A new survey from Sapio Sciences finds today’s researchers are largely frustrated with electronic lab notebooks (ELN), driving loss of time and money, repeated experiments, inefficient data use and a growing reliance on unauthorized shadow AI.
For the survey, Sapio Sciences interviewed 150 life sciences professionals who run experiments and work inside ELNs across biopharma R&D, manufacturing, diagnostics and CROs.
The shortcomings of ELNs
When ELNs first entered the laboratory, they were hailed for their ability to replace paper and support documentation. At the time, it was a feat; but, times have changed. The laboratory is now entering its artificial intelligence (AI) era. A byproduct of this is that yesterday’s ELNs may not live up to the expectations of the new modern laboratory.
Traditional ELNs focus on recording experimental outputs—but not interpreting the data. In many labs, scientists are still waiting on IT tickets before they can run or adjust standard assays. In fact, 67% of survey respondents said they rely on IT to configure ELN assays or templates more than 25% of the time. Similarly, 67% rely on data scientists or biostatisticians to interpret ELN results more than 25% of the time. Overall, just 5% of respondents report being able to analyze experimental results without specialist support.
Additionally, duplication is a persistent issue. Nearly two-thirds of scientists, 65%, say they have had to repeat experiments because prior results were difficult to find or reuse, driving avoidable costs and delays across teams.
“The survey clearly shows a growing mismatch between modern scientific practice and the capabilities of traditional ELNs,” said Mike Hampton, chief commercial officer at Sapio Sciences. “Today, scientists are working with increasingly complex data and are expected to move from results to decisions faster than ever, yet many ELNs still function like glorified filing cabinets.”
The rise of shadow AI
These frustrations are reshaping behavior in the laboratory. Almost half of scientists surveyed, 45%, say they use public generative AI tools through personal accounts to support their work, despite the security, IP, and compliance risks associated with shadow AI. In fact, 77% of survey respondents still report the use of ChatGPT, Claude or Gemini in their work. On the other hand, 32% use public generative AI via company-managed logins.
“Scientists aren’t turning to public AI because they want to bypass governance. They’re doing it because existing lab tools can’t help them analyze results or determine next steps efficiently,” said Sean Blake, chief information officer at Sapio Sciences. “When AI capability isn’t available in governed environments, people will find it elsewhere, even when they do understand the risks. Shadow AI is a signal of unmet need.”
However, today’s public AI tools do not have a deep knowledge of laboratory workflows—especially your specific laboratory needs.
Even with their popularity, only 27% of survey respondents say current generative AI tools meet their scientific needs very well, and 15% say they are poorly suited to scientific tasks and workflows.
“Generic models are good generalists, but the lab needs a specialist,” read the Sapio report. “AI [tools] only become credible co-scientists when they are built around scientific data and integrated with lab systems in ways that support regulated work.”
The AI electronic lab notebook
When asked what they want from the next generation of ELNs, scientists consistently emphasized interaction, guidance and interpretation. Ninety-five percent want conversational, text-based interfaces, while 78% want voice interaction. Almost all respondents, 96%, said future ELNs must help interpret data, not simply capture it. Scientists also want built-in, field-specific AI capabilities, with demand varying by discipline.
“The findings suggest scientists are not looking to relinquish control but to work with AI tools that actively support reasoning and interpretation within governed lab environments,” said Rob Brown, head of the scientific office at Sapio Sciences.
Thinking about an AI lab notebooks, scientists highlight analysis and connectivity as the foundation. They named the following features as key AI developments over the next 12 months:
- ability to analyze and interpret results and highlight trends, anomalies and relationships (52% of survey respondents)
- intelligent experiment planning and design (39%)
- ability to connect data across instruments, ELN and analysis tools to reduce manual work (45%)
- generate novel insights by linking structured and unstructured lab data (39%)
Additionally, 79% of survey respondents said they want AI-driven SOP execution directly from protocols, illustrating that scientists also expect the notebook to help run the work, not just report on it. Today’s scientists are envisioning the next-generation ELNs as intelligent research partners that support hypothesis generation and experimental design rather than data input engines.
“The mandate from the bench is clear: an AI lab notebook that works with the scientists, not around them. Its role is to turn the hypothesize, design, plan, act and analyze loop into a connected workflow across instruments, data and analysis, not just provide an AI veneer on top,” concludes Sapio’s report.