We’ve made it to article seven of seven in this “AI Zeitgeist” series. It’s been a while building up to this, and I’ve kept the competitive dynamics of AI as the topic of this seventh article because to me everything builds up to this.
Everything we do here at Emerj is always about who’s going to win in the market in the future, and so we need to give an inordinate amount of thought to this topic.
(Note: If you haven’t read the first article in the AI Zeitgeist series [linked below], you should familiarize yourself with it first. In that article, I explain what the 3 phases of “Emergence,” “Adoption,” and “Dispersion” actually mean).
For those of you who may not have read the previous installments in the “Zeitgeist” series and may be interested in exploring the topics we cover in this article more in-depth, you 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)
- The Competitive Dynamics of AI – Now and in the Future (AI Zeitgeist 7) <— You are here.
We’re really going to sink our teeth into this. This is what it’s all been about for me, and we’re going to kick things off with how the competitive dynamics are going to rear their heads in the “Emergence phase:”
Phase 1 – Emergence
Roadmap of Readiness
Companies are going to compete to some degree, even before they adopt AI, on an understanding of a “roadmap of readiness.” The companies that prime themselves to leverage AI when the technology becomes available will have a robust understanding of some of the critical elements of bringing AI onboard and using AI in business.
The education of executive teams is going to be a serious dynamic for companies that are paying for events and preparing themselves to adopt AI. Business leaders are going to need to understand three core facets of the Roadmap of Readiness:
- Data Infrastructure
- Time and Iteration Requirements of AI
Leaders in firms need to understand the potential value of data. How can it be used in machine learning applications? How can it be stored? How can it be harmonized? How can it have a uniform format or mode of making use of it in the future? How can it be accessible between and across silos? How can it be stored the right way the first time so that it can be used in the future?
A broad understanding of data infrastructure and an understanding of what kinds of data could be valuable and why is one element of the learning and preparedness of companies that are able to move quickly and move to serious advantage when they enter the “Adoption” phase.
Companies that are smart in this area understand that they can’t rely on consultants or upscaling their C++ programmers to become machine learning engineers.
There’s going to have to be strong in-house data science chops. If a company wants to work with a vendor or integrate some fancy AI tool, they’re going to have to have in-house data science talent. That involves not just people with a background in data science but having their leadership understand broadly what AI can do.
Companies that have all the money in the world to pay great PhDs won’t have a competitive advantage if none of the people those PhDs talk to at the company speak the language or know how to integrate the tools the PhDs create.
It’s at the companies that have to some degree an understanding of data baked into their leadership where that new data science talent will actually have a home, will actually be able to deliver something and learn about the subject-matter expertise of the leaders, will actually become a contributing part of the team.
It’s not that consultants are bad, but without in-house talent, companies aren’t going to see an advantage. They can’t win by just hiring PhDs.
Many AI tools and libraries are becoming increasingly more accessible to people without a background in the hard sciences, and this may allow less mature Emergence-phase companies to begin developing some familiarity with AI. Marco Lagi is Tech Lead of Machine Learing at Boston-based HubSpot:
Many tools are becoming easier to use… that is happening now. Lots of libraries are being developed as a layer of abstraction over the powerful and clunky APIs in the machine learning space. You look at Ceres or FastAI and they are getting so much traction and success that I think a lot of people are entering this space… there are forums and communities online where more and more people are picking it up.
Time and Iteration Requirements of AI
Companies that really understand a “roadmap of AI readiness” are not companies that thin, “Let’s get some AI; let’s plug it in and see some immediate benefits.”
These are very naive assumptions. To be ready for AI, to be able to take advantage of it when it becomes accessible, companies have to have a grasp of honestly how challenging it is. Machine learning applications are going to often involve training off of a company’s own historical data and potentially their own real-time data to reliable recommend better products or detect fraud, for example.
Oftentimes, this takes a while. Sometimes vendors come in with pre-trained systems, and that process is a little bit easier, but oftentimes there’s going to be a lot of uniqueness to how a company’s data is structured. Companies have to embrace the iteration process of months and months that potentially gets to a place where they don’t see an ROI beyond what other technologies can do.
Companies that “get AI,” that will be ready to grasp this as a competitive advantage in the future, understand that having a willingness to experiment is inherent to “doing machine learning.”
Another aspect that will give companies a competitive advantage is being able to spot opportunity. A company that wants to do this has to have subject-matter experts at the company that “get AI” and understand the “roadmap of AI readiness.”
But leaders within the company also have to understand the key processes within their business where AI can be applied. They have to have their heads in the AI world enough to know that if they deal with a lot of payments, they know which kinds of payment processing concerns are low-hanging fruit for AI and seem to have traction, for example.
This is not just a conceptual understanding of AI; it’s enough of a grasp of the functions of their business and enough of a grasp of AI out in the world where business leaders can see where they mesh.
When business are in a space where they might not be actively adopting half a dozen machine learning application tools, they might at least know that when they make the step into “Adoption,” they know where ROI might be strongest for their business.
Companies pre-“Adoption” that grasp this fully will at least go in knowing what it’s going to take and knowing where AI could hit the ground running to give them that competitive advantage. They at least don’t have to learn these lessons backwards.
Most companies in the world are going to gain their early advantage in the competitive landscape of AI in the “Emergence phase.”
Phase 2 – Adoption
The first area of the competitive dynamics of the “Adoption” phase is in data dominance. There exists a dynamic of “winner-take-all,” where there’s a pocket of profitable, valuable data that allows companies to not just have a more robust solution to better solve the customer’s problem than any other solution on earth.
We see this play out in companies like Google. The company gets a tremendous volume of search going through their search engine, which means they’re able to consistently train their algorithms to deliver and test more relevant search results than any other companies. By doing that, more people will search using Google because Google will have more relevant results. When they have more people searching with them, they can make more money. When they make more money, they can invest even more money in crawling websites and iterating and testing their algorithms for presenting relevant content. As such, more and more people are going to use Google.
We discuss this further in our article, The AI Advantage of the Tech Giants, and this is going to extend to other spaces. Below are some examples:
Data Dominance Example: HVAC
There might be a company that masters IoT sensors and IoT data for HVAC maintenance and efficiency. By plugging into millions nad millions of HVAC systems around the globe, they’ll be able to better create efficiencies and savings and better do predictive maintenance than any other HVAC company in the world.
As a result, because of how much data they have, they can promise better efficiencies and greater uptime with HVAC systems than any other company. They’ll then win more customers and build a better reputation and have better results to the point where it would be impossible for anyone to create IoT sensors and software that could better help with the maintenance and efficiencies of HVAC systems than this company.
Data Dominance Example: Anti-Money Laundering
There might be a company that becomes so good at integrating themselves into a position to be the anti-money laundering vendor of so many big banks that their system can just reduce that laundering cost so much more than any other anti-money laundering system on the market. They might be able to go to any other bank and say, “We’re going to be able to get you better results, prevent more fraud, save you more money, than any other solutions because we’re drinking in the patterns of all of the biggest banks in the world. So if you want to do anything with anti-money laundering in banking, you have to go with us.”
In doing this, they can continue to swell that data dominance plume.
The second aspect of competitive dynamics within the “Adoption” phase of the AI Zeitgeist is having a differentiated vision or enhancing one’s strengths.
For any given company that’s thinking about AI, there’s going to be an intersection of what their customers want, what AI can possibly help to enhance, and what the company is best at. Finding a combination of those factors is going ot be its own art form.
We might find that across the banking space, there are some banks that have a much more modern customer support infrastructure. A bank in that position that is already known for great customer service and has accessible data in that space might use that to master customer service with AI. They might get in early with some early chatbot applications to help with low-hanging fruit questions and be able to feed that into analytics systems that might already be pretty advanced.
In the grocery business, there might be a grocer that is particularly good with inventory management processes and technology. They might be good at predicting demand and rotating their inventory, for example. They might decide that in leveraging AI, they can further entrench their differentiation. They want to use that to continue to compete in the market.
Being able to form a differentiated vision is what we’re going to see companies that win with AI actually do.
Part of our work in doing business strategy with enterprises is about crafting this. Sometimes it’s done a little before the “Adoption” phase; sometimes it’s done a little after the “Adoption” phase. Companies that want to win with AI in the future are going to want to craft a vision that’s coherent.
Talent Acquisition and Retention
The last dynamic is in the world of talent. Below are some questions that companies are going to be asking themselves for the sake of figuring out where they stand against their competitors:
Is the Data Science and AI Work That We Have Unique, Purposeful, and Exciting?
Is there an interesting, compelling problem that only the company can solve or an aspect of a business area that a company is at the cutting edge of? It’s neat when we can tie it to a cause, but “purposeful” can also be about how unique and exciting something is.
Companies that don’t have unlimited budgets are going to have to come up with these kinds of ideas for successful hiring strategies. Even companies with unlimited budgets will have to compete with how exciting it is to work at Facebook or Google.
Answering this question successfully is going to be critical. Business leaders in the sectors hat we are heavily involved, insurance, pharma, banking, retail, know they need to hire data science talent, and in some cases, they have a use for that talent, but their job listings are often bland descriptions; they’re like any other job description. To be bale to articulate this well is going to be criticla for not just winning the talent but actually retaining them.
Is This Making a Business Impact?
I’ll tell you one thing right now. When I talk to top talent that’s working within big companies or small companies, one of the things they’re most excited about is one part of career advancement: what talent gets to do once they hit the ground running on day one of joining the company.
If the company hires a bunch of PhDs to sit in a room and there’s no one to tell the PhDs what problems to work on or anyone to answer the questions they might have, they’re not going to be moving the needle very much. Maybe they make a great paycheck, maybe even one that’s better than their offer from Facebook, but they don’t go home feeling satisfied and they don’t go home feeling like they’re learning.
This is a fast-moving field. Making a business impact is part of staying relevant in the field. If a company is hiring an AI team that’s four times as big as a business knows they can make use of, they may be able to hire them, but they may not be able to retain them for more than three months before they get bored and move somewhere. This is part of why these PhDs like to move to Silicon valley.
If they hop in at Google, Facebook, or Amazon, they’re going to have something productive to do with AI that’s actually going to have an impact at the business. This is in opposition to older firms that might be hiring PhDs just do AI for AI’s sake
Is the Connective Tissue of Subject-Matter Experts in Place?
As mentioned, on the big questions here is whether or not the leaders at the business can speak data. The people that a data science interacts with at the company should speak their language, know why their work is important, and take the time to speak with them because they should know how to be part of the team with them. If the data scientists at a company are hard to retain, it’s likely because the connective tissue isn’t in place.
Do We Have Data Accessible?
There are all sorts of horror stories in the Ai world of companies that hire AI experts and are told the kind of problem the company wants them to work on, and the PhDs now have to spend a lot of time wrangling with people at different departments to collect the data and clean the data so they can actually use it.
If the company doesn’t have some basic understanding of how to access their data, they can’t expect to have PhDs bicker via email with people in the IT department about setting up a data silo, for example.
Do We Have Career Paths for Data Science People That Are Compelling?
When someone comes in, is there anywhere for them to grow? Does the company understand the kinds of hierarchies that are going to develop in the different departments within their business? They can’t just hope to hire Ai folks like a bandaid. It’s going to have to be a career trajectory. These PhDs are going to want to know that they can pursue their interests, to lead projects, to make more money.
Within data science, sometimes there isn’t even an initial preliminary thought to this. There isn’t much thought into what that job might look like in two years.
When it comes to competing, a company needs to have a clear vision of how they’re going to attract and retain talent. We speak more about this in our third installment of the AI Zeitgeist series, which readers can navigate to above.
There’s always going to be a limited number of data scientists, and if a company wants data dominance to take root, they need the talent to take root with them.
Michael Segala is a PhD from Brown University, and Founder of SFL Scientific, a 30-person data science consulting firm based in Boston. He sums up the shift from AI nascent beginnings to a more necessary strategic consideration, even for mid-size firms:
As innovation decreases the costs for big data and increasing the ability to achieve highly accurate models, it will become more evident that a holistic AI strategy is imperative for organizations to thrive in this era of digital transformation. As seen in traditional consumer-driven industries and more recently mobile tech and SaaS firms, companies will feel pressure to develop “me-too” products and services in this space; some relatable examples in recent memory are machine learning-based playlists for song recommendations and voice assistants.
When these features work, they attract and engage users, build brand loyalty, and provide sticky name recognition. Today, outside of industry leaders and businesses founded on AI-based solutions, companies that see low-hanging fruit are making small investments into data science teams and tools in the hopes of seeing a lift. As they test and prove themselves, sustained investment into AI teams will become commonplace and executives will have the comfort and clarity to make larger strategic investments.
Phase 3 – Dispersion
The dynamics of data dominance and the differentiated vision we talked about in the “Adoption” phase are still very much going to be in place in the “Dispersion” phase. But there will be other facets and aspects that business leaders are going to have to bear in mind.
One of these is a dynamic called “data ubiquity.” At some point, the tracking and storing of all data streams will be more or less combined, accessible, useful at a company.
The companies that are going to be able to move into whatever enhanced capability are going to be the companies that can access the data to do that. Ai will not be seen as different or particularly challenging. Data will be seen as oil; data will be used as oil in all business functions.
The companies that are going to compete and win are going to be the ones that have that fluid ability to access whatever they need to access. Payment information, on-site activity on their webpages, inventory, and financial systems, for example, will all be able to be merged and purged and breathe life into whatever capability the company wants.
In the Dispersion phase, the companies that are data ubiquitous will be at an advantage clearly because they’ll be able to pivot, to adjust, and to make use of whatever new capability that they can think of. When I say “data ubiquity” I also imply companies having the ability to access data they might not have in-house. They’ll know what kinds of data they could crowdsource or buy third-party data, for example.
A data ubiquitous firm is going to know what to crowdsource, what to buy, and how to access absolutely everything within their organization that could be leveraged to drive a certain capability.
This feeds into the other element of the competitive dynamics of the “Dispersion” phase: efficiency and agility.
Efficiency and Agility
What does data ubiquity allow a company to do? It allows them to move and pivot and take advantage of whatever seems like would permit them to garner an advantage. A heavy investment in a single strategy is going to be more dangerous than ever five years from now when dynamics change. Efficient and agile firms that ren’t too sunk into any one way of the world will be able to use data to pivot.
But this goes a bit further. If the world is changing very quickly, then certainly we need teams and staff that’s able to learn and adjust fast, but also companies need efficient operations. Companies need operations that aren’t too hung up in any one area.
If they can use data and AI to automate and augment as much as possible, then instead of hiring 200 people they can hire four. That’s a much safer bet because what does the company do with those 200 people if that whole operation is moot in five years?
This has to do with a number of different factors. Some of the elements of efficiency and agility are as follows:
Infrastructure and Data Infrastructure
How are a company’s systems set up? Are they married to a particular way of the world or a particular means of interacting with a buyer or are they able to adjust and bend?
I have a longer article that was based on a Ted Talk that I did at the University of Rhode Island about three factors of job security in the age of AI. This to some degree has to do with talent’s ability to adapt, but there will be teams that are able to learn whatever new technologies or means are possible. If a company has a marketing department team that’s married to email, they’re probably not going to be there in the “Dispersion” phase.
What we see at companies at Google, for example, is a relatively small team to profit. For Google, each employee is worth a lot of revenue and a lot of profit because they have a tight, compact, relatively high-paid group of people working on core things. Google outsources ancillary tasks. More and more firms in the future are going to have core teams of high-paid, high-skilled folks that are being hired for that adaptability.
I’m not sure how society is going to adjust to that, but certainly business will have to adjust to that.
Services and Products
Apple is always reinventing itself. Amazon is always finding new means to get more customer engagement even before their other channels fail. It’s not like email marketing doesn’t work for Amazon, but they also do a tremendous amount of recommendation work. Then they offer Alexa and let you buy from this “robot” on your kitchen counter. Google is constantly expanding its set of offerings. The company is very well aware that search, as it is today, won’t be here forever. They’re always trying to get an idea of the intent of their users even before search the way it is today becomes irrelevant.
Companies need the ability to be efficient and agile with how they deliver their services and products. This will not only be possible in the “Dispersion” phase, but will be an inevitable dynamic across sectors.
First and foremost, most firms reading right now will be in the “Emergence” phase, but I will say that if you’re in a firm of that kind and you plan on adopting AI, be considering some factors of competitive advantage in the “Emergence” phase. It might seem like until you can get your hands on data scientists, there’s really no way to stay ahead of the game competitively in terms of AI, but actually, a lot of the advantage of AI in business is about the level of awareness of the leaders of a company.
So take that seriously. Believe me, a lot of the headway happens well before one buys AI.
For business leaders in the “Adoption” phase, you’re probably already thinking about talent quite a good deal. Consider some of the questions I brought up in this article. Think quite seriously about them and take advantage of the opportunity to be a better home for data science talent. And if you haven’t already, be formulating that data dominance strategy.
This is something that companies are going to be doing in their board rooms. They’ll be thinking through what they’re good at, what they market wants, what AI enables, and how in five years they’ll be so far ahead of the competition that their competitors will be in their dust.
As for “Dispersion,” I don’t have a crystal ball, but I do believe quite thoroughly that data ubiquity will be a big deal in Dispersion, and so companies should set themselves up for that will be at a greater advantage and will be able to be efficient and agile.
I hope sincerely that this has been helpful. A lot of these ideas have been the culmination of the executive conversations and researchers conversations that I’ve had for the last couple years. “AI Zeitgeist” is really a forward look at the progression of AI in business from years of these conversations, and to finally shake them out in a series has been really fun and really rewarding.