopt them. And it's something that I think is important for people to understand and not be resistant to or feel like that's a barrier to adoption for their side of things.

- [Michelle] So that's why we always tell the organizations, first of all, it's not to replace your workforce. On the opposite, it's to augment your workforce, right? So far we have been very successful on that. Because once they see what the AI does, and they're very happy, then even when we have those recruitment specialists and marketing specialists, they say that, that's great.

So I don't need to stay up at two o'clock in the morning to answer emails. I say, sure, you don't need to do that anymore, right? And So that's really the power of automation, I would say. And and still, one of the things I also wanted to stress it is it is automation, it is the augmentation of the human time.

But in the meantime, we still put human in the driver's seat, right? The humans can dictate what the AI assistant can do and cannot do and what to do, right? But not how to do. That's the whole point of this difference. Thank you. You, as I said, I keep using this intern, like a junior, or a junior assistant role.

It is you, because they have a certain intelligence used to tell them you do X, Y, Z. But I'm not going to micromanage you to tell you every step on the way how you're going to do it.

- [Ryan] Where do you see this all going? Like how do you see this evolving? How do you see chatbot assistance, AI assistance, these not just being adopted across different industries and different roles, benefiting businesses in many different ways, but how does this evolve? How is AI going to, or are these tools going to become more responsible, more empathetic, make a better connection, provide more value? Where do you just see this going? Or what's your kind of view that, outlook on all this kind of over the next number of years.

- [Michelle] I think in the next five to ten years, not far away down the road, I do believe this earlier vision of this computer scientist called J.R. Licklider from MIT invention, it's going to happen. It's called a human machine symbiosis, right? We call it a human machine or human AI teaming. So we really believe that every human, every one of us would have our own assistant.

So the assistant is not just a personal assistant to help you turn on your music, turn down your temperature. The really understands you to your core, probably understands you better than you understand yourself, your mom and dad understand you, and they use that insights to help you achieve many things that's almost sounds too expensive or too hard to afford, too expensive to afford in today's situation, right?

So really help you be your, suggest what might be your career path, what majors you should be in, what learning programs you could use to rescale, upscale, and as well as what kind of adopters will be the great match for you. And what kind of financial investments it's would be, again, best for you based on your personality, maybe somebody who are not very adventurous and somebody are very adventurous, right?

So it's really gives you that kind of suggestions, but still, again, you are the ultimate decision maker, but you do have that AI assistant at your disposal.

- [Ryan] Neil, any last questions, any last words?

- [Neil] I think it's been a great conversation and seeing how Michelle and her company are advancing artificial empathy. It's also, I think, interesting talking about a little bit about human symbiosis because at least in my world, we often talk about hybrid intelligence, where we know that people are good at certain things, and we want to augment our own capabilities, machine capabilities.

But that's not quite the same thing. The Holy Grail of assistance has been to have an AI essentially personal assistant that knows you as well as you know yourself, which like the Black Mirror episode White Christmas and sounds like this is a good step towards that.

- [Ryan] For our audience that's listening to this and wants to learn more about what y'all have going on and maybe follow up touch base in any way, what's the best way they can do that.

- [Michelle] I think they were just, if they want to know more about our work or something, the best way it is to follow us on LinkedIn, because we actually building an AI community on LinkedIn, and we welcome all kinds of input, suggestions, and feedback. And I think so because our theme is we also talking about democratizing AI and the democratize AI really takes a village, right?

Takes the world. Effort. We want people to come to chiming in and helping us, and we can help together. We can help the world. It's cool. It sounds like a cliche, but it's real. To make a better world.

- [Ryan] When we started IoT for All, now with AI for All, it's really been about helping make these technologies easier to understand, more accessible to people, so that they don't fear them and that they understand their value and can really wrap their heads around what's trying to be accomplished that we can work on this together as a society.

So fantastic stuff you have going on. Really appreciate your time chatting with us today. It's been a great conversation as Neil said, and we're excited to get this out to our audience and hopefully we'll have you back at some point in the future.

- [Michelle] Thank you very much. Ryan, Neil, and Nikolai. Yeah. Thank you.