The direct, personal involvement of a company’s CEO is essential if any significant transformation is to succeed.  Yet for AI, most have stayed on the sidelines, unclear what exactly they must do, or assuming AI belongs to IT. This will not stand. This article proposes five non-delegatable tasks belonging on their To-Do lists. These include:

1. Boldly setting their company’s ambitions
2. Learning to live with uncertainty
3. Making data a first-class citizen
4. Building data and AI into the organization, rather than “bolting it on”
5. Acting swiftly but thoughtfully

All things AI press on at an almost mind-boggling pace. Tech companies – and the Trump administration, Europe, China, and many countries – are making enormous infrastructure and other investments.

Stories featuring game-changing potential (e.g., AI co-scientists or co-workers) appear daily; AI providers announce breakthroughs with increasing frequency; companies (e.g., Duolingo or Shopify) announce stunning new AI initiatives; and consultancies project multi-trillion-dollar benefits in just a few short years.

The story would not be complete without geopolitical implications, as concerns about Nvidia chips, Taiwan Semiconductor factories, the availability of rare earth minerals, and progress in China (e.g., DeepSeek) illustrate. It is hard not to get caught up on the excitement!

In stark contrast, AI is yet to deliver on its promises: Too many companies don’t know how to assess implications or where to start, many others can’t get a simple pilot into production, data continues to be a second-class citizen, and resistance comes from all quarters. On top of this, many companies mistakenly believe they already have an AI strategy simply because they have activated tools like Microsoft Copilot.

When they fail to see the productivity gains they were expecting, they conclude that AI has little to offer them. There is a lot to learn and do in a very short time. After all, a technology can run on hype for only so long.

When it is nearly impossible to separate the signal from the noise, what should a company do?

We’ve contributed to, led, and advised on plenty of AI and data efforts — some proved transformational, some moderately successful, and some outright failures. In our view, progress has been shackled by a lack of both understanding and imagination, paralysis in the face of uncertainty, a failure of courage, bad data (both structured and unstructured), half-heated measures, a “we build it and the come” misconception, and organization charts best described as “unfit for data, people and AI.” It is time to take the shackles off!

We’ve also learned that: 

• AI and data efforts are way more difficult than their supporters admit.
• Only the boldest, those with the best leadership, those willing to question everything, and the courage to make fundamental changes succeed.
• The effort goes no further than the top-most person deeply involved demands.

Combining these perspectives leads us to five non-delegatable responsibilities for CEOs hoping to garner their company’s share of the benefits cited above:

1. Don’t limit your ambitions.
2. The uncertainty is enormous. Do not get paralyzed – learn how to live with it.
3. AI is data. You must make it a first-class citizen.
4. Build data and AI into your organization, rather than “bolting it on.” 
5. Act boldly and soon, but thoughtfully – The race to create value using AI is on, and it is a marathon, likely with many sprints.

Let’s explore each in turn.  

1. Don’t limit your ambitions

Considering the constant flow of AI “news,” we understand why CEOs might prefer to make smaller bets on easier projects. That can be fine for a first step, but don’t limit your ambition of where you want to eventually reach. Consider this sequence of increasingly bold possibilities.

• Process productivity: AI is already helping automate processes and improve productivity, just as other information technologies have done in the past.  One recent example involves TFAS, a financial advisory firm that announced it had improved productivity by 25% by providing access to a GenAI assistant. The longer-term target is a 70% productivity improvement.
• Individual and team productivity: Anyone who has tried ChatGPT, Gemini, or CoPilot has seen firsthand their potential to boost individual productivity—like having a trainee with superpowers. A study by Harvard Business School and BCG confirmed this productivity gain, showing that consultants using GPT increased their output by 12% while also improving quality.
• Improved product and service: Two generations ago, renowned business advisor Stan Davis introduced the concept of “informationalization,” making core products and services more valuable by building in more data and information. The tire gauges built into today’s automobiles are one everyday example. We can’t think of a product or service that can’t be “informationalized.” AI supercharges the concept, new products appear everyday with integrated AI to enrich their features, such as the glasses proposed by OrCam for the visually impaired, or a facial recognition system built into our smartphones.

What if? Indeed, AI enables organizations to rethink entire business models. The best current example is the impact AI is having on warfare. As pointed out by the CNAS, “AI is emerging as a significant asset in the ongoing Russian-Ukrainian conflict” and “translates into more precise and capable responses to adversary forces, movements, and actions.”

 Another example involves the Singapore National AI Strategy that focuses on “fundamentally rethinking business models and making deep changes to reap productivity gains and create new areas of growth.” Every company and government agency should ask itself extensive “what-ifs” involving AI and competitors, new entries, long-standing business problems, and relationships with customers.

Of course, at any point in time, CEOs must decide what projects to actually fund, or to fund first, in the face of Board pressure, competition among subordinates, and employee concerns. Don’t let politics hamper your ambitions.

2. The uncertainty is enormous — Learn how to live with it

The cold, brutal reality is that everything about AI is uncertain: Estimated benefits and feasibility vary greatly, most “projects” so far have either failed entirely or failed to get out of pilot stage, trust in AI is often low, with many fearing it, and national governments are unsure what and how to regulate.

And who knows what impact the new U.S.-China “space race” will have? It has even gotten to the point that insurance companies such as Munich Re offer insurance products for new AI ventures.

Not a comfortable environment for managers who’ve spent their entire careers trying to reduce uncertainty. Making matters worse, we don’t see any relief in sight. Our best historical analogy involves the printing press, where it took two full generations for some semblance of order to emerge.

Senior leaders, especially CEOs, are well-advised to embrace uncertainty, though “learn to live with it” is probably more apt. Indeed, increasing uncertainty, especially rapid-fire “unknown-unknowns,” creates winners and losers. To do so, CEOs and companies must learn to experiment, in everything from technologies to funding mechanisms, to people and organizational structures.

All experiments take time and resources, so it is best to plan them carefully. Some experiments, such as those that involve moving from model to adoption will be large.   And plenty fail. Importantly, a part of experimentation involves reaping the benefits of successful experiments and learning from failed ones.

Said differently, rather than investing in a long-term strategy “get out there,” sort out your company’s capabilities, learn, grow, and adapt.  

One final point: Adopting a “wait and see” attitude qualifies as an experiment. An even worse one involves “dabbling,” perhaps motivated by FOMO-fear of missing out. Such halfway measures are not big enough to succeed, but may well convince you, “AI is not for us,” or “the technology is not ready,” and lead to missing out.

3. AI is data: You must make it a first-class citizen

As James Betker of OpenAI observes, “The ‘it’ in AI models is the dataset.” And today, data availability and quality are generally regarded as the number one limiting factor for AI, as highlighted by the recent failed launch of France’s AI chatbot, Lucie.

Of course, very few companies will actually develop a foundation LLM such as DeepSeek, Chat, or Gemini, nor can they do much about data scraped from the Internet used to train LLMs. Instead, companies must focus on the data they use to augment LLMs and to train and operate predictive AI models.

The issues are broad and deep: Much data is simply wrong, poorly defined, or even hard to find; a host of “right data” issues, such as relevancy and bias, bedevil individual projects; and models themselves provide bad answers (inferences), even hallucinate.

Exacerbating this, most data is “unstructured” (i.e., emails, contracts, forms, recordings of meetings and so forth) and subject to less scrutiny than data in corporate systems.   Finally, many individuals have been well aware of the problems for some time (e.g., Garbage in, garbage out) and recognize that their companies’ data is not ready. But companies have misdiagnosed data quality as a technical problem, not the management problem it really is. No surprise, data has not received the attention it deserves!

AI brings data and data quality to the fore. We strongly suspect that data quality will go a long way in separating winners and losers. 

The implications for CEOs are clear: They must launch and fully support aggressive data quality programs, likely akin to those of the quality revolution in manufacturing, and featuring end-to-end focus on processes that cross multiple departments to supply AI applications. Fortunately, companies such as Aera Energy, AT&T, Chevron, Gulf Bank, Hello Fresh, and Shell have shown that data quality improvement is a big winner!

4. “Build AI into” your organization rather than “bolting it on”

As Arthur Jones observes, “organizations are perfectly designed to achieve the results they achieve.” AI presents some tough challenges, such as sorting the various roles for IT and “the business.” But such issues are just the tip of the organizational iceberg.

The litany of issues includes too many people not understanding their responsibilities, issues that fall “in the white space,” too much “up and down” management, and not enough “left-to-right,” and skepticism of analytics generally, and AI specifically.

In some respects, such issues should surprise no one. After all, today’s organizations were designed for industrialization, not data and AI.

Here is where the CEO’s role is especially crucial: To meet these challenges, CEOs must actively take ownership of reshaping the organization so it is ready for a world where data and AI are central. This means driving a transformation that promotes the right organizational qualities: clear accountability, cross-functional collaboration, adaptability, and a commitment to continuous learning.

The CEO must ensure that the organization becomes capable of systematically addressing data and AI challenges, breaking down (or bridging across) silos, encouraging experimentation, and scaling successful innovations. Importantly, they must foster a culture where learning from both success and failure is valued, where teams are empowered to test and improve, and where the organization as a whole becomes more agile, resilient, and ready to evolve.

In short, the CEO’s mandate is to redesign the organization not by replicating a particular structure or best practice, but by building the capacities and virtues needed to embed data and AI into the heart of the company’s operational and strategic work.

This organizational work is going to require a lot of effort, considerable experimentation, and some painful choices.  But there is no getting around it.  

5. Act boldly and soon, but thoughtfully – The race to create value using AI will be a long one!

We’ve pointed out that the stakes are high, though uncertain; that there is a lot of new, unfamiliar, and likely unpleasant work to do; competing priorities; and plenty of excuses to “wait.” For example, the tech community proudly proclaims, “Today’s AI is the worst AI you will ever use.” Why not wait until it is ready for prime time?

Acting on AI requires urgency and boldness, but not haste. Many organizations today fail to address fundamentals, leading to stunning inefficiencies. For example, employees spend an average of 30% of their time dealing with data issues. The impact on AI will be worse. At the same time, both customers and employees are already experimenting with generative tools, often contrary to their organizations’ policies. This signals not just demand, but risk: if leadership waits too long or acts without direction, others will set the pace, sometimes in ways that compromise trust, quality, or competitiveness.

CEOs must therefore lead with intent, shaping the conditions for responsible adoption. Success doesn’t come from rushing in, but from moving decisively with a clear understanding of where value lies, what the organization can absorb, and how to evolve its culture and capabilities to deliver lasting impact. For example, many companies “let a thousand flowers bloom,” but fail to get a single application into production. While you must experiment, only implementation counts as a win.

While the success rate of AI projects is low, some companies are breaking through. As best we can tell, direct CEO or Board involvement features prominently. So, even if you judge the likelihood that AI is “just a load of hooey” to be high, you’re still smart to jump in. The risk associated with “wait and see” is simply too great.

Final remarks

By and large, senior leaders have remained on the sidelines when it comes to data, analytics, and especially AI. We get it — the technology is daunting, and the pace is dizzying. But “staying apace with the latest technology changes” did not make our top five. Of course, learning some basics and having a perspective on where AI will fit is important, but CEOs have more consequential matters to attend to.

Similarly, the responsibilities we prescribe may seem new, unexpected, unfamiliar, and demanding. But play the movie forward: Do you see companies truly succeeding with AI if they do not adopt it?

About the Authors:

Thomas C. Redman, “the Data Doc,” is President of Data Quality Solutions. He helps companies and leaders chart their courses to data-driven futures with special emphasis on quality, organizational structure, and analytics. His latest book, People and Data: Uniting to Transform Your Organization (Kogan Page 2023), urges companies to increase the power of their data programs by getting everyone involved. Redman has a Ph.D. in Statistics and two patents.

Gaël Gioux is the Co-Founder and Managing Director of NoeSysAI, a strategy consulting firm that helps organizations leverage data and AI to shape bold business strategies and deliver measurable results. Covering the full journey from early discovery to long-term AI governance, NoeSysAI enables sustainable adoption by aligning ambition, delivering impactful use case portfolios, and embedding AI into operations. Prior to founding NoeSysAI, Gaël served as Chief Data & Analytics Officer at a leading Swiss insurer, where he led enterprise-wide data transformation and implemented high-impact AI use cases across business lines.

Earlier in his career, he held global leadership roles at Alcoa, driving operational efficiency and digital transformation across operations, finance, supply chain, and shared services. With over 20 years of international experience, Gaël combines strategic clarity with hands-on execution. He holds a Master of Science from MIT and an MBA from INSEAD.

Theos Evgeniou is Professor of Technology & Business at INSEAD, with nearly 30 years of experience in AI. He directs Executive Education programs for boards, C-suites, and senior leaders, focusing on AI strategy and governance. His research is published in top academic and business journals and he has been cited by major outlets such as the FT, Bloomberg, and Fortune. He has been a member of the OECD network of experts on AI, advisor for BCG Henderson Institute, and academic partner on AI for the WEF. He is also co-founder of Tremau, a B2B SaaS, and NoeSysAI, an AI strategy consulting firm. He regularly advises CEOs, gives talks, and consults globally. He holds four degrees from MIT.