Archives

Why AI’s Biggest Blind Spot In Pharma Isn’t Technical

This article was originally published by Anna Forsythe in Forbes on 03 March 2026. I spend a lot of time working with artificial intelligence, and I am constantly struck by a contradiction: On one hand, AI has become remarkably fluent. It can summarize dense material, surface insights from massive datasets and produce text that looks confident enough to pass for expertise. On the other hand, in some of the places where accuracy matters most, AI remains almost entirely clueless about real decision making. Pharmaceutical evidence generation is one of those places.

A Systemic Blind Spot

Despite years of excitement about AI in healthcare, there are still no widely used systems that autonomously maintain living, submission-ready evidence for regulatory or reimbursement purposes. This is often framed as a technology gap. In my experience, it is much more than a gap; it is a systemic blind spot not easily remedied. It exposes a misunderstanding of what regulated evidence actually is and what today’s AI systems are built to do. Systematic literature reviews are not summaries. Nor are they answers to questions posed after the fact. They are strictly governed by carefully regulated processes. They begin with predefined protocols, require transparent inclusion and exclusion logic, and must withstand scrutiny long after the original analysis is complete. Regulators and payers do not just ask what conclusions were reached. They ask how those conclusions were formed, what was excluded along the way and whether the same decisions would be made again if the review were repeated. They need to understand how those decisions were made. This is where conversational AI systems struggle in ways that cannot be fixed with better prompts or larger models. Large language models are optimized for plausibility, not traceability. They are designed to produce likely responses, not to preserve the reasoning trail behind each decision. While they can describe evidence convincingly, they cannot reliably explain why one study mattered more than another, or why a marginal trial was excluded at a particular point in time. When the literature updates—as it constantly does in oncology—those inconsistencies compound. I have seen teams experiment with using chatbots to “speed up” evidence reviews, only to discover that what looks efficient at first quickly becomes indefensible under scrutiny. The problem is not that the chatbot models are unsophisticated. It is that the task itself—the special sauce behind the systematic literature review—is not generative in nature. That special sauce needs to be procedural, auditable and accountable.

A Change In Design

What ultimately breaks that stalemate is not better AI, but a different way of designing work. In regulated environments, the only approach that scales is human-in-the-loop intelligence. Machines do what they are good at—continuous surveillance, structured extraction, pattern detection—while humans retain ownership of judgment, interpretation and accountability. When designed properly, this does not slow teams down. But it does change where expertise is applied. What surprises many leaders is that this challenge is not unique to pharma. Several years ago, I had a conversation with an executive in commercial aviation who described a similar tension. Modern aircraft are astonishingly automated. They can take off, navigate complex airspace and land with minimal human input. Yet aviation has never tried to remove pilots from the cockpit. In fact, as automation has increased, pilot training has become more rigorous, not less. The reason is trust. When something goes wrong at 35,000 feet, no one accepts “the system thought it was likely” as an explanation. Automation is expected to assist, not absolve. Human oversight is not a fallback; it is part of the system’s credibility. Evidence generation works the same way. Regulators do not reject AI because it is new. They reject opacity. They expect to see where judgment was applied, and by whom. Systems that blur that boundary undermine trust, even if their outputs look impressive.

What’s Holding Organizations Back?

What ultimately holds organizations back from building these hybrid systems is rarely technology. It’s culture. Most companies are still organized around projects with defined endpoints, not living assets that require continuous stewardship. Evidence is treated as a document to be delivered, not an infrastructure to be maintained. AI initiatives are evaluated on novelty and visibility, not on whether they quietly reduce friction year after year. Changing that requires leadership restraint. It means resisting the temptation to deploy tools that demo well but cannot be defended later. It also means investing in governance, workflow redesign and cross-functional ownership—none of which make headlines, but all of which determine whether AI creates real value. The most important lesson I have learned working at the intersection of AI and regulated decision making is this: Fluency is not the same as reliability. The organizations that succeed with AI will not be the ones that generate the fastest answers, but the ones that can explain and stand behind those answers when it matters. In pharma, as in aviation, intelligence is only as valuable as the trust it earns. And that means human beings in the chain of command.

Pharma’s Biggest Missed AI Opportunity Is Living Evidence

This article was originally published by Anna Forsythe in Forbes on 29 January 2026. At this year’s J.P. Morgan Healthcare Conference, the largest healthcare investment symposium in the industry, it is no surprise that artificial intelligence featured prominently across a wide range of discussions from drug discovery, target identification and molecule design to clinical trial optimization and operational efficiency. AI applications are now fully embedded in each of these core pharmaceutical R&D strategies. What was far less visible, however, was the role AI could play in the systematic evaluation of scientific literature that underpins nearly every strategic, regulatory and reimbursement decision in modern pharma—or evidence generation. This omission is notable, in fact critical, at a time when AI-assisted evidence generation represents one of the industry’s most immediate and measurable opportunities for return on AI investment.

Where AI Is Being Applied Today

Current AI adoption in pharma tends to focus on highly visible areas closely associated with innovation, such as accelerating discovery timelines, improving trial execution and supporting internal productivity. These use cases already demonstrate long-term value and competitive differentiation. Still, the majority of high-stakes decisions in pharma do not hinge on discovery algorithms alone. Instead, they depend, as they have for decades, on structured assessments of existing evidence about disease burden and unmet need, historical endpoints and comparator performance, safety signals and evolving standards of care. These traditional assessments inform decisions ranging from trial design and asset valuation to regulatory strategy and pricing. Despite their importance, evidence workflows remain largely manual and highly fragmented.

Navigating Using Outdated Maps

A useful analogy is navigation. When trying to reach a destination, no one relies on an outdated static map (remember MapQuest?) printed years ago. Roads change, traffic patterns evolve and more efficient routes emerge constantly. Modern navigation relies on GPS systems that update continuously and reroute in real time. Pharma, however, still navigates critical decisions using static evidence reviews. Systematic Literature Reviews (SLRs), which have long been the gold standard for evidence synthesis, continue to be conducted as project-based exercises. This one-off approach is expensive and time-consuming, and the results are quickly outdated as new publications appear, guidelines are revised or new therapies enter the market. Once completed, these product-based exercises often live in disconnected siloes, requiring tweaking or partial reconstruction to support the next decision In a scientific environment that evolves daily, this reliance on static evidence is an increasingly poor and outdated solution, especially at a time when living, continuously updated maps offer a cost-effective solution.

Increasing Regulatory And Reimbursement Pressure

The limitations of static evidence are becoming more consequential as medical reimbursement systems evolve. In the United States, Medicare price negotiations are now in their third cycle under the Inflation Reduction Act. Medicare Part B drugs in oncology, for example, once largely insulated from pricing negotiations, are now fully in scope as of 2026. Manufacturers are expected to justify pricing not only based on evidence available at launch but also relative to new comparators and changing standards of care that continuously emerge over time. In Europe, the Joint Clinical Assessments (JCA), designed to create a unified, cross-national analysis of the efficacy of new drugs, raise needs and expectations further. Companies must consider all relevant comparators across all EU member states, address multiple subpopulations and present comprehensive, transparent evidence syntheses that can withstand scrutiny across multiple jurisdictions. In both settings, evidence is no longer assessed at a single point-in-time. At a time when regulatory and reimbursement demands are continuously being re-evaluated, conventional static snapshots struggle to keep pace with these demands as they evolve.

The Cost Of Fragmentation

Despite this pressure, evidence generation in pharma remains highly fragmented. Different functions (R&D, regulatory, health economics, market access, commercial) often commission their own literature reviews for similar questions. Reviews are modified, repeated and localized across regions, frequently by different external vendors and internal teams. Assumptions diverge. Institutional knowledge is lost. Redundancy accumulates. That redundancy is costly. A single high-quality SLR routinely costs six figures and takes months to complete. For global organizations with large portfolios, the cumulative cost of duplicated effort is substantial. More importantly, fragmented evidence increases the risk of inconsistency at moments when alignment matters most.

Why General-Purpose AI Falls Short

Generative AI tools like ChatGPT and chatbots are often cited as a solution. While useful for summarization or exploration, they are not designed to produce regulatory-grade evidence. Regulatory and reimbursement decisions require predefined methods, transparent inclusion criteria, traceable citations, reproducibility and alignment with established systematic review standards. Outputs must be auditable and defensible. General-purpose AI systems prioritize fluency over traceability and cannot replace structured evidence synthesis. In low-risk settings, speed may outweigh rigor. In regulated environments, rigor is non-negotiable.

The Case For Living Evidence

The alternative is a shift from static reviews to living evidence. A living evidence approach treats evidence as shared infrastructure rather than as a series of isolated projects. Evidence is continuously updated as new data emerges, centrally governed, and organized by indication, population, comparator and endpoint. Updates are incremental rather than repetitive, and changes are transparent. Functionally, this mirrors how GPS systems work: always current, responsive to new information and capable of supporting multiple routes and decisions from the same underlying map. Such an approach could support better decision-making across the product life cycle, reduce duplication and improve consistency under increasing regulatory and reimbursement scrutiny.

Why The Shift Has Been Slow

If the potential benefits are clear, why has adoption been limited? One reason is organizational structure. Evidence budgets are typically allocated by function, by brand and by project. Living evidence, by contrast, is shared longitudinally and is cross-functional. Adoption requires investment at an enterprise level rather than ownership by a single team. Living evidence is also, by its nature, less visible than discovery breakthroughs or novel technologies. Yet visibility and return are not the same. As AI continues to reshape pharma, the most impactful opportunities may lie not only in discovering new drugs faster, but in navigating the increasingly complex evidence landscape more intelligently.