51% of respondents describe themselves as advanced when it comes to implementing AI.
Why AI Projects Fail in Life Sciences
Article
And How to Make Yours Succeed
Life sciences are not short on AI ambition. With their knowledge-intensive workflows, deep data assets and highly structured operating environments, pharma companies are uniquely positioned to realize value from AI.
Across the industry, leaders are investing in AI:
14%
Companies expect to spend 14% more on AI initiatives than in years past, according to EPAM’s 2025 AI report.
And the industry is confident in its AI adoption thus far:
51%
8%
8% of Life Sciences firms describe themselves as disruptors.
Successful AI initiatives within pharma and biotech hinge on three areas:
01
Identifying value and building the right frameworks
02
Crafting — and implementing — a winning business case
03
Quantifying value and scaling success
Here we examine the first of these three pillars with a focus on strategy, governance and data. Why do businesses struggle to identify value? What foundational elements are missing? And most importantly, how can pharma companies avoid those pitfalls while building an AI strategy designed for true impact?
THE NUMBERS
74%
74% of companies struggle to achieve and scale value when adopting AI.
9%
While 63% of companies are piloting GenAI, only 9% are deploying it at any significant scale.
65%
65% of disruptors agree or strongly agree that they know the skills necessary for AI adoption, yet 46% of those disruptors believe their staff needs to upskill.
THE PROBLEM
Many organizations adopt a scattered strategy where they test hundreds of disparate use cases before strategically evaluating their existing technological maturity. This approach dilutes resources, creates technical debt and prevents meaningful scale.
THE IMPERATIVE
Successful Companies:
Adopt a focused, high-value approach.
Conduct a rigorous audit of their current data infrastructure, talent capabilities and technological readiness.
Prioritize projects that align with their immediate capabilities while setting the groundwork for long-term structural improvements.
THE NUMBERS
39%
Despite 88% of organizations citing the presence of a data strategy, only 39% agree that their data strategy is completely aligned with key business objectives.
2.4x
Businesses with a defined vision are 2.4 times more likely to achieve their objectives effectively and drive sustained growth (Forbes).
THE PROBLEM
AI initiatives frequently launch as isolated experiments rather than strategic business drivers. Leadership engagement and a clearly defined AI vision aligned with business objectives are essential to success. Without leadership-driven change, businesses fail entirely.
THE IMPERATIVE
Successful Companies:
Align every major AI initiative with core business goals to ensure relevancy and impact.
Have executive-level leadership buy-in to drive accountability and inspire organizational support.
Ensure both business and IT ownership and drive cross-functional alignment.
THE NUMBERS
88%
88% of companies report regular AI use in at least one business function.
40%
Over 40% of agentic AI initiatives are expected to be canceled by 2027 due to unclear business value and governance challenges.
3x
Companies with senior leadership commitment are 3x more likely to achieve AI value.
THE PROBLEM
Centralized AI teams often work in isolation, developing solutions based on technical potential rather than real business pain points. When business stakeholders are excluded from the early ideation and design phases, projects risk misalignment with practical workflows and actual end-user needs. This disconnect means critical business context is overlooked and adoption barriers are underestimated.
AI solutions that are imposed top-down without early buy-in are far less likely to solve core problems or gain traction. Misunderstandings about workflows, data availability and regulatory realities can result in expensive initiatives that stall or get shelved, compounding skepticism about future AI investments.
THE IMPERATIVE
Successful Companies:
Involve business stakeholders from day one by embedding them in project ideation, requirement gathering, and solution design.
Establish cross-functional steering committees to ensure alignment between AI strategy, end-user needs, and compliance requirements.
Co-create and iterate AI solutions in partnership with business owners to maximize relevance, drive adoption, and ensure measurable impact on operational and clinical outcomes.
THE NUMBERS
51%
51% of life sciences organizations cite resistance to change as the top barrier to AI adoption.
58%
58% of employees haven’t used AI at work.
24%
Just 24% are confident in their organization’s AI policies.
THE PROBLEM
Even when executives endorse an AI project, adoption can stall if end users don't trust or understand the technology. Clinical researchers, data scientists and compliance officers may see AI as a disruption to their daily work rather than a tool that can help them.
THE IMPERATIVE
Successful Companies:
Champion Aggressive Change Management
Involve end users early in the development process to ensure the tools solve actual daily friction.
Provide Comprehensive Training Programs
Establish clear communication channels to foster understanding and skill development.
Demonstrate Early Wins
Build momentum and trust among your workforce by showcasing the technology's initial successes.
THE NUMBERS
18 months
On average, businesses expect it will take them 18 months to roll out an effective governance strategy for AI.
68%
68% of companies acknowledge lacking the necessary governance to manage AI risk.
37%
Only 37% have established an AI ethics committee to guide these initiatives.
THE PROBLEM
In order to see change, businesses must establish governance now. Waiting a year and a half leaves organizations exposed to severe security risks and regulatory breaches. This reactive approach is a critical misstep in an era defined by rapid technological advancement.
THE IMPERATIVE
Successful Companies:
Define clear roles, responsibilities and decision-making processes.
Consider which type of governance model works best for your organization.
Address ethical considerations, bias mitigation, model management, risk management and regulatory compliance early before they become disruptive.
Governance in Centralized, De-centralized & Hybrid Models
Decentralized or Federated Governance
Decentralized or Federated Governance
Central CTO or enterprise teams provide guidance, and divisions own workflows.
Common Context: Typical in organizations with strong divisional independence and diverse business models.
Hybrid Model
Hybrid Model
Central AI strategy with business unit-level autonomy, often supported by external expertise.
When it Fits: Common in organizations with independent business units and differing needs or ways of working.
Centralized Governance Committees
Centralized Governance Committees
A single cross-functional body oversees AI.
What it Controls: Approval workflows, risk tiers, documentation standards and ongoing monitoring.
THE NUMBERS
57%
57% of life sciences professionals say data silos are their top technical challenge, a barrier that consistently derails AI initiatives and curbs value creation.
80%
80% of organizations report poor data quality or availability, leading to an average loss of $12.9 million annually.
49%
49% of experts cite ontology gaps blocking FAIR (Findable, Accessible, Interoperable, Reusable) data management.
THE PROBLEM
AI requires accessible, high-quality and harmonized data, but most life sciences organizations struggle with data that is fragmented across silos, subject to inconsistent standards and rarely aligned with FAIR principles. Critical information is scattered across disconnected legacy systems, hindering collaboration and making it nearly impossible to build the reliable datasets needed for robust, compliant AI models.
This fragmentation slows innovation and erodes trust in AI outcomes. When data cannot be easily found, trusted or reused, teams spend excessive time cleaning and prepping information rather than developing new solutions. The net effect is wasted resources, stalled initiatives and missed opportunities.
THE IMPERATIVE
Successful Companies:
Invest in enterprise-wide data integration platforms and federated data models to break down silos and enable secure, centralized access while protecting compliance.
Establish robust data quality frameworks and company-wide semantic standards, making information consistent, interpretable, and reusable across the organization.
Embrace and operationalize FAIR data principles (Findable, Accessible, Interoperable, Reusable) to enable seamless data discovery, sharing, and scaling of insights throughout the enterprise.
THE NUMBERS
85%
85% of pharma leaders report inadequate data and technology foundations hinder AI adoption.
50%
50% of genAI projects were abandoned after proof of concept due to poor data quality and shifting business priorities.
81%
81% report achieving analytic accuracy and operational efficiency as the top barrier to implementation.
THE PROBLEM
AI initiatives in life sciences often fail because slow, fragmented development cycles can't keep pace with the rapidly evolving landscape of research, regulation and technology. By the time solutions are ready for deployment, the original business problem may have changed or the technology is already outdated.
This "too little, too late" dynamic is worsened when teams work in isolation or stick to rigid assumptions, leading to "solution obsolescence." This often results in wasted investment and increased technical debt.
THE IMPERATIVE
Successful Companies:
Commit to agile, iterative development processes that allow rapid adjustment of project goals and technical approaches as business needs and technologies change.
Schedule regular alignment reviews with business stakeholders to ensure ongoing relevance and value.
Build modular, upgradable AI architectures that enable swift integration of new data sources, algorithms, and compliance requirements as they arise.
Foster a culture of continuous learning and adaptation, so teams can pivot quickly and maximize the long-term value of their AI investments.
THE NUMBERS
71%
71% of CTOs and CIOs reported facing unrealistic ROI demands from leadership regarding AI initiative.
80%
While 80% of respondents set efficiency as an objective for AI initiatives, the most successful organizations also prioritize growth and innovation, avoiding a narrow focus on cost-cutting.
50%
50% of GenAI budgets go to sales and marketing, but back-office automation often yields better ROI.
THE PROBLEM
AI is often seen as a cure-all for complex business problems, but this sets projects up for failure. Underestimating the need for quality data, iterative development and human oversight — while overestimating AI's speed as a "plug-and-play" solution — drives poor planning, stakeholder disengagement and increased regulatory risk.
THE IMPERATIVE
Successful Companies:
Set clear, realistic expectations at the outset by educating stakeholders about the true capabilities and limitations of AI, including where human expertise, oversight, or consensus remains vital.
Encourage an experimental, learning mindset — frame AI adoption as an evolving journey, not a single transformational leap — so initial setbacks become opportunities for improvement.
Build in periodic project reviews focused on aligning technical progress with business value and communicating results honestly, fostering transparency and organizational learning at every stage.
CONTRIBUTORS
BERT
CARDOEN
Senior Consultant,
Advisory
DOMINIK
RIEDT
Senior Consultant,
Advisory
SASCHA
ROSENBERG
Senior Consultant,
Advisory
DR. MAXIM
POLIKARPOV
Senior Manager,
Advisory