top of page
Search

Leadership Principles for Successful Adoption of AI: Lessons from Innovation Leaders at UPS, Michelin, Amazon, Apple, and OpenAI


Gemini generated image using the prompt: Generate a clean image of a transformative leadership in the age of AI.


Artificial intelligence is no longer a technology of the future. Global AI adoption in companies surged to 72% in 2024, up from 55% in 2023 (Arkcoll, 2025). Yet high adoption rates alone are deceiving: only 12–18% of companies have captured meaningful ROI from their AI investments (O'Reilly, 2025). The gap between organizations that deploy AI and those that derive real value from it is widening — and the primary variable separating them is leadership.


This is not a new pattern. The organizations examined in this article — UPS, Michelin, Amazon, Apple, and OpenAI — each pursued ambitious technology transformations. What made their successes and stumbles instructive was not the technology itself, but how leaders shaped the culture, strategy, and people around the technology. A McKinsey survey of 3,613 employees concluded bluntly: "The biggest barrier to [AI] success is leadership" (Meyer et al., 2025).


This article synthesizes six leadership principles — drawn from case studies and current research — that distinguish organizations that successfully adopt AI from those that fall into the “AI theater” trap: the appearance of adoption without any organizational rewiring to capture value (see my previous article on Navigating the AI Frontier).


Principle

Core Behavior

Organizational Risk if Absent

1. Lead with Vision

CEO/C-suite model AI use publicly; anchor AI to business strategy

Adoption stalls; teams are skeptical

2. Commit to Innovation

Invest in infrastructure, talent, and R&D before expecting ROI

AI pilots never scale

3. Establish Learning Culture

Reward learning over usage; treat failure as data

Risk aversion kills experimentation

4. Take Incremental Steps

Phased roadmap with KPIs at each stage

Enterprise initiatives collapse under own weight

5. Align People and Culture

Reskill continuously; ensure experts lead AI decisions in their domains

Change resistance; talent gaps; culture clash

6. Govern Responsibly

Build ethics, privacy, and oversight into the platform from day one

Reputational damage, regulatory exposure, erosion of trust

These six principles are not independent levers — they form an integrated system. Vision without governance creates recklessness. Investment without a learning culture produces expensive shelfware. Incremental execution without organizational alignment stalls at the first cross-functional friction. The leaders examined in these case studies succeeded when they operated across all six dimensions simultaneously — and struggled when they neglected one.


Principle 1: Lead with Vision — AI Strategy Must Start at the Top

What the Research Shows

McKinsey's 2025 State of AI survey found that AI high performers are three times more likely than peers to strongly agree that senior leaders demonstrate ownership of and commitment to their AI initiatives. High-performing organizations report that senior leaders are actively engaged in driving AI adoption — including role-modeling the use of AI in their own workflows (Quantum Black, 2025).


Leadership in Action

At UPS, digital transformation stalled until Oz Nelson became CEO in 1986. Between 1986 and 1996, Nelson personally championed a $11 billion investment in information technology — not as a technology project, but as a strategic imperative to reclaim market share from Federal Express (Ross, 2002). The result was a modern IT infrastructure and the eventual creation of ORION, UPS's AI-powered routing platform that saves roughly 100 million miles and 10 million gallons of fuel per year (UPS, 2020).


At Michelin, CEO Jean-Dominique Senard and Chief Digital Officer Eric Chaniot set a clear technology transformation vision: harness AI and data analytics to enrich their core mission of safer, more sustainable mobility. Their vision explicitly identified subscription economy models, 5G technology, and AI as existential trends that Michelin had to embrace — not react to (Chaniot, 2019).


"When leaders fail to model usage, set expectations, or communicate a clear purpose for AI, adoption stalls and skepticism grows. Conversely, when leaders actively demonstrate how AI fits into the company's priorities and incorporate it into their own workflows, teams follow with greater confidence and consistency" (Arkcoll, 2025).


The leadership principle is clear: AI vision cannot be delegated. Leaders must articulate not just what AI will do, but why it matters to the organization's core value proposition — and they must be seen using it.

 

Principle 2: Commit to Innovation — Investment Signals Culture

What the Research Shows

Harvard Business Impact's 2025 Global Leadership Development Study found that 55% of organizations now prioritize generative AI and machine learning as their top digital transformation initiative, up from 43% in 2024. Yet the organizations capturing the most value are those that treat AI not as a cost center but as a strategic investment — allocating resources to talent development, infrastructure, and governance alongside tool adoption.


Leadership in Action

Michelin has operated 9 dedicated R&D centers with 6,000 employees focused solely on innovation since 1965. When Michelin's leadership committed to becoming a data-driven company, they did not purchase an off-the-shelf solution. They organically built a data platform team, an internal data custodian network, a data community, and incrementally constructed a multi-cloud architecture designed to scale. The upfront investment in infrastructure was a prerequisite for every AI capability that followed (Gagnet, 2023).


Jeff Bezos at Amazon institutionalized this principle by treating ideas as assets. His structure drove innovation from the top and established a risk-taking culture where experimentation was constant, data was valued, and learning was formalized. Even when Amazon Go underperformed commercially, the data and insights from the Just Walk Out Technology improved real-time inventory tracking and demand forecasting across the broader retail portfolio (Tucker, 2018).


Commitment to innovation is demonstrated through budgets, organizational structure, and personal behavior — not press releases. Leaders who only signal commitment verbally will see their organizations hedge rather than invest.


Principle 3: Establish a Learning Culture — Failure Is Data

What the Research Shows

McKinsey's research on overcoming generative AI adoption barriers identified a "gardener's mindset" as essential: the most successful managers identify where innovation is already happening organically — employees, teams, or departments experimenting with new technologies — and nurture that growth. Leaders who instead take a "carpenter's mindset" — planning every detail of technological transformation from the top down — find themselves unable to keep pace with the rate of change (Sternfels & Atsmon, 2025).


The same research found that the most effective rewards for AI adoption focus on learning rather than just usage. Organizations that reward employees for demonstrating new competencies, sharing insights, and helping others navigate the learning curve outperform those that simply track AI tool logins. Social recognition — respected team leaders publicly acknowledging they are still learning — reduces psychological barriers for everyone else.


Leadership In Action

Michelin hit three documented data platform barriers during its AI transformation: ineffective metadata governance, a degraded software engineering practice after the shift to cloud, and the absence of data versioning. Rather than treating these as failures, leadership converted each barrier into a documented standard, published it internally, and required all teams to adopt it on a clear timeline. This is learning culture in practice (Gagnet, 2023).


Sam Altman at OpenAI adopted a discovery-driven transformation strategy: releasing AI tools first to small groups, then larger groups, then publicly — collecting feedback at each stage to align product performance with organizational values. When the governance crisis surrounding his November 2023 firing revealed a misalignment between OpenAI's capped-profit business model and its nonprofit board's mandate, the resolution — restructuring the board and giving Microsoft an oversight seat — was itself an organizational learning event (Allyn, 2023).


"When respected team leaders share their AI learning journeys and publicly acknowledge that they're still learning, it reduces the psychological barriers for everyone else" (Sternfels & Atsmon, 2025).


Principle 4: Take an Incremental Approach — Reduce Risk Through Iteration

What the Research Shows

McKinsey's 12 best practices for AI scaling include establishing a clearly defined roadmap with phased rollouts across teams and business units. Fewer than one-third of organizations have implemented most of these practices, which partly explains why so few are realizing full ROI. The organizations that do succeed redesign their workflows incrementally — embedding AI into one process, proving value, then expanding (Singla et al., 2025).


Leadership in Action

Michelin's data platform journey followed an explicit incremental model:



Michelin explicitly stated their principle: "Test what works, obtain new information, and reduce risk" (Gagnet, 2023).


UPS deployed ORION beginning in 2012, then added UPSNav in 2019 for turn-by-turn directions to hard-to-find locations, then added Dynamic Optimization in 2020 to update routes in real time based on changing conditions. Each increment added a measurable capability without risking the entire routing system on an unproven feature (Ross, 2002).


Tim Cook at Apple incrementally expanded Core ML, increasing the number of pretrained models available through Create ML over successive iOS releases — rather than launching a fully autonomous AI platform at once. This gave developers time to build literacy and gave Apple time to refine review processes for AI-powered apps (Podolny & Hansen, 2021).


The practical leadership implication is to resist the temptation to boil the ocean. AI transformations that attempt enterprise-wide simultaneous deployment almost always stall on governance, change management, or infrastructure bottlenecks. Phased rollouts with well-defined KPIs at each stage sustain organizational momentum and build the institutional knowledge needed for the next phase.

 

Principle 5: Align People and Culture — Technical Readiness Is Not the Constraint

What the Research Shows

McKinsey's 2025 workplace survey of 3,613 employees found that C-suite executives are more than twice as likely to cite employee readiness as a barrier to AI adoption than to blame their own leadership. Yet employees — including those McKinsey calls “Gloomers,” its least-optimistic segment — report high levels of openness to generative AI. The research conclusion: Employees are ready. Leaders are the gap (Meyer et al., 2025).


Deloitte's 2025 AI adoption research reinforced this finding: AI workforce transformation and talent readiness is becoming a strategic differentiator. Organizations without in-house AI expertise become vendor-dependent and experience slower adoption during upskilling phases (Ammanath, Kulkarni, and Ritter, 2025).


Leadership in Action

Tim Cook maintained Apple's functional organizational structure — experts leading experts — even as the company scaled 40x in size. Apple's machine learning breakthroughs, including on-device AI that runs with limited compute and no internet connection, required functional leaders with deep AI, compute, and software domain expertise to hold decision-making authority in their respective areas. The functional structure ensured the people closest to the technology made the decisions most consequential to the technology (Podolny & Hansen, 2021).


Michelin confronted a cultural collision when its shift to cloud and data-focused skill sets degraded its software engineering practice. Data scientists and data analysts brought different quality standards — focused on data quality rather than code quality — creating friction with engineering teams. Michelin's response was to explicitly define what engineering excellence looked like in a data-first organization and ensure that both cultures' tooling needs were met in the modernized infrastructure (Gagnet, 2023).


Academics studying AI leadership have identified empowerment and skills development as a critical leadership competency: leaders who emphasize continuous learning create a workforce that is both adaptable and resilient in navigating technological change. This reflects adaptive leadership — the ability to keep people at the center of transformation rather than treating workforce adjustment as a downstream HR task (Zamir, 2025).

 

Principle 6: Govern Responsibly — Trust Is the Foundation of Scalable AI

What the Research Shows

McKinsey's 2026 AI Trust Maturity Survey of approximately 500 organizations found the average Responsible AI (RAI) maturity score at 2.3 out of 4 — an improvement from 2.0 in 2025, but still with significant gaps. Only about one-third of organizations report maturity levels of 3 or higher in strategy, governance, and agentic AI governance. Governance structures are struggling to keep pace with the rapid expansion of AI use (Asaftei et al., 2026).


The consequences of governance failure are not theoretical. McKinsey found that 47% of organizations report having already experienced at least one negative consequence from generative AI. Fewer than half have taken concrete steps to mitigate even their most urgent risks, including inaccuracy, cybersecurity, IP infringement, and privacy violations (Singla et al., 2025).


Leadership in Action

UPS addressed IP and data governance proactively during its digital transformation. As it built an open ecosystem where vendors and customers integrated with its tracking API, it required user registration and established contractual agreements to limit how UPS software could be used externally — protecting its intellectual property while enabling broad adoption (Ross, 2002).


The OpenAI governance crisis of November 2023 is the clearest modern case study of what happens when AI governance fails. OpenAI's non-profit board of directors, whose mission was humanity over profit, fired CEO Sam Altman after judging that he was not candid with them. What followed — 97% of employees signing a letter demanding the board's dissolution, Microsoft hiring Altman the next day, and OpenAI reinstating him 72 hours after the firing — was a governance system that had not kept pace with the organization's growth from non-profit AI lab to the most commercially influential AI company in the world. The resolution required restructuring the board and giving Microsoft formal oversight responsibility (Kerr, 2023).


Researchers studying AI leadership ethics have identified that unlike traditional technological adoptions, AI raises distinct ethical concerns — bias, data privacy, and accountability — that require leaders to establish governance frameworks ensuring responsible and transparent AI practices. These are not legal or compliance tasks to be delegated. They are leadership obligations (Zamir, 2025).


"Trust underpins two critical outcomes: it enables organizations to realize value from AI investments by supporting sustained adoption, and it is essential for successfully managing an expanding and evolving risk landscape" (Asaftei et al., 2026).



Conclusion

AI deployment surged 400% across enterprises between 2024 and 2025. Yet only a fraction of organizations are capturing meaningful value (O'Reilly, 2025). The technology is not the constraint. The organizational capabilities to absorb, deploy, and scale AI — built by leaders who apply these six principles — are.

Organizations that treat AI adoption as something that happens to them — tool by tool, team by team, without strategic intent — will find themselves shaped by the technology rather than shaping it (Hoque, 2026). UPS, Michelin, Amazon, Apple, and OpenAI — each with their successes and stumbles — collectively demonstrate that the organizations setting the pace are those that rewire their leadership, culture, and operating models alongside the technology. As McKinsey put it: "The value of AI comes from rewiring how companies run" (Quantum Black, 2025).


The leaders who will define the next decade will not simply be those who deployed AI first. They will be those who led their organizations through the harder work of becoming ready to use it well.



 

References

 
 
 

Comments

I Sometimes Send Newsletters

© Jorly Metzger. Powered and secured by Wix

bottom of page