SDS 240: State of Artificial Intelligence in Business

Welcome back to the SuperDataScience podcast, ladies and gentlemen, welcome to episode number 240. Super excited, because today, we’ve got some interesting research that we’re going to talk through, some interesting findings, and we’re going to talk about artificial intelligence. Before we start, though, I wanted to say you might hear some noises in the background. I’m standing here at the beach in the evening, just wanted to go for a walk in Gold Coast, Australia. You might hear some waves, because we’ve got Cyclone Oma coming through Queensland at the moment, so waves are crazy. It’s actually pretty hectic. Haven’t seen the ocean like this in a while. Anyway, apart from that, to be pretty clear, no distractions, so we can dive straight into it.

What is currently going on in artificial intelligence in business at this moment? So right, now it is still February 2018, the very end of February 2000 … Oh, ’19. End of February 2019. To what extent has artificial intelligence been adopted in business? What are some of the great benefits, what are some of the drawbacks, some of the setbacks, roadblocks, and so on? We’re going to be talking about all those things, and where are we getting the data? This is important. Where is all this information coming from? I sat down and did some research into this very interesting topic, and I found three main reports that we’re going to be getting our insights from. These are, reports are very reputable sources. Two of these companies that did these studies are consulting organizations. You may have heard of them. Probably you have.

One of them is PwC, PricewaterhouseCoopers, and the second one is McKinsey, both billion-dollar companies, massive, massive companies, operations around the globe. The third company that we’re going to be getting our insights from is RELX, a massive company. RELX Group is a British multinational, information and analytics company headquartered in London. Again, another billion-dollar company. All of them are very reputable, and they did their research. The interesting thing is that their findings, why I liked looking at all three reports is that their findings agree. If you’d like to get these reports, you can get the links of course at the show notes at www.www.superdatascience.com/240 if you want to follow along or if you want to reference them in the future.

With that in mind, let’s kick it off. First things first, what is artificial intelligence? What are we going to be meaning by artificial intelligence when we’re referring to? Because there can be a lot of confusion around, what is AI? What is deep learning? What is reinforcement learning, machine learning, and data science? All these technologies can get mixed up, so for the purposes of this podcast, and in general, the definition of artificial intelligence that I stick to is the most broad out of all of the items just mentioned, that artificial intelligence actually encompasses in itself machine learning and a lot of other things, and then machine learning encompasses deep learning as one of the subbranches.

It’s very important to keep that in mind that artificial intelligence isn’t the same as deep learning. In fact, artificial intelligence doesn’t have to always be even machine learning level. It can be something much more simple, like and as a very simple if-else rule-based technology can be artificial intelligence, or is considered artificial intelligence under this definition. For instance, RPA, robotics process automation, is something that falls under artificial intelligence. However, it is not machine learning or not necessarily does it have to include machine learning. Most of the time, it’s much simpler than machine learning. Something just to keep in mind that artificial intelligence is a very broad term and is not necessarily just deep learning and neural networks.

Okay, so what do our studies say about artificial intelligence and what, the state that it’s gotten to in business up until now in 2019, February, almost March, 2019? One of the things that the studies agree on is that most executives know that artificial intelligence has the power to change almost everything about the way they do business. And artificial intelligence could contribute to up to 15.7 trillion to the global economy by 2030. That’s 15.7 trillion with a T can be contributed by artificial intelligence to the global economy by 2030, which is not that far away. It’s 21 years away, less than 21 years away. That, just to put it in perspective, 15.7 trillion is more than the combined current GDP of China and India together, so if you put the whole output of China and India together, it would be less than that number, so as you can see, it’s a massive, massive contribution that artificial intelligence can make to the global economy.

Nobody’s saying that that is exactly what will happen, but what we can see is, the potential’s there. In fact, it might even be more due to the exponential nature of artificial intelligence. It’s really hard to predict what’s going to happen in the next 10 years, but even now, experts are seeing this happen.

What is going on in terms of adoption? What have executives, so all these three firms have surveyed, have studied responses of executives specifically in organizations, and have come to these insights, conclusions that we’re going to be talking about, so we’re going to be talking about lots of facts, lots of percentages, lost of insights from these surveys, from these analysis and these surveys are definitely statistically significant. If you go on the websites and actually go through these surveys, you will see that thousands of executives were surveyed and, like in the case of McKinsey, I think it was like 2,300, or over 2,000 executives participated in the survey, so definitely statistically significant findings that they have here.

Let’s go. One of the findings, a very interesting one, is that 88% of senior executives agree that AI and machine learning will help their businesses be more competitive. That’s from the RELX findings, from the RELX study, so 88%, majority, vast majority, almost all senior executives agree that AI, machine learning will bring benefit to the businesses. 47% of business executives say that their companies have already embedded at least one AI capability in their business processes, and that is from the McKinsey study. 47%, that’s half of the executives surveyed say, or we can even say half the companies, say that AI is embedded in at least one of the business process. Actually, 21% of them say their organizations have embedded AI in several parts of the business. 47 say that AI is embedded in at least one. 21% say that, in two or more processes.

Important here to, again, remember that AI is not just deep learning. This doesn’t mean that half the companies in the world have neural networks in their businesses. No. It means some sort of, some form of artificial intelligence, some form of system that can, intelligent system that can make decisions is embedded, so some companies might be just starting out with very simple forms of artificial intelligence. Some could be taking it to the next level of RPA, robotics process automation. Some might be already embedding machine learning, and some might be using deep learning and neural network technology and things like that, but the point stands that businesses are seeing the value of this and are going in that direction.

Next finding is that 30%, so we’ve talked about companies that already have AI in one business process and maybe 21% have in multiple business processes. The next one is, 30% of executives say that they are piloting AI, so in one way or another, they’re also, so they are either have it already and they’re piloting it, or they’re just piloting it to see how it’s going to go in the organization, so 30% of executives are already piloting artificial intelligence.

Those numbers should kind of demonstrate that executives are really thinking about this, and it’s not a surprise, because we live in a capitalistic world, and inevitably, market pressures, competitive pressures, are going to enforce artificial, are going to help those companies that are, so companies that are using artificial intelligence are going to become more efficient. They’re going to cut costs. They’re going to innovate better, faster, be stronger, and through competitive pressure, such competitive pressures, all businesses will eventually adopt artificial intelligence, or they will really see their market share be taken away from them, and it will be hard for them to stay competitive.

Another interesting finding is that, how is artificial intelligence used? Well, right now, according to RELX, the three most popular uses of AI and machine learning are to increase efficiencies or worker productivity, so 51% of the companies reported AI to be used in that way, increasing efficiencies or worker productivity. Second one was to inform future business decisions, 41%, and the third one was to streamline business processes. 39% of companies reported using AI for that purpose.

Another finding, which I found very exciting, was that according to PwC, 20% of business executives say their companies will deploy artificial intelligence across the business in 2019, very deep finding or very deep insight that is happening now, is happening, all this is happening now. 2019 is going to be a year where a lot of businesses are getting on board. This illustrates the exponential nature of this technology that it’s not one of those things that we hear about it and it’s going to happen 10 years, 20 years from now. It’s happening now. We live in a world that is so fast-paced, so fast-moving.

Like I don’t know, 2010, we were just starting hearing talks about self-driving cars and things like that, and there was maybe, like still in inception mode, now we already have self-driving Ubers in some states of the U.S. How crazy is that? Again, 2018 is when it’s happening for many, many businesses.

The next one is that 55%, this is about government. 55% of government officials say they are aware of AI, but it is not being utilized. 37% say that they are utilizing AI, so even governments are getting into this space, and we can see that from examples of smart cities such as what San Diego is trying to do or is aiming to do to build a smart city, and other cities around the world. Not just cities. Any government function is slowly starting to, is seeing the value of AI and is slowly starting to implement that, as we can see from these numbers.

There we go. Those are the adoption rates and how executives and senior management are seeing artificial intelligence, the role of AI and what they’re doing about it. Now, what are some of the challenges associated with adopting artificial intelligence? What is happening in this space? Why is it not as easy to jump onto this bandwagon of artificial intelligence and turn the key, and that’s it, you can use artificial intelligence? Well, there’s quite a few challenges associated with implementing such a innovative technology, such a fast-paced thing that something that hasn’t been around for that long, how do companies get on board with that?

From what I read in these studies, I identified four main challenges, four top challenges, associated with artificial intelligence adoption. Here we go. Top challenge number one is strategy. 43% of executives cite lack of a clear artificial intelligence strategy as the major roadblock for them adopting AI in the business. That’s no surprise, because it is such a new space. It is so hard to see how artificial intelligence can benefit your business. There’s lots of moving parts associated with artificial intelligence.

The data, where do we get the data? How do we … Is our data prepared? Is it in the right state? How do, what tools are we going to use for artificial intelligence? Are we going to go open source? Are we going to go commercial software? How do we protect our data? How do we ensure privacy? What are we going to do with the algorithms? What about the intellectual property of the things that we develop? Is it safe to deploy these algorithms? Are they explainable? How do we deploy these algorithms? In what order do we solve the challenges that a business faces? How do we build a team? How do we build an artificial intelligence team? Do we build a team right away? Is it going to be a siloed team? Is it going to be integrated within the business with separate artificial intelligence experts in every division?

There’s a lot of things that are involved in artificial intelligence strategy, and a lot of ways to approach this problem, and, of course, without having prior experience or any kind of frameworks to guide executives and businesses, a lot of businesses or executives see this kind of like as a, like poking in the dark or trying to come up with an approach and rather than just having a standardized way of how this can be done, for instance, like accounting or finance, things that have been around for decades or even centuries that are so crystallized and can be, there are certain standard practices that businesses can follow. Artificial intelligence isn’t like that yet. Part of that is what makes it so exciting, actually.

One of the top challenges or top challenge number two is talent. 42% of executives point to a lack of talent, of AI talent, according to McKinsey. Very interesting, because it’s very controversial. At the same time, what we’re seeing in the marketplace, in the job marketplace, in the talent marketplace, is that there’s actually plenty of talent. There’s lots of people who are capable of creating artificial intelligence algorithms, building machine learning software or algorithms and models, doing data science, doing analytics, writing deep learning scripts and code. There’s plenty of people like that. If you just go on Kaggle, there’s over a million people registered on Kaggle, participants on Kaggle. There’s even in SuperDataScience on our Udemy courses, we have over 700,000 students.

There’s lots of evidence to suggest that there’s plenty of people who are capable of the, who have these skills, and yet talent is one of the top cited problems, and that comes from something I recall the data science gap. Here at SuperDataScience, we have this term, the data science gap, which stands for the mismatch of, or kind of like the, yeah, the mismatch of expectations and what is actually required for the job.

Because it’s such a new field, artificial intelligence, data science, often time recruiting managers and hiring managers, they don’t, they can’t really articulate what is exactly required, especially when there’s no strategy around artificial intelligence. How can you articulate what exactly your needs are, and therefore it seems as there’s a lack of talent? However, as we’ll see just now in a few minutes, in this podcast, once you sort out the strategy problem, the talent problem is actually quite straightforward. It’s no longer a problem once you know what you’re looking for and you know what you want, what your business needs.

Top challenge number three out of four is opportunities. Only 17% of organizations have mapped out their AI opportunities, so this is an important thing. We’ll talk about this in a second. Basically, only 17% of organizations know how artificial intelligence can be beneficial to their business, and we’re talking about things like required level of investment, difficulty of implementation, and potential value at stake, but again, we’ll get to that in a second.

Top challenge number four for adopting artificial intelligence is budget, so 50 … This is where the studies kind of conflict a little bit. 58% of businesses say less than one-tenth of their company’s digital budgets goes towards artificial intelligence, and 71% expect AI investments will increase in the coming years. That’s going to McKinsey, and according to RELX, the numbers are a bit different, or the number is a bit different. RELX states that only 18% of senior leaders plan to increase investment in AI and machine learning. On one hand, you have 71% expect to increase budget, according to McKinsey. RELX says only 18% expect to increase investment in AI. Whatever the case is, the reality is that budget is still a constraint.

How, as an executive, if you’re an executive listening to this, how can you solve these challenges? How can you overcome these roadblocks in adoption of artificial intelligence in order to reap the benefits and be on the train of artificial intelligence, be in that list of companies that is rapidly adopting AI and not fall behind and get those benefits that, or be part of that group that we talked about at the beginning, where we were talking about the 88% of executives that see the value in AI, 47% of businesses that already have AI in at least one business process. 30% that are piloting AI. 20% of businesses that will deploy AI across the whole business in 2019. Things like that.

How can you overcome these four challenges? Here’s a potential approach that you can use. There’s probably lots of different approaches, but this is some of the things that I would do. If I were a business executive and I had these challenges ahead of me, how would I solve these problems? Well, to start off with, I would go for the challenge with the challenge number three, opportunities. Right now, we see that only 17%, again, this is according to McKinsey, only 17% of organizations have mapped out their AI opportunities. That would be my starting point to map out your AI opportunities, whether to do it yourself internally, whether to hire some consultants to come in and actually look at the business and see where their potential opportunity is.

According to Peter Drucker, the father of modern consulting, there are only two things in business. Every business is a combination of two things. That is marketing and innovation, or innovation and marketing. Everything else is a cost. So basically, the question is, how can artificial intelligence help with your marketing? How can artificial intelligence help with your innovation, whether it’s products, services, processes, how you work with your employees, culture, and lots and lots of things, and how can artificial intelligence help with your costs, reducing your costs? Those are the three main components. That would be my starting point. All right, how am I going to think about artificial intelligence, and how, what can it help you?

Of course, it’s necessary to know some things about artificial intelligence, so it’s a matter of education, self-education, getting to know what artificial intelligence is. What is it capable of? Maybe reading some case studies and listening to podcasts like this, understanding what artificial intelligence is capable of and then seeing how that can be applied in your business, or, of course, alternatively, you can just hire some consultants to come in and do that work for you, but nevertheless, that step needs to be done, because without that, it’s really hard to just go and start thinking about strategy, the team, the tools, all of those things, they actually flow out from here.

Once you’ve mapped out your artificial intelligence opportunities, the next step would be to build a strategy. 

We’re not talking about a comprehensive, full-fledged strategy for artificial intelligence for the next five years, like is common to be done in executive teams and board meetings. We’re talking about a strategy to get started with artificial intelligence, because all you need to do is, you need to turn your map of artificial intelligence opportunities into a roadmap to proving success of AI. Here, I would talk about finding out of your opportunities, doing a standard kind of like putting them on a scatterplot, basically creating two axes, how much value you will get out of this opportunity on the Y axes, and how difficult is it to implement on the X axes.

Finding the most, the easiest win, the quick wins, what is the quickest win that is going to bring the maximum value, or to be concrete with this, the low-hanging fruit which can be described as easy wins that require relatively little investment, I would say, up to $100,000, are easy to implement, take under six months to roll out and actually put into production in the business, and have massive potential. In some cases, you can get 10X, even more, 100X. That’s much rarer, but 10X is probably doable with artificial intelligence projects. There’s a good chance you can find a project with a 10X ROI in your business with artificial intelligence and put that into play. That would be my starting strategy.

Once you have one of those projects, you roll it out, 3-6 months, less than $100,000, you get a 10X ROI, that is your proof of concept that artificial intelligence can actually work, can benefit your business. It’s not just the pilot pilot. It’s a pilot that generates revenue that you can put towards more artificial intelligence initiatives. Rather than thinking out a long-term 10, 5, 3-year strategy for artificial intelligence, I would get started like this, like Google does, like with their skunk work principle, where you just, like you fail fast. It means you fail, you fail fast, you get up, and you go again. You get up, and you go again. Being more agile about this whole process.

Then once you have that one first successful project proof of concept and budget that’s coming in, that’s when you start thinking more concretely about artificial intelligence strategy once you have the buy-in. This is very important to have the buy-in from your organization to start building this culture where data-driven decisions, and artificial intelligence, and other technologies such as machine learning are accepted and people are curious about it. People can see that this can really help them in the work, these can help the business.

Once you have that, then you start thinking about the artificial intelligence strategy in the big picture sense, in terms of the comprehensive strategy, which involves all those questions we talked about, like do we have the data? What state is the data in? What can we do about it? Should we build a data link? Do we need Big Data technologies? Do we not need Big Data technology, we need to buy data somewhere else? What kind of tools, or how do we break down data siloes that we might be seeing in the organization? What kind of tools do we need to adopt? Is this going to be open-source? Is it going to be commercial software?

Are we going to be on-premise, or are we going to be in the cloud? When and how are we going to build our artificial intelligence team? Is it going to be one person? Is it going to be multiple people? Are they going to be a functional AI team? Is it going to be a functional machine learning team? Are they going to be sitting within different branch or different functions of the organization, existing functions, such as IT, marketing, support, operations, or is it going to be separate? Lots of those questions are going to come up, but they need to be answered in due course at the right time when, once the first project’s out and the organization is seeing value of artificial intelligence.

That answers some of the problems that we were talking about. We already talked about opportunities. That answers the strategy problem. That answers the talent problem as well, because now you’re thinking about AI talent when the time is right, when it’s time to think about AI talent. Finally, that also addresses the budget concerns, so if you start off a quick win project which has a massive ROI or has a potential to have massive ROI, then all of a sudden, you can actually subsidize to a level those investments in artificial intelligence. It can be a self-fulfilling prophecy. Of course, may, not always that first project will work out. Maybe you’ll need to do couple of those first projects, first initiatives, but once you hit it, like data is a gold mine for any organization. Your own data is a gold mine for you, so once you strike gold, that’s when you can continue digging and getting to the bottom.

Also, it’s important to make it clear to the, whoever’s doing your artificial intelligence work, that this is not just a pilot. You’re not just interested in creating results like finding insights or building AI for the sake of building AI, because it’s hype, but the point of all of this is the bottom line of the business. It’s very important to make it clear to the relevant people that the bottom line of the business is important and that there has to be return on investment. That is part of this approach.

There we go. That’s where artificial intelligence stands. I probably should have said at the start of the podcast this is geared more towards executives and business owners, but regardless of your level, I hope this was useful to you. If you have a manager or executive that you know, forward them this episode and help share this knowledge. Spread the word with them. A call to action for all the executives who are listening is to start by mapping out the artificial intelligence potential in your business, so what are the potential opportunities for artificial intelligence in your business?

Very good exercise, even if you’re already doing AI to some extent, as we could see, 47% of businesses are already doing AI, but only 21% of businesses are doing AI in more than one business process. Basically, there’s a high chance that you fall into that, not into that 21%, but into the remaining 79, so perhaps you’re not using AI across your whole organization, and that is good starting point to map out, what are potential opportunities for artificial intelligence? As we discussed, one of the ways to go about it, and not how I’d go about it, is, think about Peter Drucker’s advice.

Business is innovation and marketing. The rest are costs, so why not break it down into those three categories and see, “All right. In terms of marketing, in terms of innovation, in terms of costs, how can artificial intelligence be applied in my business? What are some of the case studies that exist in artificial intelligence in my industry, in my type of business, or in adjacent industries?”

Well, I’ll finish all this podcast by saying that as you probably have seen, if you’re following the Data Science Insider or if you’re just reading our email list or like our emails that we send out, perhaps you’ve seen in other places. Hadelin and I, Hadelin, my business partner who you probably already know, we started a consulting startup, consulting business in the space of artificial intelligence, about a year ago in February 2018. We’ve been working on some interesting, exciting business challenges since then, and now, we’re finally excited to announce that we’re open to helping more businesses, helping more clients.

If you are interested in working directly with us, if you’re interested in getting our view on your business, on your current situation, your state of, the state of artificial intelligence in your business and how it can be improved or how it can be enhanced, how artificial intelligence can help, what are the opportunities in your business, then do let us know. We’d love to hear from you. We love working on interesting challenges.

Right now, we’re looking to take on two or three more projects in addition to our current portfolio. If you’re interested, the best place to contact us, to submit this inquiry, is at the company’s website. Company’s called Bluelife AI. Website is www. Bluelife, one word, .ai. That’s bluelife.ai. There’s an inquiry button, inquiry form there. We’d love to hear from you. Love to help you out with your business challenges and get you onto this artificial intelligence train and skyrocket your business.

On that note, hope you enjoyed today’s podcast and got a few valuable insights from there. All the show notes, as usual, at www.superdatascience.com/240. Bluelife’s at bluelife.ai, and I look forward to seeing you back here next time. Until then, happy analyzing.