Three things are clear from a recent survey of chief AI and data officers from over 125 major firms: 98% reported they are increasing their Gen AI investments in 2025 – up from 82% last year. 91% said investments in data and AI are a top priority, up from 88% last year. Perhaps most importantly GenAI systems in production have increased almost fivefold from 5% of firms to 24%. It seems Gartner’s claim that this technology is at the height of inflated expectations may not square with the facts.
What I are wondering, as this investment and implementation wave continues, is who will lead these GenAI efforts? The more I examine the problem, I think it will be spread throughout the organization where those who own the processes, the data and the profit and loss statements will drive the implementation of this new capability. In turn, this means that the technology executives, especially the CIO and the Chief Data Officer should be ready to deliver in a world with diverse needs and distributed authority, perhaps in a more flexible model where certain decision rights are held at the center, but most of the action is within the divisions or functions.
Why do I think this technology wave will be driven by executives outside the technology function? In my analysis there are five core reasons.
The first reason is that GenAI helps organizations fundamentally change work that is made up of Words, Images, Numbers and Sounds. In our October 2023 HBR Article Where to Get Started with GenAI, we called this WINS work. WINS work is spread throughout any organization. Marketing creates analyses, images, stories and more. Sales responds to RFPs, contracts, and pitches for customers. Human Resources finds, deploys, upskills and judges talent, and the list of WINS work goes on and on. The knowledge of how these functions and processes work, and the wonderful low code, no code nature of GenAI tools, means that both the knowledge of how to improve them and the technology to improve them, can live within the function or division.
The second reason is that improvement will be outside the IT function is that the key barrier to adoption according to the head of Microsoft’s Co-Pilot division and a senior executive from OpenAI is that people need extensive training on how to use these powerful tools, in order to get their deep value, and to not be intimidated by them. Alan Kay, one of the key geniuses who brought us the personal computer and invented object oriented programming – the basis of most modern consumer software noted that it takes time for anyone to learn a powerful tool. Language, mathematics, a musical instrument, all take time to learn. So too with these models. The good news is that unlike math or a piano, these tools can not only perform but they can tutor you on how to use them too. This investment in training is unlikely to ever come out of a central IT budget, but is more likely to live within the division or function of the firm.
Peter Drucker the legendary management theorist once said, “The purpose of a firm is to create and keep a customer.” My third reason is inspired by Drucker’s observation. 80% of firms report their primary focus for their efforts is growth and innovation. In many organizations, power tends to reside in those functions that can drive growth and profit. In technology firms this is often the engineering organization. At a financial services firm, it is often finance. At a CPG firm, it’s marketing. It is the rare organization where IT is the leader for growth and innovation. Any good IT organization supports those key goals but does not usually lead them. Given the focus of GenAI investments in the growth and innovation of firms, we expect that they will be in a supporting, not leadership role. Many of the early use cases of GenAI use in firms are in support of customer service professionals, ranging from training and improving call centers to entirely automated chat bot interactions where the chatbot has a reasonable IQ and can answer real customer questions as well as escalate quickly and effortlessly when a customer gets frustrated. The powerful functions also tend to “own” the data and processes needed to create productive models.
It is true that OpenAI’s ChatGPT still processes the vast majority of large language model queries, and by itself has rocketed to one of the top ten web properties in a mere two years since their ChatGPT 3.5 release. However, my fourth reason derives from the fact that we believe we will see more and more GenAI solutions aimed at functional and industry needs . Harvey is aimed at transforming the practice of law. At the recent Consumer Electronics Show Qualcomm, in conjunction with IBM and Honeywell has introduced its Qualcomm On Prem AI Appliance Solution that can automate processes in environments from quick service restaurants to hospitals to dealerships, etc., across the full life cycle from data collection to model maintenance. The market will tell us if this product will live up to its claims, but it is just one of thousands of new solutions aimed at functions, divisions or entire businesses. This type of solution is purchased by the owner or line manager, and they will consult with IT, but the ownership will stay with the business executive.
Last but not least, ambitious executives often build their careers by championing a new product or service, or transforming an existing part of the organization. A colleague and friend of mine Betsy Holden became co-CEO of Kraft for many reasons, not the least of which was that she was the product manager for Tombstone Pizza, the first rising crust frozen pizza with take out taste. It sold like hotcakes, and Betsy rose with it. Likewise, the strong talent in organizations are starting to realize that if they are willing to do the hard work, this new set of GenAI technologies can create massive productivity through capital substitution for labor, it enables better innovation acting as a thought partner and content generation team, and provides vastly more scalable organizations. As people who have been senior executives in large service organizations we know first hand how powerful that last point about scalability can be as markets move up and down over time. Many CEOs of existing businesses five years from now will get there by using GenAI to transform the cost, innovation and growth of a major part of their own organizations. It is unlikely that the majority of this leadership will originate in IT.
Elon Musk has recently said, “I’ve never seen a technology that has improved as rapidly as AI.” For those of us in the industry, it literally changes day to day. Some organizations would say, let’s wait and let things quiet down. However, we believe that there is a learning effect with this technology. Using these models is an organizational capability – not just a technology. Staff need to be trained, processes changes, unstructured data accessed and harnessed. Unstructured data buried in pdf’s, customer comments, RFPs, contracts, etc., etc. makes up 90% of most firm’s data. In addition, these models get better and better with use as organizations improve the quality, range and accuracy of these models. There is a learning effect that will be hard to chase, and a process of progressively structuring unstructured data that can only be done with time, inside your firm. Put another way, time will not make your firm’s journey easier, and it will give the competition time to bolt ahead.
Often the CIO’s function is to rationalize and simplify an organization’s technology base, and get the most secure, reliable, lowest cost solutions. However, when there is a cambrian explosion of innovation as there is in this domain, the CIO must add an additional capability – sense making. The CIO’s must be able to scan and evaluate key trends, emerging vendors, and vital service providers in a volatile environment. This sense making and evaluation function is available in the most sophisticated technology functions, but even those need to up their game in this new wave of capability.
The Chief Data Officer’s Role (CDO), was traditionally to reduce cost and mitigate risk, by creating orderly data repositories, and CDOs enabled companies to increase operational efficiency, reduce data duplication, and prepare for more predictable modeling efforts. However, over the years with data analytics and traditional AI, the CDO has been more outward looking and growth oriented in addition to the efficiency role. However, this wave of new data types, pictures, sounds, videos, voice, pdf’s, and more need an entirely more open approach. Also the engagement method with the business leaders has to be one of engagement and progressively solving the problem. Put another way, it is the symbiosis of the models and the type of data that they are fed that can help take knowledge that was so scattered and fragmented and make it accessible. Think of tens of thousands of RFPs that firms now have available for review and analysis at a trivial cost. Consider tools that can review tens of thousands of contracts in their original form looking for savings and risk. Imagine all the customer conversations that are ongoing day in and day out being available in real time, distilled for their trends and implications. This process is active engagement with the process and the data it kicks off, it is not an exercise of investing heavily for clean data in a data lake with centralized control. The value is much closer to the business process and CDOs need to be able to operate in this new world and focus on making new data available to the power of the models.
In short, because the real value of GenAI comes when business processes, products and services are radically reinvented, and due to the fact that these real value comes when firms create an organizational capability to mix technology and talent we expect this wave to be lead by executives outside of the IT function. For the technology leader, this can be very intimidating, or invigorating – depending on whether they are motivated to control the firm’s approach, or to unleash the vast potential for transformation in this new world.
