Nir Kaldero, author of Data Science for Executives, is the Head of Data Science at Galvanize, and he’s trained a bunch of executives from Fortune 500 companies on how to use machine intelligence and data science to transform their companies and bring a positive impact into the world.
We spent the first part of our conversation talking about a lot of those myths around AI. Then we spent the next part of our conversation talking about what happens to the companies who really incorporate AI and data science into how they operate. What’s going to happen to these Fortune 500 companies if they choose to do it or to not do it? Nir Kaldero has some thoughts.
Nir Kaldero: I started working in the field when I joined the Israeli Military Service at the age of 18, and I was leveraging data to increase efficiency and pretty much save lives. I truly saw the impact and power of leveraging data to make a positive impact on the world.
Data science, machine learning, and artificial intelligence is a very new field, and when I first heard about it around six to eight years ago, I felt I can make an impact and help reshape this field—especially through education, science, technology—by leveraging my experience that I gained to really make this world a better place and explain to people the benefit of using data and modeling techniques and how it can actually extend to every part of their life.
Dispelling Myths about Data
Charlie Hoehn: When you go in to these meetings, what are some of the things that they think that you have to dispel?
Nir Kaldero: Yeah, there are many of these things, especially around people truly believe and think that artificial intelligence is a technology that will replace us. Some people will think the robot will conquer humanity and destroy the world.
Charlie Hoehn: Really? I want to press you on that Nir because I mean, two very smart people have said it’s a potential existential threat. Stephen Hawking and Elon Musk—why is that a myth?
Nir Kaldero: Long conversation, but I’ll try to make it short. So working closely with the largest technology companies, the one that actually creates all the chips and the technology infrastructure to run machine learning, data science, and AI, I see that they build this technology with three principles in mind. The purpose, transparency, and ownership.
If you really think about the purpose when you build this technology and hardware specifically, you see that companies like IBM, Dell, Intel, and Nvidia build this hardware with the purpose to augment our human intelligence—not necessary to replace us.
We are living in an era of a wealth of data, where our brain cannot process the vast amount of data anymore.
“We are in era of human plus the machine.”
You can really think about AI and all of this sophisticated modern techniques as a brain helper. AI is a tool. It’s a tool that is here to service us, to augment our intelligence, to help us make better decisions on a daily basis, not necessarily to replace us.
We need this technology to basically digest and process the vast amount of data that we have, that we get as an input on a daily basis, and crunch it down and shrink it to a much smaller subset that we can actually read, remember process and act upon.
Currently, with all the data that we have that we tried to process, we cannot make anymore smart decisions. Again, looking at the purpose of the people and the companies that are building the technology. The true purpose here is to augment human intelligence and not to replace us.
Give you an example. Quantum computing has a good side and a bad side. The good side is that now, we can solve problems that we could not find solutions before. Even with super computers. Think about, for example, drug discovery. There are so many drugs out there, people have different preferences and different reactions to certain drugs. How can we find the best optimal drug based on everyone’s conditions and all the drugs available?
“Even super computers today cannot solve that.”
This is the binary approach. Quantum computing is moving away from this zero and one binary approach to think about an infinite interval between zero and one.
This very strong computational power can actually find an answer to what drug to recommend to a specific patient. Everything about finance, everything about drug discovery, healthcare issues, quantum computing can actually provide us an answer to it.
But quantum computing also has a dark side. For example, with enough usage, quantum computer can encrypt The Pentagon, it can encrypt its system. When companies like IBM and Google and Nvidia are actually building quantum, they’re thinking about this anecdote about this kind of like misbehavior and situations where the public can actually misuse the technology. Looking at this anecdote, they put barriers to the technology.
The technology cannot and will not, especially at this point, even looking like five to ten years ahead, will not be able to either outperform us or replace us.
Also, you don’t want to replace human intelligence and judgment, because machine learning and data science are purely looking at past behavior and data. We have way more than that.
Although I’m always trying to tell people like you should always leverage data to make data driven decisions that do not think that in this process or equation, I do not think we should ignore our past experience and even our gut feeling.
The entire idea here is to crunch a vast amount of data and present us with much smaller optimal size that we can actually act upon and make a decision with respect to that.
Who Needs Nir’s Book?
Charlie Hoehn: Tell me what kind of companies you really want to benefit from the teachings in your book?
Nir Kaldero: Yeah, I think every company’s size can actually benefit from trying to figure out how to deploy and implement artificial intelligence to serve their client, their user better, or even to enhance their operations.
I think that we are in a crucial time in history where incumbent companies should act right now, today, and really transform themselves to become not just data-driven companies but also model driven enterprises.
I truly believe that if the transformation will not happen in the next, let’s say, one to three years, these companies will basically vanish from the world. They will see their market share eaten up.
Competition and speed of change in this forward industrial revolution is really exponential.
Charlie Hoehn: Give an example of how this might play out—one company who doesn’t do anything and one company who does incorporate machine intelligence.
Nir Kaldero: I’m a data scientist, so let me present to you some data. I typically ask executives in the workshop if they know the life expectancy of a Fortune 500 company today. How many years do we expect a Fortune 500 company to survive in the index?
In 1955, the life expectancy of a Fortune 500 company in the United States was 75 years. In 2015, it dropped to 15 years, and today, the life expectancy of a Fortune 500 company is around eight years.
“If change will not happen immediately, these companies will not survive.”
Now, let me give you an idea about the companies that they work with. These companies are Fortune 500 companies, some of them are 185 years old. Just think about it. They employ between 60,000 to 400,000 employees globally. They generate around 240 billion dollars every year. If you think about it, it’s one billion dollars every business day.
Think about how much time and how much effort and how much money needs to go in to this initiative to really transform this gigantic beast.
I truly believe that companies today can transform themselves quickly.
Just because of the large size and globalization, it might take more time. I think we are already in a point where it’s late in the game. Look at companies like Facebook, Tesla—all of these data driven companies are basically growing and eating market share from other companies that have been doing and been in this business for many years.
If incumbent companies will not be able to transform themselves immediately, they will not be able to survive and compete within the competitive landscape of this fourth industrial revolution.
Not Just for Tech Leaders
Charlie Hoehn: Are there any companies that aren’t technically tech companies but are following in the footsteps of these tech giants?
Nir Kaldero: Without disclosing names because I’m working with them, I can tell you that there are many banks in the US that are moving rapidly to become the model driven enterprise that I talk about. There are a lot of healthcare companies in the United States, especially relative to Europe, relative to companies in Japan for example. That are really catching up with these trends. But again, they just started the journey maybe a year, two years or three years ago.
They are gigantic, and the transformation takes time, you know? In general, this initiative of becoming a data driven and model driven enterprise is not an easy one.
“I truly believe that organizations can transform themselves.”
It’s just a matter of how much time it’s going to take them to really survive.
Charlie Hoehn: They need some shepherding, right?
Nir Kaldero: There are no guiding principles or rubric or a playbook, how to become a data driven and a model driven enterprise. What I did in this book is try to give them the guiding principles, definitely not the playbook, of how to start the journey.
Harnessing data to become a data and model driven enterprise is not an easy one.
It takes a lot of effort and change mindset, so I try to give them the habits, the tools, the principles. Hopefully they will make their own with respect to their own culture, their own mission, their own organization goals to make this transformation and journey successful.
Principles of Change Management
Charlie Hoehn: Maybe give a quick overview of some of the principles that you talk about or some of the workflow and change management stuff that you talk about?
Nir Kaldero: Yeah, you know, when I typically talk about AI and data science, it is important to understand the workflow. What is a data science workflow? I typically try to explain to executives, what are the steps that are happening within this data science workflow. Not because I want them to become a data scientist or a scientist in general. I really want them to understand what is the expected contribution and responsibilities within each of the steps that they should know about?
For example, if I look at the overall data science workflow, I think there are four major steps. The first thing is you want to ask the right questions that can be solved by leveraging data. And then you want to acquire the data, then you want to analyze it, and then you want to act on it.
If you really think about it, the act phase is mainly driven by business people, the acquire phase is collaboration between some business people and some technical people, the analyze phase is all about business people like data scientists and machine learners and engineers. Act phase operationalizing all of these problems.
If you really think about the change management required for implementation and operationalizing data science project to realize the ROI, it’s a really a collaboration effort between business and technical people. Which is a completely new thing to the world.
If you think about how business people would make decision in the past, it was mainly driven by executives and senior leaders.
Today, the participation of technical people, especially data scientists and engineers is part of this new change in management. Executives and business people have to start figuring out how to involve them in the process to operationalize all of these projects so the organization can actually realize the ROI behind it.
Working with Nir Kaldero
Charlie Hoehn: What is it like from start to finish when we hire you and work with you?
Nir Kaldero: I try to first and foremost try to explain what are the current and future trends in AI machine intelligence so they understand how the frontier line looks. Then I’m trying to help them figure out why and how they should start transforming their organization to become a data and model driven enterprise. Especially looking at these six principles that have been iterated many times and so companies successfully act on.
And then the second part or maybe the third part here is to talk about the data driven journey and try to demystify what is machine learning. How your organization can benefit from machine learning.
Also, understanding the data science workflow and their contribution and participation within each of the phases within this workflow.
At the end, I typically write a very tailored case study that is highly tailored to their type of business problems and the data they have. I try to showcase them and guide them through this data science workflow so they can understand what is expected from them, how they can develop the mindset, and the critical thinking within each of the steps to make sure that the project is going in the right direction and the organization can really benefit from leveraging data and modeling techniques to basically innovate.
Last Encouragement for Listeners
Charlie Hoehn: Are there any results that you’re particularly proud of or an impact within an organization that you’re really thrilled about that incorporated these learnings about machine intelligence?
Nir Kaldero: Without saying names, look at Fortune 10 companies in the United States. These are gigantic companies. When I started working with them around three years ago, there were some companies that had zero data scientists or maybe one data scientist.
Again, for a company that generates around 200 billion dollars a year, that was a scary moment for me to hear that. And now, after almost two years and they have an army of a hundred data scientists with more than 250 hardcore machine learning projects that have been implemented, and they are benefitting from all of that. I definitely see how these companies evolve and become, not the leaders, right? But they’re now innovative. Now they can actually create solutions and products that will take the organization to the next step.
“Definitely be proud of that.”
Charlie Hoehn: I want to wrap up our interview here with a couple of more questions. The first one is how can our listeners either get in touch or follow you?
Nir Kaldero: You can always feel encouraged to reach out to me. There is a bunch of material on my website, there is a lot of stuff on my LinkedIn that I post pretty often. Feel free to reach out.
There’s also a Facebook page that I try to, with a little time that I have to maintain. And you know, if you are really curious about data science and want to transform your life or your career to become a data scientist, I think Galvanize is a really great place to start the journey.
There is tons of information and tons of processes online as well.
Charlie Hoehn: Yeah, the website is Galvanize’s website or your website?
Nir Kaldero: You can do both. I think on both, we have enough information about how to start the journey to become a data scientist.
Charlie Hoehn: My final question for you Nir is, give our listeners a challenge. What is one thing they can do from your book this week that will have a positive impact?
Nir Kaldero: Yeah, my true hope is that they will read the book and then go to the appendix. In the appendix, there is the data science work sheet that I typically hand to executives. This data science worksheet will guide them through the steps and actions on how to start the journey.
I want everyone to finish within the book and go to their office and enterprises and start leading a change.
The journey is to become a data and model driven enterprises. It’s not easy, but it’s the only path, in a way, to grow and survive in this fourth industrial revolution.