Introduction

So I know there is a lot and lot of conversation around agents. People have said that this year is going to be year of agentic AI. And other speakers would be talking about the agents in detail.

We have had this conversation a couple of months ago as well. So today, I thought let's focus on something which is more close to us.

Agents in Day-to-Day Life

Where do we stand when the agents and AI systems become our commonplace in our day-to-day life, which is happening? And how do we lead the next revolution of collective intelligence, bringing in the human intelligence the intelligence which comes out of the AI systems and autonomous agents.

In case if you'd like to connect, this is my LinkedIn QR code. Happy to continue the conversation as we move ahead in this journey together.

So, yeah, agents are everywhere from our... mobile apps, to our social media feeds, as well as the various news articles which you read.

So I'll say, let's now focus on how and why the human centricity of AI and AI systems would be important and why that is the focus of today's conversation. So if you remember, when a lot of us were younger, there was this match between Garry Kasparov and Big Blue and unfortunately Garry Kasparov had lost that match after a lot of discussions and there were a couple of bouts and then they decided that now the chess should be played with humans and machines together. And that is where this idea of having a combination of humans and machines for solving our problems came about. 1And they realized that if we have this combination, it would do much better for us.

Anybody who would like to cite any examples from what you read, what you experienced, or what you watched in some movies about how human-machine combination. When I say machine, it could be AI-based machine, it could be a computer, or it could be a mechanical machine, where a combination of human and machine was able to do things which humans alone or machines alone could have not done. Anyone who would like to share your thoughts? Something you read, something you did in your coursework at school, or in your office work? Anybody? Any thoughts? Anything which comes to your mind, anything which you read, any fiction, anything you watched on the television or on the Netflix?

So essentially, if we try to put human intelligence, definitely it remains the super intelligence, the best of all, because we can perceive things around us, we can take decisions. When we take decisions, they are towards a particular goal, we take certain actions, and we have a lot of reasoning power, we have a lot of power of reflection. Yes, humans are trying to make machines as good as what humans do in terms of reasoning. People must have heard about R1, DeepSeq R1 model, right? Large reasoning models, DeepSeq R1, OpenAI's 401, O3, Claude 3.7, they all are reasoning models which have been trained specifically to reason. that's happening. But other than that, if you go beyond that particular dimension, you will find that there's another component of bringing this machine intelligence and human intelligence together so that we can do a lot more things as we go ahead in our day-to-day activities.

Understanding AI Agents

I will very briefly touch upon agents. So anyone in the room right now would like to talk about what do they understand by AI agents, autonomous agents. What will you call not an agent? Yes, please. Yes, so essentially, If there is a system which can bring about sense from surroundings using different sensory inputs, do some actions perhaps improve its performance to deliver certain task to achieve a goal autonomously without human intervention based on what has been programmed for it would be called an autonomous agent. Anything which does not do this could qualify as an agentic workflow but would not be an autonomous agent.

And then there are There are chatbots, there are conversational bots, which we have been traditionally using. So you will find there's a shift towards autonomous agents, but there are a lot of other things in between the two, which we have been all leveraging in our workplace, in our personal usage. So the idea is to create a system which has all three sets, humans, AI systems. They could be large language models, large multimodal models, or small language models, because most of the models have now become multimodal. I hope everybody would align with the idea of what a multimodal model means.

So if you recollect one or two years back when you would use ChatGPT, it would perhaps only take text inputs. Subsequently, they enhanced it from 3.5 GPT-3.5 to GPT-4. GPT-4 revolutionized a lot of things. You could upload pictures. It could calculate your bill. If this is a menu card and you had one beer and one sandwich, how much will it cost? You are in Ontario. This is the HST. If you are in BC, HST is different. It could do all that, right? And it also saw enhancement in the OCR plus visual language model. That is a shift, right?

And now leveraging that technology, you will find a lot of advancement in these tools which we are using in our daily usage personally and at work, right? So this is something which is a study done by this professor who is also an entrepreneur. And it proves that if we have a combination of AI plus humans, you will be able to attain a lot more than either we are just using AI or we're just using a human, right?

You will be able to find this. So he has done this research with, he teaches at MIT, he teaches at Stanford, and the research is public. So you will be able to access that and you'll see that Initially, they did it with students, then consultants, and then working professionals in a few of the retail industry organization, and they found that a combination of AI plus human was able to perform much better. The results were a little more aligned to what was expected, and definitely people had better productivity, so they could devote their time to do something more creative, more strategic. Right?

Historical Context

So we talked about Garry Kasparov. So this particular article appeared in Harvard Business Review in 2021. And the other author, David Kramer, he is the dean of Northwestern University School of Business, Boston. And they realized, and this is Garry Kasparov himself mentioning that, that AI would be augmenting human intelligence and should be used to enhance human intelligence, not replace it. And that is what is the whole purpose of this conversation.

And that is where we should channelize this whole AI revolution to, that it should be utilized to kind of augment what we do instead of replacing any one of us. For sure, it should replace certain repetitive tasks which will allow us to do more strategic, more high leverage work. And that is why all you leaders who are in this room should prepare yourself and your teams to kind of embark on the journey where you have a direction towards augmented collective intelligence, right? And if you again borrowing it from Professor Eric, he says that the most valuable firms will be those that find ways for machines to complement workers, not substitute for them. Right.

The AI Revolution

And Any transformation, like previously we have had a lot of transformations. So if you just go back in history for the last 10 to 12 years, cloud, before that mobile technology, smartphones, before that internet, before that semiconductors. So you see any transformation, human society took some time to adopt. But the adoption definitely for AI, especially gen AI, has been quite fast. And that is why you will find that there is a phase which we are right now crossing. Tool phase is kind of over.

We have utilized independent tools like ChatGPT. All of us use Cloud. Cohere. Cohere is our own Canadian reply to open AI. And they do quite well in terms of the performance and other parameters. And now, as we start building these systems, these multi-agent systems, a lot of you would be using them or you would be witnessing them in use, either in your personal or in your professional lives.

Anybody who has used deep research by any of the providers, what is your experience on deep research? Which one did you use? Open AI's, Google's, perplexities? How was it?

So essentially, there are some subgroups of agents, and you are dividing tasks amongst them. So let's say you are a manager managing a particular effort in your team at work, and you have four different teams. One team does initial research, another team does some synthesis of analysis of data, the third team produces those very, very visually appealing artifacts, and the last team does proofreading. So similarly, you can divide these kind of tasks among these agents, and they bring you the best. So you could actually deliver them on using different foundation or small models so that each team or each team of agents does very specific work, and it brings you something which is very, very comprehensive. That is what we are seeing right now.

Human Intelligence in AI

And the next step should be that we bring in our human intelligence, our human experience together with these systems to create collective intelligence. There are some examples where industry is already using it. So Mayo Clinic has been kind of putting this thing together where they're trying to leverage physicians' domain expertise along with some AI tools to bring more I would say, value to the patients as well as other stakeholders in the whole value chain.

Then DARPA has been using it for their own requirements. Siemens is another company which has been leveraging these collective intelligence means for different tools which they have been working on. But what is very important and which is why I would like to urge all of you, essentially all of us are leaders. And that is why we have to focus on the human factor of it.

And we have seen this in the past. This is not the first time that we are undergoing a digital or an AI transformation, right? As I mentioned, cloud, before that mobile, before that internet, we've all gone through it. 1And what is the focus is actually to build that trust within our internal team members and with our stakeholders, our customers.

Navigating Transformations

And then... For us, individuals will have to continuously learn and evolve because a lot changes in few weeks because technology is still not settled. Essentially, we are flying a plane as we are building it. So it is important that we continue to evolve and be up to date. and provide the psychological safety not only to ourselves, but also to people around us.

And that is what will actually help us take this in the right direction. Yes, there are challenges in terms of responsible AI, ethical practices of AI, regulation. All of them are catching up. with the technology, but I think we will have to up the ante and keep ourselves at the forefront so that we lead it and we do not get kind of lagged behind, right?

And it can be true for at every level. It could be at individual level. It could be at the level of group, this group of ours. It could be at our community. It could be at our province. It could be at our nation. So Canada remains at the forefront of AI research. And I think it is all up to us here that we leverage that research to commercialize, to innovate, to create value for all of us, for our families, our neighbors, our community, our country, and become a world leader in creating responsible and ethical and safe AI.

A lot has been done, but I think we can achieve a lot of Better things in this realm. And connecting that to the industry. Because I think all of you, some of you may not be from AI background, from CS background, from data science background. Doesn't mean that you should not leverage AI.

I think you are the right people because you know your domain very well. Nobody knows your domain better than you. You just need to map the whole workflow and bring in somebody who is building that agentic flow or that particular AI solution to tell them that this is where I find problem. This is where I have a lot of issues.

Like all of us have to answer some 100 emails every day. Or if I have to reply, send so many emails, perhaps that part can be automated. That part can be handed over to a set of agents. And we leverage on what we need to do in terms of strategic planning and getting things done.

Perhaps talk to people and bring those insights out so that we solve their pain point. And all of us have to do that. We do it. I mean, even if we are not in sales, we have to do sales, right? We have to talk to people, understand their point of view, and create those solutions.

That could be the human factor which can help us create those things which machines alone cannot create. Anybody who identifies this particular person? Dr. Fifi Lee. Anyone who would like to mention who she and... So she is the godmother of deep learning.

Deep learning, the current generation of AI which we use, which we leverage on our phones, on a chat GPT. It has three godfathers, right? Professor Geoffrey Hinton, who is a Canadian professor who got Nobel Prize last year. Professor Jan LeCun. And... Professor Yoshua Bengio, who sits in Mila, which is in Montreal. But this is the person who created ImageNet, which was a competition which created artificial neural network, which helped creation of the current generation of artificial neural networks, which are the base for deep learning and what is right now what we have today, what we use, right?

She's saying this particular statement means a lot because what she's trying to indicate to us is that it's not the AI which is going to do anything to our roles and our jobs. But at the same time, people who are able to leverage AI better will actually be a competition to us, right? So we should start leveraging it. All of us use electricity for a lot of things, right? Our laptops, our mobile phones, everything at home generally is on electricity.

Leveraging AI for Value Creation

We are the consumers of electricity. We create business. We create value out of it. But do we go and build electricity? Do we go to a nuclear power plant or a hydro power plant or a coal power plant? No. We just consume it. We buy it from the provider who supplies it to us. Similarly, to leverage AI for creating value, we do not have to build AI. We just have to know how to use it to solve our problem, right?

So that is what I would urge all of you, those who are even not from AI or CS background, to start leveraging it for your value creation. And at the larger scheme of things, both at work as well as in person, we all would be like an orchestra conductor. So we are the orchestrator, and I'll talk about it a little later.

in a little while. I just wanted to ask anyone who has used wipe coding. Wipe coding, it's, yes. Anybody would like to share their experience, how it has transformed whatever you were doing? I'm planning right now building various mockups . We're just talking about it. It brings it to the 90%, but we still need to know some coding . You need to fix and debug. It's not perfect. Yes, it's not perfect at all. It's very easy to read from zero to a mockup. Yes. Yes. Yes, absolutely. Would you like to share something? I'm just using for my personal unit task. Probably email, stuff like that. I give that. I'm using . I feel like it's going to bring in some information Thank you, both of you.

So essentially, yes, it helps us take our idea to 90%. So something for which you would have perhaps waited for many years that I have this idea. I will hire a team. I will use my vacation to actually build this particular product or this entrepreneurial idea which you had. Now at least it can bring to that level that you can demonstrate to somebody else. It could be your partner. It could be somebody in office that this is what I want to build. Here it is. This is my demo. This is my MVP. Would you like to work with me? Would you like to invest in it? It can happen for a personal project. It can happen for a professional project. So that is what is the change. And these tools are enabling us to do that.

Yes, they are not foolproof. They may not pass the master of cybersecurity. They may crumble upon when you have five or more than five users because they cannot sustain that. But yes, it brings us to the 90% level. And then we work over it and take it to the next level. So I'll just talk about this particular framework, which we can call orchestrate.

So essentially, as humans, we will need to orchestrate things around the value. And it could be that we have these components in place where we understand what is the organizational architecture. Who is doing what? What is the role? What should you give to the autonomous agent flow or autonomous agent system? What should you keep human?

That is the clarity which you should have as a leader. Continuously develop the skills of your team members because obviously they should not get outdated. Ethics, there is no qualms about it. There is no alternate about it. We will have to ground it to human ethics because anything which you create should be harmless to another human. It should be... Usable 10 years from now as well. It should be totally aligned to human ethics. And definitely you have to have a good infra and a data pipeline.

Continuously learning on this and managing those risks which get created and then evolve and adapt it. and continuously build trust both for internal and external stakeholders. And have a culture of experimentation because today it is live coding, tomorrow it will be something else, and it will keep evolving. You just have to find what creates value for you, what is safer for you, and continue with it.

Future Skills and Adaptation

Very briefly, I'll touch upon that how can we individually or collectively navigate through this renaissance of technology and how it actually helps us equip ourselves to get better tomorrow. This is a very important picture. If you see this, this is something which appeared around eight weeks back from a report which came from World Economic Forum.

This is the mapping of current and future skills which are required by employers across the globe. And I think because it is World Economic Forum, I think it has some weightage because they generally are asking employers to tell what are they expecting in their future employees or current employees five years from now. So if you see the shift, the core skills are plotted on the right upper quadrant, and top most is definitely AI and big data, but a lot of other things which are human-centric, creative thinking,

A lot of people ask us, a lot of students would ask, should I finish my undergrad or should I not? Because AI will do everything. So let me start getting into AI directly. But I think everybody should do the education part because you need creative thinking.

You need to design that prompt which will actually curate and create whatever you would like to. Similarly, for wipe coding also, if you want that particular system to give you what you want, you will have to prompt it. You will have to have those four components of a very good prompt so that it does what you want it to do.

Otherwise, you will keep playing around and you will not get what you would like to. And you will see this. If all of us give this... Given the same problem, given the same set of data and same system, we'll create different outputs because we all are different. We all think creatively differently. And that is where our creative thinking would help us use these tools to transform whatever we want to transform.

And if you will notice, most of these skills perhaps are very human skills and we in our race for whatever we were trying to achieve so far we have forgotten about them. So it is time to put those so-called soft skills back into our bucket list and start taking all of them because we need to get ready for what's in future in like four years, five years from now is that what the shape is. If you see a lot of other so-called skills which we right now give a lot of importance to, they're coming down in this particular priority list. And because that can be offloaded to an AI system.

Anybody who has any questions or comments on this particular slide or aspect?

OK, this is something I think we will experience because our next two speakers will talk about it. But you will see a rise in use of agentic flow as well as autonomous AI agents in the next few years. And a lot of companies are planning to use these tools for their workflows.

Skills of yesteryears, data entry, administrative task, routine information processing, it's already getting kind of changeover to skills of tomorrow, which are more human-centric for leveraging AI, right from AI prompt engineering to agentic workflow handling. Because somebody has to create that workflow so that the agentic system can actually take it on. So this is something which is totally our lookout as human managers, right?

Hybrid Teams

Again, there will be a team, so it may happen, may not happen this year, but by next year you will find in your offices you will have a hybrid team. You will have some humans and you will have some AI systems who are reporting to you directly, right? Some digital employees and human employees reporting to you.

So I think you will have to evolve your leadership style in a way where you can actually handle both of them. And it will be your lookout what you should task the AI system for, what you should task your other human colleagues for. And again, you will have to modify your leadership style.

I'm sure all of you have had good experience of these kind of transformations previously, and you will come along with it. So as you pivot, there may be few of you who would think of pivoting your careers. There are very interesting roles which are coming up as we move ahead.

Career Pivoting

So you can actually leverage a few of them because what is required is the domain expertise in your respective field of industry, whatever your vertical is. Because the best of the... AI builders will not be able to build that particular solution, that particular agentic system without knowing how that particular industry, that particular role works.

We have foundation model, we have agentic systems, we have orchestration layer, but we cannot build it for one particular industry, let's say mining industry in Northern Ontario. Somebody has to tell us what has to be built there, how to leverage those, let's say, satellite imageries to do some kind of prediction, some kind of maintenance schedule. So somebody who knows that system can be very, very helpful for making those larger system changes or organization changes.

Again, human skills will remain primarily important. Technical skills, I'm sure you can continue to sharpen yourself. And the part which is about being the hybrid domain expert is something which will come handy.

If you want to launch your career, this is a very interesting career to get into, where you are the one who's staring, who is in the driver's seat, who's making sure that both the digital as well as the human team members are moving in the direction where you want to take it. And this definitely can change a lot of our issues which we deal with on day-to-day basis, both individually as well as as community.

Why I would urge you, because this is something very interesting. So as a human being, we can do a lot of things, right? You might have started your career as a doctor, as an engineer, or as a lawyer, or as a sales professional, but nothing stops you from pivoting, right?

Because we are humans, right? We can always pivot and do better things any time of our career, our lives. Because we can be journalists, and I think it is time to be a journalist and pivot as it is required.

Conclusion

Thank you very much, and if you have any questions, any thoughts, please go ahead and share.