Artificial Intelligence is the future, and the future is here.
Our brilliant & brainy Director of Data Science, Nir Kaldero, led this session at TechCrunch Disrupt NYC where he lays out a framework and tools to help you and your team avoid common mistakes and take advantage of the Cognitive Revolution.
So, for those of you who haven’t got brainwashed from the video that were running around, and for those of you who do not know Galvanize, Galvanize is the learning community for technology. We are the market leader in data science, data engineering, software engineering, education. We have eight beautiful campuses and very vibrant community. I welcome you all to check–
Also our brand-new New York campus, on Hudson Street. I’m also very glad and happy to announce that we officially open our admission for our data-science-immersive boot camp, in New York. So check this out.
So, before we get started– I see there are some people at the back– I have created for you a worksheet that basically will help you achieve two goals. One, it will help you to follow me, talking around 15 minutes. It’s going to be very fast. I can talk about this for eight hours, if you’re interested. We can take it afterwards.
But also the worksheet’s going to stimulate you to think about how you, within your function within the organization, can make an impact right after this talk in the workshop and start the journey to navigate your startup organization to become more data-driven, that will be able to enable and unfold the AI capabilities.
So, this is the worksheet. Some of you have this one. The people at the back, if you have your laptop you can actually quickly download it from this link. There is either the PDF version of it or the Keynote version.
So, it’s time to get serious about artificial intelligence. You know, academia and industry have been talking about artificial intelligence for many years. You know, Shannon wrote the paper in ’49. IBM is doing a lot of research, and IBM Watson win the Jeopardy. Last year, AlphaGo win the Go game.
But, more than that, it is just, with talking about artificial intelligence, the message that I want to deliver you today is that artificial intelligence and machine intelligence is already here and now. And it’s time to get serious about it. And it’s also a very interesting time to think about investing in AI, because there were early adopters in the field, companies like Google, like Facebook, like Tesla, that really adopted machine intelligence very early on. And what we can do today is basically take the investment and figure out what was the ROI on this investment and how this, basically, how machine intelligence helped them grow their organization.
So, AI is everywhere. You know you have a quote, here, from CEO of Microsoft that basically said that we need machine intelligence, and we need artificial intelligence, in order to help us solve the most pressing problems that we have in the world. And, you know, it’s quite simple. We do need this helper brain. We do need this machine intelligence, because of a very simple fact– our brain is very limited.
We are living in an era where information flows to us in a tremendous amount of volume, and we cannot anymore treat the data in a very egalitarian way. Our brain have a very short, limited memory. We take data that is– we recognize data that is with certain colors, with a certain message, and we don’t treat everything in a very equal manner, which the machine can actually perfectly do.
And the problem is that it’s affect our decision-making within our organization. We don’t make any more optimal or even suboptimal decision-making. So we do need AI, and we do need machine intelligence to help us solve the most crucial problems, that obviously all of these problems have a tremendous amount of data around them.
So AI is transforming not just specific industries. It’s everywhere. And if I had more time, I will give you a few examples from all of these industries that we mentioned, here. But the idea is that many industries have accumulated a lot of data that today can be leveraged in order to create some business ROI.
So AI is basically everywhere. Now, I encourage you to think about your industry, about your company, about your startup– how can you actually leverage the data that you have, in order to create some business benefit? I want to show you some data, before I will get into the interesting part of the presentation, which will be give you some tools. I want to show you some data about how big and how vast is the transformation.
So we have 300% year-to-year investment in artificial intelligence raised in 2017. There is around– actually it’s going to be– I believe it’s going to be higher than that– but at least 20% lift in revenue for companies who are leveraging AI capabilities by 2020. And this number is just growing, more and more.
It’s really time to act on it. And the next slide I hope will convince you why. It’s time to act on it. It’s time to start the journey of thinking about, how can I, within my company and within my role, can take and leverage the data that I have, in order to push and move my organization forward, that the organization that basically will be able to survive this fourth Industrial Revolution.
More than a third of businesses will not survive in the next 10 years. How many of you know what is the life expectancy of a Fortune 500 companies, these day? Any idea?
I wish I could hear you. [LAUGH] I will show you, the next slide. Anyway, it’s [INAUDIBLE] to act on it. There is around 140 startups acquired since 2016. Look at the market value. More than $2.5 billion. This is completely transforming entire industries and the way we work.
So, here is the answer. And this, I think, is actually a very interesting slide, and I hope it will stimulate your brain to think, how can you actually survive this new industrial revolution?
So, in the past 15 years, around 52% of Fortune 500 companies have disappeared from the index. If you look at the life expectancy, in 1955, of Fortune 500 companies, it was 75 years. In 2015, it was 15 years. Today, it’s actually less than 9, but I wanted to write “10” just for the sake of it.
But, basically, what it means is that you have to adopt. You have to change. You need to start making the journey. Otherwise, you might not survive in the future.
So, no, we can talk about some example. Think about Kodak, right? Kodak was, like, part of the ’55, right? What happened to Kodak? Start asking yourself, did they make the adoption to the market? Did they listen to their customers? Why they are gone by now?
That’s some questions that you need to ask yourself about your specific organization. So, again, we are at the beginning of the fourth industrial revolution, which is the AI revolution. And what people are not quite sure that they understand is that, we talk about artificial intelligence, but artificial intelligence is not just about talking about what happen in the future. The future is already here and now.
And, if you ask me and many others, leaders in the field, what really sparked this revolution, it was actually, just in 2016, just almost a year ago, when AlphaGo created a very generalized algorithm that, with curated data, basically won the AlphaGo game. Which kind of, like, started– we all started to think about, wow, we actually thought that artificial intelligence and machine learning, all this robotic, all these, like, self-driving car, will arrive in 2020. But what we have discover is that it’s actually here, and it’s actually now.
And we see many application. You know, I show lots of, like, Fortune 500 companies. We meet with a lot of their C-level executives and senior management. I show them robotics. Sometimes they freak out.
But robotics are here. You know, go to Japan airport. Check out Pepper. See how many households in Japan have Pepper, have robotics that interact with them.
Self-driving car. You know, we have it in San Francisco. We have it in Pittsburgh. We have it around the US. You know? It’s already here. It’s already now.
And what I want you to take from here, from these slides, is that you really need to start think about how you can act on it within your organization.
Now, we are a very techie conference, which I love, because I am coming from tech, so let’s talk about innovation. So, today, most innovative companies– and I’m sure you’re all familiar with the companies, here– rely on data-science machine learning, and artificial-intelligence capabilities to drive business process and user experience, to add value. You know, Amazon is a great example that perfected it. It’s basically a gigantic supply-chain company that uses artificial intelligence and machine learning in every piece of this chain.
And they really perfected it. You know, we can talk about it forever, from how to organize the packages in the warehouse, the pricing optimization, to the estimated delivery time, to let driver know how to deliver the package in the most efficient way. All of this is basically artificial intelligence. It’s all about machine learning. It’s all about data science.
But what is more interesting– and, you know, we can speak about content recommendation at Google and some other. But let’s look about the future. Let’s look about what’s next.
What does the future of business innovation look like, 5 or 10 years from now? Future of business innovation has an artificial-intelligence component. And it’s a very core structure of the organization. And there are some companies that did it long time ago– you know, Pandora, 15 years ago, started with the Genome Project. You know? It’s been quite a while.
They completely created the business around recommendation system, about artificial intelligence, about interacting with their users and provide them the best content, based on what they have.
Nest. You know, it’s very complicated to create a thermostat that can actually provide you the best temperature, based on your preferences, just by writing a very long code. Right? Right now, Nest, what did, is basically created a model that is very generalized. You play with Nest, for two weeks and a half. It connects to your phone and router. Every time you get back home, it recognizes you. It collects weather data from your phone and from the internet. And voila– after two weeks and a half, after you’re playing with this basically very generalized model, you get the best temperature, based on your preferences.
So these are some companies that basically positioned machine intelligence and artificial intelligence as the core component of their business. Now, I encourage you again to think about how you, within your startup, within your organization, can start making the transformation. And what I want to impart you is this line of thought.
So, every time– this is quite funny. Every time I talk about artificial intelligence, some people love it. Some people completely hate it. So, [LAUGH] it’s very interesting, especially when you talk to a C-level executive that have been managing companies and organization for many, many years.
So people love artificial intelligence, because they do see and recognize the opportunities and the benefit that it unfold within their organizations. But many people also are afraid of it. When we speak about AI, we speak about robotics, people, like, freak out.
They look at the numbers, here. 36% of people believe that AI will eventually take over or destroy humanity. When I show people robotics, they freak out. Oh my gosh, what is that? Is it going to take my job?
I don’t know if it is going to take your job, but if we speak about the future we do know that machine can replace the human, in many areas, and can take some of the responsibilities that we currently have. It doesn’t mean that the robot or machine intelligence or artificial intelligence will completely eliminate your job or replace your job. It just takes some of the responsibilities.
And I just want to give you one example. Let’s talk about lawyers. Do we have lawyers in the house? Awesome. I feel safe. [LAUGH] So, usually, when I speak about– again, with Fortune 500 companies– there are some lawyers in the C-level suite, and they are all like, oh my gosh!
So let’s speak about lawyers. Natural language processing is doing a fantastic job writing contracts. Why lawyers need to spend so many times within their day writing contracts, if the machine can do it in two hours, one hour, maybe two seconds. Right? It doesn’t mean that the job of a lawyer will not be here 15 years from now. We’ll still need lawyers. But the responsibilities of the lawyers will be different.
Maybe lawyers will not spend so much time writing contract. Maybe they will do negotiation– which the machine is not that good, at this point, at, you know, interacting with humans or with other machines, at this specific moment. So, yes, robots and artificial intelligence will replace some of the responsibilities that we currently have. But we will be focused on some other or new ones.
So, this is our love-hate relationship. Honestly, I do not believe that artificial intelligence will take over or destroy the humanity. And I can give you my arguments, right after this talk, if you’re interested.
But what’s the basis of machine learning, data science, or artificial intelligence? It’s all about data. And I put this slide here because I have been working with lots of startups company that have a serious issue with their data.
So, let me give you an example. Data is the king. Data is, I think, the most expensive commodity in our days. And I have been talking with lots of startup companies that really wants to leverage this AI and machine-learning capabilities. They started the journey a year ago, or even two years ago, but then realized that, wow, shit, I don’t have data for that.
What I want to tell you– and some of you probably are founders or cofounders or CTOs of startup companies. I want to encourage you to think how you can actually start collecting data from day one, even before you launch your product. Because eventually, once you have the product launch, and you have some feedback from users, you have some data, you will be able to rapidly create these AI capabilities and create these machine-learning models that will help you better understand how to interact with your users and eventually will help you create and drive some business opportunities and ROI, back to your companies.
I call it the “startup-data fallacy,” where companies don’t think about how they collect data from day one. And this is something you should really pay attention to and take close to your heart.
So, data is the king. Without data, you cannot, basically, create all of these fancy predictive models. It’s the foundation of everything. And strong data culture is the basis of artificial intelligence. And I will talk about it in just one second.
So, I will not read you this, but most powerful tech CEOs betting big on AI– and for good reasons. Companies here are actually companies who adopted AI very early on. And they can actually quantify today what’s the ROI on the investment.
We should not be afraid of AI. AI will actually help us build and solve some of the most crucial problems that we have in our days, that humanity face. And we do think about– our brain is limited. How much data flowing towards you, every day? Can we still, in our era and days, make an optimal decision to survive? That’s going to be tough without using a machine that will treat everything in a very egalitarian way.
So let me share with you some of my experience working with Fortune 500 companies. What have I learned from working closely with them? So we look in all of these companies, and some other companies that I didn’t put here just because of confidentiality.
But when we look on this organization, we recognize three main pillars, or aspect. All of these organizations have in common strong data culture, strong tech teams, and strong technology infrastructure. And when we speak about how to survive the game, you need to get out of here today and figure out how you can actually start to invest in these three pillars, in order to unfold the benefits from artificial intelligence.
And this is what we’re going to talk about. But, before we jump into the three pillars, you need to make sure that you check V on one big thing. It’s not a small thing. It’s about navigating your organization and start the journey to become a data-driven organization.
It’s not an easy task. That’s the truth. Again, I have been working with many small and large companies. It’s not an easy journey, but I want to give you some tools to basically start thinking about it.
And, if you have the worksheet with you, start thinking about how your organization looks today. What’s the future state that you want your organization to be? And think about your role, as a functional leader, startup founder, cofounder, CTO– how you can actually– what actions you can take, today, right after, by the end of this organization, to make sure to navigate your organization to become data-driven that will be able to unfold all of these opportunities from artificial intelligence.
So, I just want to quickly show you how the distribution look like. I tried to draw a distribution, there. But a distribution is moving rapidly towards the left side, were companies back then, maybe 15 years ago, were data-indifferent. I think small amount of them were data-denial.
But, in recent years, there is a massive shift towards the left side, towards data-driven. Companies basically use data and decisions at the same place and time. Where your organization fits within this framework, within this access? Is it towards the left? Is it towards the right? What you can do today to shift it towards left.
So why it’s so hard, and why [INAUDIBLE] is so hard to become data-driven organization? Well, firstly, data science and machine learning is a top-down decision. It has to start from leaders in your company– basically because they need to navigate the organization to be data-driven, and believe in the amount and believe in the capabilities of data.
But also because it’s required large investment. And you have to get the economics and finance and the money before you actually start applying all of that. But it’s hard, because there is a lot of, like, management problems, in this era, with traditional management.
Lots of managers have a big fear of data. You will not believe how many managers don’t know even how to deal with the overflow amount of information that flow to them. There is a lot of overload of information that creates a lot of fear.
Nature. We do not feel comfortable with numbers. Sometimes, you know, there is a lot of C-level executive and senior management that see all of these numbers, and the predictions from the models, and all the graph, and they don’t know how to interpret it. They don’t know how– they’re not accustomed to basically making decisions right away with data.
Ignorance Again, people aren’t accustomed to solve problem with data, in general. Even if they have the data, sometimes the natural habitat is, like, yeah, I can make the decision to navigate my organization to here or there, because of my gut feeling.
This is not healthy anymore, in our days. People want control, especially managers. Even though data speaks at a different level and different way.
And lack of patience, which is really orthogonal to the way that artificial intelligence and machine learning works. You need patience. Results now will not work, with these models. These models are an iterative process of learning trial. Right?
It’s like the brain. We learn from failure. We learn from experience.
So, we have some barriers. But let’s take a look how management in the big-data era should look like. So, I’m sure we have lots of leaders here. And what we have identified is that companies succeed in this big-data era because their leadership set very clear goals, define what success look like, and ask the right questions.
And all of these three components are basically the core of machine learning and artificial intelligence. You know, as a data scientist I need to figure out how to create a model that predicts success. So what success looks like is a very important metric that I need to get from my leadership team. Clear goals and ask the right questions are very fundamentally important for unlocking the capabilities of AI.
Talent management. It’s a big thing, and we will speak about it today. Data has definitely become cheaper. Also, technology’s becoming more and more inexpensive. But the skills required to interact and work with these technologies are very expensive.
You know, it’s very expensive to hire a really good data scientist. It’s really expensive to hire a software engineer and data engineer, these days. And, for some companies, it’s a real problem, because it also takes a lot of time.
So what we see today– and I will show you some slide– is companies actually growing talent from within and basically using upskilling programs. For example, at Galvanize we provide a lot of, like, upskilling program for companies that can actually use their own talent for new responsibilities and for new job. So, technologies are now available. It’s a well-charted territory. It’s generally quite cheap.
Decision-making. And here, I think, it’s actually– these are the most interesting bullets, here. Effective organizations and leader in this organization. Taking data and decision, and putting these two right at the same place. How many of you– and ask yourself. I will not ask you to say it out loud or to raise your hand. How many of you are really accustomed to make a decision by looking first at data? How many of you don’t distinguish between data and decision? How many of you put decisions of data right to the same place, when you need to make a call?
Company culture is also very important. And this is what we have seen. Companies that are data-driven ask themself “What do I know?” before “What do I think?” When you look about the future, when you want to make decisions, you should ask yourself, these days, within this new revolution, “What do I know?” first, before “What do I think?” And you will be surprised.
There is a lot of Fortune 500 companies that are very far away from perfecting what I’ve just shown you. So don’t think it’s just your organization. I promise you, there are many, many big ones– I will not, obviously, call their name– that haven’t perfected putting data and decisions at the same place, and companies that do not ask “What do we know?” before “What do we think the future might look like?”
So, just to recap that, data-driven decisions tend to be better decisions. And a leader will either embrace this fact or will be replaced by others. And that’s the hard truth.
So, how can you actually start navigating your mind and organization to become data-driven? So let’s speak about the three pillar of artificial intelligence. And here I want to give you some tools. And if you will see, in the worksheet, you have these sections about, where is your company looks today, and where it should be in the future, and what steps you, as a functional leader, can take today in order to make this transition?
So, data culture. It’s very, very important to maintain healthy data culture. You basically need to create an environment where data science, machine learner, software engineers can actually thrive.
So the way to think about it is, we are all– you can think about this as, data scientist or machine learner are like researchers. We love to work in teams. I see companies, they say, yeah, we’ll hire one data scientist. He will probably be there for like six months, and everything will be fantastic.
No. It doesn’t work like that. Data scientist needs to work in team, and we need to get feedback on what we do. When you think about data scientists, think about researchers. And now you can actually create an environment for them to thrive.
The second thing is that you need to get into the habit of data-driven decisions. And we spoke about it. The third thing is, basically, democratize your data. And this is something very important, because I see many companies, even big ones, making this mistake of basically just showing to data scientists or to the data analyst or the BI people the data.
No. You should let every person within your organization touch and see the data. We all, from the barista to the CEO, we all make decisions every day. We can make better decision by looking at the data.
Democratize your data. Make it publicly available to all of your employees. You will see how decision-making is improving.
“Prioritize investment with the highest data ROI.” This is mostly on technical people and senior management, to figure out where you should firstly allocate the money. Where is the data warehousing, where is the modern capabilities, hiring talent, et cetera.
And do not forget– and, again, I saw many large companies that didn’t have a very clear data governance in their organization. This is what we call “creating a healthy data culture.” Now, think about your organization. Are you excelling, applying all of those bullets? How your organization looks today– how you can navigate, to make sure that it looks like that tomorrow.
I do have some questions for you, just to spark your brain and to think a little bit. How can you as a leader, as a functional leader, as a cofounder, as a founder, better embrace data-driven decision-making? What can you do today? What can you do by the end of this workshop?
What are obstacles that doesn’t allow you to do it? And how can those obstacles can be overcome?
All right. Let’s talk about the second big thing. So, I will ask you just to raise your hand, just so I will have some proportions. How many of you have struggled with getting the data that you want for your organization? How many of you struggle with data? All right, there is, like, five, seven people. Very typical.
How many of you have struggled with getting the technology in place? There’s like, three people.
How many of you struggle with hiring technical people within your organization? Way more. So, that’s the big thing. Lots of companies struggle getting and maintaining the data talent.
So I’m sure you all have seen this before. Data science and machine learning is basically a three-legged stool skills. There is a lot of math and stats. There’s lots of computer science. You can think about the subject-matter expertise as having an MBA.
And, if you will find someone that master all of these four components, we will call this guy, a machine learner or a data science, “unicorn,” which is very rare to find. I maybe saw, like, few people in my entire life. You will basically look like that.
But what I want to tell you, here, is that these are four different disciplines. We’re speaking about mathematics, we speak about computer science, we speak about an MBA, we speak about computer science. To master one of those disciplines, it takes a lifetime.
Now, think about mastering all of those together. And figure out what the intersection between them, to create capabilities. That’s almost Mission Impossible.
So what I want you to take from this slide is that, when you’re creating your data talent teams, you need to figure out how to create a heterogeneous team. These people will come from the math and the stats background. These people will come from the computer science. These people will come with some subject matter or business acumen. And basically, together, if you mix their skills all together to one team, this will actually create a very healthy data environment.
So, let’s speak about how to grow talent and attract talent. It’s very expensive. That’s true. What we’ve seen, one of the most significant data point that I will have to share with you today is that many companies around, small one and big one– again, Fortune 500 companies– are coming to Galvanize, so we can actually help them upskilling their talent within their organization.
So instead of hiring, which is very, very costly– and obviously there is lots of risk and uncertainty with hiring people– not speaking about the timeline– companies prefer, these days, to grow talent within their organization. So, what do you need to look for? You need to look for people that have the curiosity and the growth mindset, people that have an interdisciplinary perspective on the business, people that have some technical acumen and business and big-picture view.
And what we see– and I will show you, in a second, the talent map, and how basically the talent map shift, or will shift in the future– this is some predictions that we make. And I encourage you to think about how you can actually grow and identify talent within your company. Now this is the interesting stuff, here.
So, this is the AI, data-science talent map. And this is how it looks currently, today. We have business analyst and data analyst and SQL developer. You have the data scientists or the data engineers, the machine learner.
But what’s the problem in this chart, in this talent map? There’s some really big, actually gigantic, holes within it. What is between the data scientist and the SQL developer? What’s going on between the machine-learner engineer and analytics engineer?
What we have seen, in the past year or so, is that companies are actually creating new roles that fills the gap, creating new responsibilities for a talent team to basically to fill those gaps that we have, here. And this is how, basically, it looks like. Don’t hold me on– you know, the names are changing. You know, just “data scientist,” you can find 150 flavors of it.
But this is what happened recently. SQL developers, companies trying to figure out how they can take their SQL developers and grow the talents, so they can actually become a data-science analyst that better serves the data scientist. Right? Data scientists should not work on a lot of, like, exploratory data analysis or should not, like, clean the data. That’s perfect job and responsibilities for a data-science analyst. So basically the data scientist can actually focus on the hard-core modeling aspect.
We have the data science engineer that, I don’t know, here in the East Coast, but in San Francisco, in the West Coast, there is lot of data-science engineers. There are senior data scientists, now, and machine learner. So what we have seen is that companies actually upskilling their talent, and growing talent within their organization, to fill these gaps.
So, this is current trends. And the truth is that Fortune 500 companies have been working on that already for almost a year. At least that’s the first data point that I have encountered with.
But the question, here, is, how this talent map is going to shift in the next five years. What’s going to happen? What will be the trend? Speak about business innovation, here, how business innovation around talent will change in the future?
So let me tell you. And again, this is what we have seen, collecting data and working with lots of big and small companies in our community. So what we have seen in the past eight months is that we got so many requests for companies to basically upgrade or upskills this lower side of this chart. Basically, companies wants to figure out how they can take their SQL developers, people who are literally touching the data, cleaning the data, and preparing the data for the data scientist, and upskill their skills to be data-science analysts. So they can basically, again, take some of the responsibilities from the data scientist and better serve them, so the data scientists can basically focus on modeling.
So what we have seen is that many jobs or responsibilities have been eliminated, or upskilled, around business analyst, SQL developer, BI. Who is doing BI, these days? BI is pretty much gone, right? 7 years ago, 10 years ago, BI was everywhere.
So this is some prediction how this talent map will grow within this artificial-intelligence revolution. And I can assure you, it’s going to go up and up and even more. So think about how you, within your organization, can basically upskill and grow talent within the people that you have.
So, here’s a little bit about some data teams. Again, data scientists works in team. These are some of the most notable models that organizations are using to create their teams. For example, the centralized one is, you had, like, head of data science, head of machine learning. And basically he prioritize all the investment and what’s the responsibility is going to look like and where we are going.
Embedded team is basically, you break down the talent to be in pods. And basically they just share practices. Maybe they meet for lunch and learn. Maybe they collaborate on some platform, to share best practices, but are pretty much autonomous.
The hub and spoke is a very interesting model, hard to execute but very, very beneficial for organization, where basically data scientists work like consultants. So, there are data scientists who go to the marketing team and act as a consultant. There are data scientists who goes to the operation teams and act as consultants. But eventually, at the end of the day, they all come together, they all share best practices, within one center group or head.
And this is actually a very healthy model. Go back and forth, in terms of learning velocity. It’s very helpful for the organization.
So let’s speak just a little bit on the technology aspect. This is well-trodden territory. I haven’t put– I think this is only 10% of the technology that is available out there. There’s different technology for different tasks. There’s many maps like that. This is just something for you to be familiar with and think about.
Python is still the king. That’s the truth. Python is great language that help us, help data scientists and machine learning, throughout our daily workflow. This is one of the main reasons that it’s still out there.
Obviously, in terms of trend, Scala picking up crazily, exponentially, lately, around big data. Hadoop has been there for quite a while. TensorFlow. If you see adoption to TensorFlow from Google, the number is outrageous.
So, definitely, the future, we have lots of Scala, lots of TensorFlow. We’ll see what’s going to happen with Python after Python. 3. But that’s how it looks like.
Again, most of these technologies are open source. They’re available for you. You don’t need to pay for most of them. This is just something for you to think about.
So, we spoke about the culture. We spoke about the talent. We spoke about the technology. And hopefully you took some notes about where your company– how your company looked today, and how you want it to look in the future, and what three– one, two, three– actions you can take today, in order to make this transformation.
But where to start is a good– it’s a good start, here. Usually, what I do, when I consult with companies and work with Fortune 500 companies, is ask their leadership team, is, what are the top five business problems in your organization? What is the crucial thing that you need to solve, today, in order to make the transition, in order to be better, in order to drive more revenue towards your organization? Ask yourself– what are the top five business problems you have? And write it down.
And then, what you basically need to do is take your business problems and translate them into data problems. So, let’s try to bucket them. And here I give you some of the narrative. Obviously, there is way more models than that.
But try to think about the business problems that you have and how you can actually classify those problems into these groups. Optimization methods. Data science and machine learning and artificial intelligence can help a lot with optimizing pricing, if you’re not quite sure what is the right pricing for the right customer, or reveal the reservation price– what is the price that someone will be willing to pay, without any intervention?
Ad monetization. How you can actually better targeting. We know that, from marketing, better targeting is always better than just, you know, spread the word out there. We are in a personalized era. Everyone wants to see something that is best fit for their preferences.
Recommendation engines and system is very, very vast. Personalized content. You know, this is what Facebook is perfecting. Product recommendation. I am sure most of you have lots of products that you are selling. How you can actually send the message or show people the product that is best fit for their preferences?
If you’re doing something with finance, fraud detection and prevention, lots of companies saving hundreds– actually, it’s more than millions of dollars– but just running fraud detection and prevention model. Just think about Amazon. How much money these artificial-intelligence capabilities saving from them? The amount is so significant and large, which is amazing.
Natural-language processing is growing exponentially. It’s been here for quite a while. If you have some business problems around customer support, or service, or personal assistant, this is becoming also a well-charted territory. And there is a lot of natural-language-processing applications out there that you can immediately apply to your organization.
If you’re thinking about how to better engage or upskill, upselling, there is a lot of churn model. I saw many companies that saved a lot of money by just identify consumers that are about to churn, using churn models. Consumer lifetime value, and on. And the list just go on and on.
But what I encourage you to do is to go back home, right after this talk, and figure out, what are the top five business problems that your organization is facing today, and how you can basically translate these business problems into data problems, so data science, so machine learning, so artificial intelligence can help you unlock the benefit of the data that you currently have. So, you have basically– and usually, again, I deliver a very long, one-day workshop with senior management and executives, called Data Science for Executive workshop– that basically people write, senior management write their data-science strategy. And, by the end of it, they can actually act.
But you have a similar version of it, right here. So here is my recommendation for you. How you can act on your data strategy.
The first thing– and this is the foundation of everything, in this artificial-intelligence era, and machine learning– you have to start thinking how to start the journey to transform your organization to become data-driven. You have to get into the habit of making data and decisions right at the same place. You need to identify business data ROI opportunities, exactly what we did a second ago. Translate your five business problems into data problem, and start investing in those.
Then you need to refine your data strategy. And what I really recommend you is, start small. You do not need to invest millions of dollars, in order to showcase wins and understand what’s the respective ROI behind the investment.
If you are a startup, I’m very well aware that you don’t have millions of dollars, or maybe you cannot even, like, hire a data scientist, you know? I saw many startups that have developers, because they need– and software engineers– because you need to make sure that the website is OK. You need to make sure that your back end collecting the data.
These developers can definitely use technologies out there. And I will just mention Watson, because developers know how to do API integration and press some buttons. They actually know way more than that, but, in terms of artificial intelligence and machine learning, this is something that developers can actually showcase result right away.
So, start with small investment. Start small, showcase wins, quantify the ROI, and then go larger and larger and bigger and bigger, over time. Then, you will have way more data to refine your data strategy, which today you have some seed of it. And this will help you eventually survive in this game.
This is a very, very exciting time, with a lot of opportunities for businesses to grow. And I hope that you will take some of the concepts that I mention here close to your heart, and think how you can actually start making the transformation within your organization, to become more data-driven and more mature when it comes to artificial-intelligence capabilities. And, with that, I thank you all. Thank you so much for everything. I hope you enjoyed it.