The promise of generative artificial intelligence (GenAI) isn’t in the code; it’s in the people. As companies move from proof-of-concept experiments to organization-wide adoption, many struggle not with the AI itself but with the organizational transformation required to integrate it effectively into daily work.
Firms across sectors, from pharma to finance, are running into the same five challenges as they try to scale GenAI. These challenges arise because of mistaken assumptions. The myths these assumptions are based on don’t just slow progress; they shape how people adopt, trust, and apply the technology. Understanding them is the first step toward meaningful, enterprise-wide impact.
In one of the largest implementations of GenAI to date, the multinational pharmaceutical company Novo Nordisk’s journey to scale Microsoft Copilot across 20,000 employees revealed something critical: the biggest challenges with GenAI aren’t technical; they’re about how people think and work with AI.
Here, using Novo Nordisk as an example, we explore the five misconceptions holding humans back from integrating GenAI.
Myth 1: GenAI is all about efficiency
“It will save us time.”
While true, it’s not what people value most. At Novo Nordisk, employees saved 2.17 hours per week using Copilot. But the benefits were broader and more significant: satisfaction was three times more strongly tied to improved work quality. Many workers reinvested their saved time into human-centered work such as collaboration, creativity, planning, and strategic thinking.
Why it matters: GenAI’s biggest ROI may not be speed; it may be the higher quality of work teams are now liberated to do.
Myth 2: Adoption will rise linearly
“Once they try it, they’ll keep using it.”
Not quite. Most GenAI journeys hit a “midcycle dip.” Initial excitement fades. Use drops. Early adopters lose steam. At Novo Nordisk, usage declined after month three. Why? People couldn’t see real use cases – or got discouraged by initial friction. But those who pushed through often reported better results later, thanks to cumulative learning.
What works: Timed training, real-world use cases, and peer champions to reignite momentum.
Myth 3: Digital natives will lead the way
“Younger employees will figure it out.”
At Novo Nordisk, senior employees outperformed their junior peers in both productivity and quality gains. Why? Experience. Seasoned professionals could better spot high-impact opportunities and evaluate GenAI outputs. They had the context and judgment that younger colleagues were still developing.
Try this: Flip the assumption. Invest in experienced users as internal AI champions.
Myth 4: Resistance is rare in digital cultures
“Everyone wants to innovate.”
Not if they feel judged, unsure, or unsafe. At Novo Nordisk, some workers feared being seen as lazy or unethical for using Copilot. Others distrusted the tech, citing privacy, energy use, or fear of hallucinated outputs. This subtle “AI shaming” quietly slowed adoption.
Fix it by:
- Normalizing GenAI use via peer demos
- Creating safe, nonjudgmental spaces for exploration
- Clarifying ethical use and ownership
Myth 5: One rollout fits all
“Let’s deploy it the same way across the organization.”
Different teams = different mindsets = different needs. STEM-heavy teams (such as clinical research) struggled with GenAI’s unpredictability. Sales and corporate staff saw big wins. A one-size-fits-all strategy missed these nuances – until Novo Nordisk tailored enablement by function, role, and mindset.
Bottom line: Personalize support. Build playbooks, use-case libraries, and onboarding by team type.