We come to the sixth installment in the “AI Zeitgeist” series, and this one is about the changing landscape of artificial intelligence priorities for business leaders.
We’ve covered all sorts of topics in the series, thus far. We’ve talked about the buying readiness of business leaders. We’ve talked about the AI talent landscape. We’ve talked about how people speak about artificial intelligence in the future, and how it won’t always be a buzzword.
But what does all this mean for business leaders, practically speaking, as we go through these three phases of “Emergence,” of “Adoption,” of “Dispersion?”
How are the mindsets of leaders going to change around AI? How are their priorities going to alter as their industry and their business go through these three phases?
(Note: If you haven’t read the first article in the AI Zeitgeist series, you should familiarize yourself with it first. In that article, I explain what the 3 phases of “Emergence,” “Adoption,” and “Dispersion” actually mean).
We’ll begin by exploring business leader priorities in the “Emergence” phase:
Phase 1: Emergence
What is AI and Machine Learning?
The first questions that will need to be answered in the “Emergence” phase is, “what is artificial intelligence and machine learning?” This goes beyond definitions. Here at Emerj, we actually have some quite popular articles on defining these two concepts: What Is Machine Learning? and What is Artificial Intelligence? A lot of these basic kinds of glossary-level articles actually get a lot of traffic because most business leaders just don’t know much about them at all.
In the “Emergence” phase, the fundamental terminology needs to be understood in a conceptual sense.
These definitions, however, have to probe deeper than short explanations. They need to entail a discussion on how AI works and how to apply it to business problems. Whether it’s computer vision, whether it’s speech recognition, whether it’s a recommendation engine—we need data and time to select the right features to build the right kinds of models.
We need some kind of a feedback loop to be able to reform the system and hopefully improve the system over time. We need, oftentimes, quite specialized talent with subject-matter expertise and data science expertise to be able to build and tinker with these models and streamline and harmonize our data.
Understanding all those required moving parts is part of understanding the full definitions of AI and ML as they relate to business. This is the first big first question in the “Emergence” phase. Grasping these full definitions is pretty darn critical for business leaders who are far from AI adoption, but who are trying to get up to speed.
What Kinds of Applications Are Available?
The second question is “what kinds of applications are available?” There’s a lot that goes into this, but we might think about a first facet here being required understanding about what AI can fundamentally do.
So, what kinds of problems can AI solve?
Our article, Where Can Artificial Intelligence be Used in Business?, goes into further detail, but basically, business leaders are looking for a broad understanding of what types of business challenges might be overcome with machine learning. It’s not everything, but broadly speaking, they want to look at a problem and be able to determine whether or not AI might be applicable there or not.
That really does require understanding what AI and machine learning are, but it also requires seeing a good deal of use cases and applications in the real world.
When we go in and do executive education work, such as business strategy work with the board of a company or with a large intergovernmental organization that’s working on understanding AI, the educational agenda almost certainly is going to involve a conceptual grasp of fundamentally what kinds of problems can AI solve. It’ll also involve a very tangible grasp of half a dozen very specific representative use cases that make those concepts click, that allow people to say, “Ah, that’s how much data might be required” or “Ah, that’s a problem that could be optimized with AI.” A combination of both is almost always necessary.
The other thing that fits under this idea of what kinds of applications are available is seeing enough use cases and having enough of a conceptual grasp of AI to get a sense of what we call the “capability space of artificial intelligence.” In other words, what function can AI serve within a given sector?
Business leaders can get a basic understanding of the capability space of Ai within a sector by having a grasp of a few dozen use cases from large companies and offerings from different vendors that have successfully sold their solution to companies within their sector.
This can give business leaders a felt sense of what pockets of their business are already being augmented and automated. Company leaders want to have a firm sense of where AI is making traction in different areas of their business, as well as where it’s not being applied and why not. This is really important, and oftentimes this involves looking at what the biggest players in their sector are doing.
Phase 2: Adoption
What do I Need for Talent?
In the adoption phase, new priorities arise when business leaders start thinking practically about applying artificial intelligence. Oftentimes, in companies that are not ready to leave the “Emergence” phase, they ask “Well, what vendor can we work with and how can we use AI?” They think that the adoption phase means asking, “How can we start using AI?”
In fact, that’s the wrong question.
That leads to what we call “toy applications,” and they result from people who want to use AI for AI’s sake. We see this in big banks and large brick-and-mortar retailers that are not really hip to very much digital transformation, nevermind artificial intelligence. They go in thinking that they’re ready to adopt AI; they know what AI is, and now it’s time to say, “Where can we use AI?” That’s almost always the wrong question.
When business leaders understand the “Emergence” phase of AI, then they know and understand the requirements of AI in terms of data, in terms of talent. They’re really only going to ask, “Who can I work with? How can I bring on AI?” once they have the required resources, talent, and data, and once they know what problems are going to make sense to solve with AI.
Oftentimes, AI is not the only tool for the job. Even in an area like anti-money laundering and banking, where machine learning has lots of potential advantage, there are solutions and methods and processes for anti-money laundering that do not involve machine learning.
In fact, we could argue that most of the anti-money laundering efforts in the world right now involve no machine learning whatsoever. There are some banks that probably don’t have the talent or the data to be able to successfully leverage those technologies or maybe even the budgets to work with some of the best vendors in that space.
Once we enter “Adoption,” business leaders start thinking about the real challenges and the real potential opportunities with AI, and that often begins with talent.
When business leaders know what AI is, they realize they can’t get a bunch of MBAs or HTML- and C++ programmers and tell them to go build advanced algorithms for machine learning. It’s just not going to work. That said, they’re also not simply thinking about hiring very smart AI PhDs or folks with a physics background that maybe worked at Amazon.
It’s not just about hiring AI leaders; it’s also about hiring data scientists and data engineers and maybe contractors and consultants. Business leaders ask, “What are the right kinds of projects that not only involve our own core team but also might involve bringing in outside help?”
They’ll also ask who amongst their technical team they can invest in with regards to their education. Maybe some of them could build themselves all the way up to be machine learning engineers. Maybe some of them cannot, but they are able to, at least, speak the language of data, speak with the AI talent that they’ve hired.
Another very important part of the mix for companies that are self-aware, particularly for enterprises that have been around for 40, 50 years or more is culture.
When we look at the banking space, for example, they have the money to buy whatever talent they want, but a culture of innovation and iteration is a real challenge for them when it comes to adopting AI. The culture of data science is something we talk about extensively in our executive guide on why Google and Facebook have a great advantage over even the largest enterprises when it comes to hiring AI talent.
What do I Need for Data and Data Infrastructure?
Business leaders have to think about data in the “Adoption” phase. Sometimes they might even think about data before talent. For example, an eCommerce is probably already digitally savvy, but they might start asking themselves, “What kinds of click stream data do we have? What kinds of purchase history data do we have? What kinds of data do we have about email promotions and does that map on top of the way we keep our customer records for onsite purchases?”
They’ll start thinking about the way they’re tracking data, how much they’re tracking, where they’re storing it, how easy it is to access it, how to merge it and purge it, how to layer it on top of each other, and how easy it is to manipulate and to use to potentially train models.
They’ll ask, “Do we want to train models to take the onsite behavior of the user and use that to prompt email or SMS messages that are going to get them to buy a product on our website?” Maybe they’ll take past purchase behavior and correlate that to who they think is most likely to buy a gift card around the holidays.
Essentially, business leaders are going to need to think about what their business needs are, and whether or not those business needs are represented well in their data.
You’ll notice we really can’t be thinking about data or data infrastructure unless we have a firm grasp of what we needed to learn in the “Emergence phase.” Namely the kinds of business problems that AI could fundamentally solve and the requirements for implementing it.
Certainly, data and data infrastructure concerns would involve thinking about how to collect data. It also might be more specific, such as, “How can we go crowdsource some important kinds of data to layer on top of the data that we have already?” Or “How can we buy third party data that we can maybe merge into the data that we have to enrich it to make it more valuable?”
Which Initiative Should I Pick?
The last question in adoption is, “Which initiatives should I pick?” Notice I didn’t say, “Where do I apply AI?” It’s really not about AI in and of itself; it’s about knowing what we need to do to move the business forward, and sometimes, not always, artificial intelligence is the right tool for the job.
What this involves is combining an understanding of the “capability space of AI” that we talked about and meshing that with our business priorities to see where they overlap.
Sometimes businesses have a really critical priority. They might realize that when they stack the potential solutions available to them side by side, a lot of the AI solutions really don’t make sense for them. They don’t have enough data or the solutions aren’t really mature, whatever the case may be. As a result, they don’t choose AI for that priority.
Then there’s going to be other situations where that “capability space of AI” overlaps with their business priority so well that it’s quite clearly the right tool for the job. In areas like fraud and money laundering, it’s pretty evident that as time goes by tools that aren’t leveraging machine learning will be pretty much obsolete in many of those spaces, particularly for large companies. In other areas, that may not be the case.
Picking the right initiatives, whether it’s AI or not, being able to inform their technology acquisition, their technology strategy, with a knowledge of AI, is going to let business leaders use it when it makes sense and not use it when it doesn’t make sense.
Then they can use their data and talent most efficiently. They can be most likely to garner a return on investment.
Phase 3: Dispersion
The “Dispersion” phase is a hypothetical future point where artificial intelligence is no longer really seen as something special; it is seen as just a natural part of tech. In “Dispersion,” it’s assumed that all the major softwares involve AI in some way.
There’s really no sector in the world at this time where AI is dispersed. The vast majority of AI applications and vendor company’s internal solutions are nascent; they’re maybe in pilot mode.
But once we get to that hypothetical point, what kinds of questions will leaders be asking?
What New Capabilities Exist in Tech?
Ultimately, their priorities will pivot away from thinking about artificial intelligence, specifically, or even thinking about data and data infrastructure, specifically. That will be normal—it’s not like they’re going to stop thinking about it—it just won’t be as top of mind as maybe it is for companies in the “Adoption” phase. It’s really going to be about the new capabilities that exist in tech broadly.
The “Dispersion” phase is when we stop thinking about AI as its own thing, and all business leaders really pay attention to is the expanding capability space of technology. They don’t think about differentiating AI from blockchain or VR or whatever tech medium; they just sort of see an expanding space of what technology can do.
In the “Dispersion” phase, there’ll be less hubbub about AI specifically and more executives asking, “What are the new things that can be automated?” Or “What are the new things that I might be able to do within my business?”
This is inherent to the “Dispersion” phase. While, again, AI is a big part of this, it’s certainly not all of it. There will be new mediums of experiencing content and of doing work. Eventually, maybe at some point in the next 10 or 15 years, people will not just entertain themselves in virtual reality; they will actually work in VR workspaces. This will allow them to interact with software in entirely new ways, and so that doesn’t even necessarily have to involve AI.
How Can I be More Efficient and More Agile?
Companies will also ask, “How can I be more efficient and more agile?” By efficient, I mean they’ll have less committed resources and overhead and more ability to adapt and change. If we think about an efficient and agile company, we might think about the classic examples of Uber or Airbnb, where we don’t really have to own the resources, but we can still make the money from all those assets, even though we don’t necessarily own them.
For example, banks will have a lot of what they currently do in-house drastically augmented or potentially almost entirely automated. Whether it comes to loans and lending, detecting fraud, or a variety of white-collar processes, much of this is going to have to be automated.
Being more efficient and more agile is going to be a preeminent concern. By the time AI is dispersed, there’s going to be tremendous pressure to make sure that we don’t have any extra headcount where it doesn’t belong or to make sure that we’re not committed in any one direction that’s going to keep us from pivoting and from changing.
The reason that this matters is because as the dynamics alter, companies are going to come to the understanding, as many of them do already, about how much quicker they’re going to have to move and adjust and adapt to where the value is in the business. When the technology changes, they’re going to have to capture profit in new and different ways, bend, and alter the business model. This has to do with competitiveness.
In our own research, we look at something called disruptiveness. In other words, how likely one AI capability is going to determine the winners and losers of a sector within the next five years compared to another.
We think this is a pivotal metric because eventually, it will be really critical for a business leader to know which capabilities are going to help them win. When everybody understands AI, it’s going to be all the more important to not just use it where they think they should use it, but use it in the areas that are going to win market share and determine their survival in the future.
Competitiveness is actually going to be episode seven of seven in the AI Zeitgeist series, so I’m not going to go any longer on it here.
To conclude, the shift from”Emergence” to “Adoption” is relatively simple and can happen in almost any sector. Business leaders can quite quickly move through the “Emergence” phase by answering those questions, understanding those principles, garnering a full grasp of that “capability space,” and then being prepared to have conversations about talent, about data, and about picking initiatives.
Business leaders can then ask if their top talent can be brought up to speed on artificial intelligence, moving from “Emergence” to “Adoption.”
The move to dispersion is going to involve a lot more factors coming together. It’s going to involve the maturity of the landscape of AI applications, which might be slower or faster than we think. It’s going to have to do with the maturity of the science.
There might be more efficient, more effective types of machine learning. There might be a new paradigm beyond deep learning that we enter into that opens up new capabilities, that goes in a different direction, and whatever those ramifications may be.
“Dispersion” is going to involve the maturity of the solutions in the market and the maturity of the science. It is safe to say that the first sectors to enter dispersion will be sectors that have, generally, very digitally savvy cultures. And there will be sectors where the winners, the dominant companies, are really indeed predicated on technology, predicated on staying on the cutting edge of technology.
For business leaders, understanding consciously what it takes to fully shift from “Emergence” to “Adoption” as a company culture can be really powerful and really useful.
Readers who haven’t read the other articles in the “AI Zeitgeist” series can navigate to them below:
- The 3 Phases of AI in the Enterprise: Emergence, Adoption, and Dispersion (AI Zeitgeist 1)
- How “AI” Will be Discussed in the Future (AI Zeitgeist 2)
- The Evolution of AI Talent and Training (AI Zeitgeist 3)
- The Increased Accessibility of AI in Business (AI Zeitgeist 4)
- Buying and Adoption Readiness for AI (AI Zeitgeist 5)
- The Changing Landscape of AI Priorities of Business Leaders (AI Zeitgeist 6) <— You are here.
- The Competitive Dynamics of AI – Now and in the Future (AI Zeitgeist 7)
Our last article (7 of 7) in the series episode is going to be about the competitive dynamics of AI in the future. In other words, how are companies competing now, and how will they compete when AI becomes more and more prevalent in all of our technology tools?
This is really critical for thinking about strategy today. The way companies do that now is actually quite important, but the way they will do that five years from now is likely to be very different. We’re going to try to articulate that in enough depth for business leaders to really be able to have a sound understanding and be better prepared for a more AI-immersed future.