This article is part 5 of a 7-part series called “AI Zeitgeist,” where we’ll be mapping out the details of AI adoption over the next 10 years and explore the critical changes in the AI ecosystem that business leaders need to understand.
In this installment of the Zeitgeist series, we’ll take a look at how the process of buying artificial intelligence solutions will change in the years ahead. Buying AI today involves wild varieties of fear and hype, and generally requires a heavy “education”-phase investment on behalf of the vendor companies.
Educating stodgy enterprises on the details of AI and data won’t be going away any time soon, but as executives in large firms become more familiar with the lingo of data science—and as they develop a more reasonable understanding of viable AI use-cases—the education phase will be shorter (and the potential for vendors to make bloated claims will be reduced).
(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).
Through each phase of the AI Zeitgeist (“Emergence,” “Adoption,” and “Dispersion”), we’ll examine both of the following considerations and how they’ll change and evolve:
- Buying Motive: The reason that companies decide to purchase from an AI-related product or service company.
- Buying Challenges: The hurdles that companies must overcome to successfully understand, purchase, and adopt AI solutions.
We’ll begin by exploring the accessibility of AI in the current phase of AI adoption: the “Emergence” phase:
Phase 1: Emergence
In the new, hyped-up “Emergence” phase of AI, buying and selling are based inordinately on the concept of “artificial intelligence” by itself.
Vendor companies want to seem modern, exciting, capable, and in line with the future, so they tout “AI” as part of their offering, even if it’s not involved in their product or service at all.
People with purchasing power within enterprises want to seem modern, exciting, capable, and in line with the future (particularly to their boss or possibly to the public), so they get excited about AI for its own sake. Vendors often don’t hear from enterprises that need a problem solved, but from enterprises that “want to start using AI” (what we’ve previously called “toy AI applications“).
Charles Martin, PhD, head of Calculation Consultants, says this about the majority of companies that come to him for AI consulting:
There are companies that are trying to do it, but they’re not serious about it. They’re putting their junior guys on it; they’re trying to learn it, but they have no idea what’s going on on Silicon Valley.
As a result, because buyer companies often don’t understand AI capabilities and the value of data, they’re getting educated on these fundamentals by the vendor company themselves (a slightly biased party).
Also, buying is driven by existing precedents. Buyers understand it if Google does it or Amazon does it or if the biggest players in their industry have done it. These precedents will be weak, however, often based on no more than here-say or a press release and not on strong evidence of ROI.
(Business leaders in the “Emergence” phase without a strong grounding in data science should read our article on selecting an AI vendor for non-technical executives before lining up potential solution providers or consider our Vendor Selection service).
As we mentioned before, many vendors companies in the “Emergence” phase are still trying to find their value proposition, and the exact product they’re selling. Early customers are test pilots – are opportunities to see if (or how) they can deliver on what they promised when they closed the deal.
This means that many vendors don’t have a succinct and clear value proposition for buyers, or a rock-solid set of expectation about what an integration looks like, and how long it will take to garner a result for the client. This can (and sometimes for good reason) scare away buyers.
In addition, vendors without robust case studies will have a harder time closing clients, which is why so many companies (even firms who have raised tens of millions of dollars) will be eager to get “pilot” programs with potential clients—often at a drastically reduced price—in order to prove their worth.
A lack of precedents of use and ROI (case studies) can make intelligent buyers hesitant to jump into the deep end, particularly if the application is complicated and requires competences that the buyer doesn’t possess.
Executives who are not familiar with AI will have a particularly hard time getting used to some of the new and important considerations for adopting AI, including:
- The requirements for data and data infrastructure in order to “feed” an AI system
- The fact that machine learning solutions won’t achieve 100% certainty, and it may take months and months of iteration in order to get a machine learning solution to a sufficient working level (or those months of experiments may yield nothing)
- Teams (technical and non-technical) must be familiar with data science concepts
In most established sectors (banking, pharma, etc.), it will be years until these concepts of AI adoption are understood, and even longer until well-established best practices exist for implementing AI in an existing, established business.
Phase 2: Adoption
Buying motives shift from “artificial intelligence” buzzword interest and more toward the actual cost and return of the application itself. Buyers are more able to make this kind of assessment in the “Adoption” phase.
The following factors will contribute heavily to this shift:
- Buyers (both technical and non-technical execs) will have a more complete understanding of what AI can and cannot do. For this reason they’ll engage with vendors in a more selective and pragmatic way – as opposed to from a place of confusion or fascination or curiosity (as they often do in the “Emergence” phase)
- Buyers will be aware of their data assets and data science staff and will have a more rounded understanding of what kinds of applications might suit their business, and which would be out of their reach
- Buyers will have experience with building and buying AI solutions in the past and will have learned from their many peers who have done the same
In other words, procurement for AI solutions will become much more like…well… regular IT procurement: a weighing of pros and cons and an assessment of the solutions that seem to warrant assessment.
Precedents of use will be more robust in the “Adoption” phase. Companies will have a reasonable understanding as to what kinds of applications worked for their competitors.
Another buying motive that will become more pressing in the “Adoption” phase is trends. In any given industry, trends will emerge which seem to be critical for success in that given industry.
A few years in the future, a specific kind of voice-based customer service may become a clear “wave of the future” in retail banking. There will be robust evidence of traction with this application, seemingly clear indicators that customers prefer it, and the fastest-growing players in the sector will be the ones using this new kind of customer service. In this case, retail banking companies would be pressured en masse to build or buy this technology, bringing new waves of buyers to market.
It should be noted that in the current “Emergence” phase, most industries haven’t seen nearly enough AI adoption to flesh out these kinds of broad “trends”, but their presence will indeed be a signal that a given industry is entering more of an “Adoption” phase.
Today (in the “Emergence” phase), vendors might be able to cajole buyers into purchasing a solution even if the buyer doesn’t have the data volume or in-house capabilities to actually use the application.
In the “Adoption” phase, buyers will have a stronger sense of what makes sense and what doesn’t in their business. They’ll push back against applications that don’t seem to make sense or that don’t mesh well with their own data assets or staff capabilities. “AI” itself will be less likely to push a buyer over the edge.
Our technical advisor, Marco Lagi (Tech Lead of ML at Hubspot, PhD in Physical Chemistry), explains what business leaders looking to buy AI solutions need to realize in order to get to the “Adoption” phase:
It’s mainly a business decision and operations decisions…more than a technological advancement…it’s recognizing it’s value. The company has to make decisions about how to serve data to the people who build the models…but it’s essentially realizing for what AI can be applied…and realizing that this is not just predictive scoring for a marketing company… it’s many many things… it’s a whole opening of the capabilities.
In the “Adoption” phase, vendors without case studies will be rarer (as there will be more established and proven players in various AI niches), and so newer vendor companies or products will have to do a better job of explaining what makes them different and what makes them worth “taking a chance” with – as there will be established players in the field already.
The “winner takes all dynamic” has the potential to see many AI players dominate specific business functions or industry applications, owning more and more data, allowing them to provide an incrementally better and better product, and in turn, continuing to win market share.
Smaller players will have to contend with larger players who are already starting to roll the “data snowball” that may allow them to own a niche almost entirely. For more detail on why large tech firms dominate AI innovation, read our article: The AI Advantage of the Tech Giants: Amazon, Facebook, and Google.
Phase 3: Dispersion
Once almost all enterprises can “speak” data and AI lingo (even if they can’t write the code), AI won’t be as much of a mystery. Basic understanding of training algorithms and precedents for AI use will be common in boardrooms everywhere once we have entered the “Dispersion” phase.
There won’t be a separate thinking process (or business process) for procuring AI-related solutions, there will simply be “IT procurement.” All tech will be smart; there will be no exceptions here.
I repeat: There won’t be a separate thinking process (or business process) for procuring AI-related solutions, there will simply be “IT procurement”. All tech will be smart; tere will be no exceptions here.
This has been week 5 of 7 from the “AI Zeitgeist” article series – only two more articles to go to complete the series.
In the coming weeks, we’ll be exploring the following AI Zeitgeist topics, in order:
- 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) <— You are here
- 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)
In next week’s article (6 of 7), we’ll take a look at how the AI priorities of business leaders will change in the decade ahead (Preview: Education and events is part of the current “Emergence” paradigm, but will fade away gradually as “AI Strategy” moves from being an obscure exercise to a critical boardroom conversation).
Catch up next week.