The Meridian Point Podcast
Interview with Suzel Wyvill-Jones, Founder of Mindshift Dynamics
Host: Kumar Dattatreyan
KUMAR: Hi everyone, Kumar Dattatreyan here with The Meridian Point. Today I'm joined by Suzel Wyvill-Jones, founder of Mindshift Dynamics and author of AI-Powered Business Transformation: From Strategy to Scalable Execution in Weeks. I think I need to get that book.
Suzel spent over two decades inside large enterprises like AT&T and Cricket, leading $120 million-plus portfolios across telecom, financial services and supply chain. Then she did something most senior leaders never do. She walked away from the corporate ladder to build her own firm, helping executives cut through the AI hype and get to real execution.
What sets Suzel apart is a single conviction that runs through everything she builds. AI should amplify human leadership, not replace it. She's also one of Atlanta's leaders in the AI Collective, a grassroots movement of more than 160,000 people pushing for responsible adoption and education. She's here to share how leaders can identify the right AI opportunities, prove ROI and adopt AI responsibly without chasing the next shiny thing.
So without further ado, let me bring Suzel on stage. Hi Suzel, thanks so much for joining me today.
SUZEL: Thank you for inviting me. It's a pleasure to be here.
KUMAR: I really appreciate you coming back, because this is our second take. The first recording somehow got deleted by the internet gods or demons, however you want to call them. So we're doing this again. I get to speak to you again about this fantastic topic.
SUZEL: I am always ready to talk about this.
KUMAR: All right, so we'll start with this. Every executive wants to say they're doing AI, but most can't really articulate the actual business problem they're solving. Why do so many companies struggle to turn AI interest into results?
SUZEL: Usually because they are looking at the shining object. They have AI, but they are not looking at what kind of problems they actually need to solve. AI can be used to solve lots of use cases, but some use cases cannot be solved by AI. If you just bring AI for whatever reason, it's like you go in shopping. You don't know exactly for what, and you buy lots of things that don't work anywhere and don't solve any of your problems. I believe many companies are doing that.
KUMAR: When you're talking about companies that do that, do you have some examples? A company that tried to implement AI and was unsuccessful because they weren't intentional enough? That would be valuable to learn from.
SUZEL: There are many companies, and as I said, they just want you to come in and resolve a problem. Some companies are looking at AI to resolve a problem with fraud, for example. AI is very effective for that. But they do not look into the environment. They do not understand their data.
Usually it's not a good idea to start AI at enterprise level. You should start with smaller organizations within the enterprise and run a pilot to see if it's going to solve your problem. You have to have the data ready. There is no AI without data.
So people bring companies in and they have a demo. Of course, demos work because you have synthetic data. Everything is perfect. Everybody comes and everybody's delighted. They swear this is going to resolve all their problems. Then they go look, and there are many components to a fraud system. You pretty much touch every single system in the company. If one of them doesn't have the data ready, there goes your fraud system. So you need to start small and grow as you continue to clean your data and become more effective.
KUMAR: That makes sense. If someone asked you where they should start with AI, how would you help them separate real opportunities from vanity projects?
SUZEL: I would have to go and do an assessment with the experts, and have an honest conversation with all of them. Understand where the data is, who could claim the data faster, and also evaluate what is repetitive. The best use of AI is for repetitive workflows.
KUMAR: Workflows.
SUZEL: Yes. If you have a dashboard you publish every week, start there. Depending on the dashboard, you can actually gather the data. Project management areas are a very good way to start because they have very repetitive tasks. Usually you just need to gather the data, and it's probably data that's already collected in a system. I would look into that first.
KUMAR: Okay. I was talking to someone about decision-making tools. What's it called, DLM? Some notation for decision trees, so when you get to a certain thing you can sort of automate decision-making. That person was saying AI is not a good use case for that type of system. When you're relying on an automated system to make decisions, it needs to make the same decisions given the same information every time. And AI doesn't work like that. AI may make a different decision based on different factors. It's more like a human, because it doesn't necessarily follow a set of rules. But what you're saying is that the repetitive work that can be automated are things that don't rely on the AI's decisions to be super accurate. It just needs to repeat a set of tasks over and over to take that burden off humans. Is that right?
SUZEL: Yes, I'm saying that. But I don't agree with the assertion that AI cannot make decisions. Planes, many times, their systems are making decisions. So it just depends on the data you're providing. It is dangerous. Well, it's more complex. Let's put it this way. If the data is not accurate, you might have a problem with the AI decision making. Also, if the data is stale, if you do not keep feeding and training the model, then you have what we call data drift. AI is making a lot of decisions. We are diagnosing cancer with AI.
KUMAR: Sure. That's interpreting data to come up with a diagnosis. I think it's different from the use case this gentleman was sharing with me. His was more around decision trees for creditworthiness, making a decision on how much credit to give someone. Do they qualify for a loan or a mortgage? Or something even more financial. For financial types of transactions, it needs to be very accurate and the decision needs to be traceable. You need to be able to tell how the decision was made, because if it was made in error you've got to be able to show auditors how that decision was made. Then you can change the logic, the flow, so it makes the decision a little differently the next time. That's where he was saying it's great at interpreting, but it may come up with a different decision each time, because that's the way AI is. It's kind of like us. We may decide differently given the same circumstances.
SUZEL: Yes, it's probabilistic. It depends on the data and the assumptions the tool arrived at based on the learning process. There are some systems that are making at least part of these decisions. This goes back to what the Pope actually published yesterday, his entire study about AI. AI can be very biased.
In the case you just presented, we have a biased credit system, right? Based on very different factors. If the data is biased, the decision is going to be biased, because that's what AI learned. This is the problem also when diagnosing diseases. If most of the data you have is from a particular race, you might miss the diagnosis or provide an incorrect one. Because our bodies are different. Process is different. So, yeah. It is dangerous.
KUMAR: That's a good point. A lot to unpack there in terms of where to use AI and how to use AI. What would you say is the single biggest misconception leaders have about AI right now, and how does that impact the decisions they make?
SUZEL: I believe leaders are assuming AI can solve everything, that we don't need humans anymore. We keep hearing about the number of people being laid off because AI is going to substitute them. I think this is going to be a turnaround at some point very soon, because we are not at a point yet where AI can substitute human decision or human interaction. I don't think we ever will. I do believe we are going to see some changes. But there is a belief now that AI can solve everything, and that's not true.
KUMAR: I agree with that. You mentioned something a few minutes ago about projects, that AI could eliminate some repetitive project work. How do you think that's going to affect the jobs for project leaders and project managers? What should they be doing right now to stay valuable if large parts of their work can be automated?
SUZEL: The part of project management that can be automated is the part every project manager wants automated. It's the boring part. The reporting. But project managers should be learning how to manage AI projects, because they are very different from regular software development.
AI projects pretty much start when you deploy the product. You are going to start feeding the model, and you actually have to be kind of a machine learning ops engineer who is making sure the data is not drifting. AI can give you results. For example, I load the data today, which is 60 days old. The data is going to stay the same. The report is going to be populating the same day. I don't know it is wrong. The data is incorrect.
This is one of the dangers. Right now, when things break with software, you know there is a problem because nothing is working. With AI, you're not going to find out until someone notices, well, there's a problem with this data. Project managers need to start paying attention to that.
One of the issues we are having is that project managers are managing artificial intelligence projects the same way they used to do regular software development. So 86 percent of the projects are failing right now. An AI project should be managed differently. It should be more iterative.
KUMAR: More agile, would you say?
SUZEL: More agile. Actually, this is the word. We are going to go back into agility, because agility got a little bit of a bad reputation. Not because agility is not good. It's because many companies were changing people's titles but not changing the process. Now we are going to have to adopt agility. If not, there is no success.
KUMAR: That's interesting. AI transformation and agile transformations are following a similar trajectory of failure at first, where AI projects really need that discovery approach, that agile approach, to be successful. Agile is perfectly suited for that. So maybe there'll be a resurgence in at least thinking that way.
You mentioned the Pope and his warning to humanity about AI. Europe is already regulating AI aggressively. The US, not so much, right? It's the wild, wild west over here. Most companies are in some state of AI awareness. Some are all in, trying everything, maybe on the wrong things. What does responsible AI adoption look like in practice?
SUZEL: I believe Europe is a little over-regulated. When you regulate anything too much, you stop creativity. In the US, it's the other side of the story. There is no regulation whatsoever. We need to regulate AI, because AI can do a lot of damage. You can easily poison water if you are using AI tools, and we have the drones. They are all AI powered.
Education is very important. I am actually starting a program of AI education for seniors, because AI now is a scam on steroids. A lot of seniors and people who are not as aware of what AI can do, they trust it. They don't know it's impersonating something. We need to educate the population about AI. This is not going back. We are not going to eliminate the existence of AI in society.
It's like the Pope said. He likened it to the Industrial Revolution. It is going to be like that. So more people need to understand it. The more people understand it, the less power you're going to have to scam, to spread false information, all kinds of manipulation we can do with AI that people might not understand.
I have family members in Brazil, and of course we have the family chat. Sometimes they put things on the family chat that, when I look at it, this is totally fake. One of these family members is a surgeon. He's older, in his 80s. He's not uneducated, but he believes everything he sees on the internet. He assumes it's true. So I do believe people need to learn how to distinguish the information.
KUMAR: That's a good point. Regulation in general can stymie progress. But as individuals, we have a responsibility to use these tools appropriately. To educate ourselves on how to use them, the dangers of using them and how to protect yourself from those dangers as much as you can, so you can actually use it for good. For your own good and for society at large.
SUZEL: It can be used for good or bad. That's true.
KUMAR: It can be. What's that new Claude model, Mythos? Claude doesn't want to release it to the public, because it can be very effective at exposing all the holes in a system that will allow hackers to get in and do bad things. They're not publishing it. But the fact they said, hey, we have this thing called Mythos and it's bad, so we're not going to publish it. Okay, but how long are they not going to publish it? Someone's going to get it. The government's going to get it. If it's in the wrong hands, it can be used for nefarious means. I think there does need to be some regulation, to your point.
SUZEL: Yes, but again, I think we need younger people, or people who actually understand it, to regulate it.
KUMAR: Yeah, people who are growing up with it or are more connected. I agree with you.
One of the things you said in our preparatory conversation is that AI should amplify human leadership, not replace it. What does that look like on a leadership team where AI is amplifying the leadership qualities of the humans on that team?
SUZEL: If we are using AI for repetitive tasks, we will allow humans to be more dedicated to human interactions, and give us more time to understand. For example, in software development, it gives me more time to understand what my client's needs are, so I can deliver what my client actually needs instead of what I believe they need.
We spend so much time in big and small companies producing PowerPoints, trying to communicate through graphics, instead of actually spending time talking to each other and trying to clarify. Right now, I can produce a PowerPoint in about five minutes if I have the data for it. That allows me to be a better professional and a better person. AI can also help people write content. You can develop content to teach other people.
As we go, the super intelligence is based on quantum computing. Quantum computing is based on possibility. Before, in regular software, you had a zero and a one. A true or false. Right now in quantum computing, you actually have to consider the zero and one together as a possibility.
Instead of dividing society, if we were to look into technology and where the technology is leading us, maybe we should look at ourselves together as a possibility, instead of the separation. About your zero, go with your team. Your one, go with your team.
KUMAR: I love that analogy. Makes a lot of sense. Good stuff. I need to do more reading on the quantum computing side of things, because I think that is going to be the next evolution of AI and computing in general.
In the intro, I mentioned the work you're doing with the AI Collective. What is that, and what are 160,000 people coming together to do?
SUZEL: It's actually more now. Every month it is growing. The AI Collective was started by a group of very young people, and they call themselves the human side of AI. It's about educating the community, about bringing more people to talk about it. Not just educating, but actually trying to do exactly what I was talking about before. When you have a lot of people understanding what the tool is doing, it becomes less dangerous.
Right now, even with the AI Collective, we are not reaching 0.01 percent of the world population. There is a lot of work to be done. We have to grow a lot and communicate a lot. We are trying to level the field and inform people of not just the possibilities of AI, but the dangers. The AI for elders program is something I am doing because of what I have been learning at the AI Collective.
KUMAR: I'm going to have to join that and see what I can learn from it and what I can contribute as well.
All right, we're going to wrap this up with some fun questions. We'll see if they're fun questions or not. You partner with development teams in Brazil in your work. What does Brazilian engineering culture do better than Silicon Valley?
SUZEL: We dance a lot.
KUMAR: I love it. So we just have to dance more.
SUZEL: Brazilians like to work lots of hours. But they have lots of holidays in Brazil too. So there is a culture of work hard and play hard. When the work ends, and usually it ends about seven or eight p.m., people just go out and enjoy themselves. They never go straight home.
KUMAR: That's cool. So you called yourself an introvert engineer who didn't talk much. Now you're on podcasts and leading a movement. What changed?
SUZEL: I think this past year and a half has been a year of self-discovery. I did not know that I like to talk to people. But I can be very passionate. AI brought all this passion, and the need to use that to make a better planet became an urgency for me. I have children, and when I go, I want to leave here a better place. Lately, it's becoming very urgent.
KUMAR: For sure, I hear you. If a CEO or an executive could only use one AI tool for the next 12 months, what would you tell them to pick?
SUZEL: An ROI calculator.
KUMAR: A what?
SUZEL: A return on investment calculator.
KUMAR: Oh, ROI calculator. Okay.
SUZEL: Yes. You need to know where you're investing your money. If you're going to be investing in AI tools, put metrics on that and measure every two to three months.
KUMAR: Interesting. For the future, what would you tell a 25-year-old to work on in terms of skill development?
SUZEL: Learn to be a machine learning engineer. Well, do whatever you want, but do learn how to train models. I believe anybody can become a doctor and also train machines for medicine. So this is a skill that is very important now.
KUMAR: Very good. Anything I didn't ask you that you'd like to bring up?
SUZEL: I don't think so.
KUMAR: Okay, I have one more. Overhyped or underhyped: AGI in five years. What do you think?
SUZEL: I think it's overhyped.
KUMAR: Okay.
SUZEL: The data on this planet is not conducive to this.
KUMAR: There's a lot of it.
SUZEL: Yes, but it's not organized.
KUMAR: Yeah, that's true. Interesting. All right, well, thank you so much for coming back. Hopefully this will not get deleted. I will make sure I don't mess up again. I really appreciate your time and you sharing your views on AI, the future of computing, and how we use these tools to better ourselves and the people we work with.
SUZEL: You're most welcome. It was a pleasure.
KUMAR: All right, thank you so much. Bye-bye.
SUZEL: Thank you. Bye-bye.
END OF TRANSCRIPT
