Why science SMEs are not using AI yet

Two in three UK science and technology businesses have not yet used AI in their operations. That statistic surprises many people who assume the science sector leads on technology adoption. The reality is more telling: the barrier is not scepticism about AI's potential. It is the absence of clear, credible use cases that map onto how operational teams actually work.

For decision-makers and operational teams in science SMEs, this gap matters. Competitors are beginning to use AI to process data faster, reduce administrative overhead, and improve decision-making. Businesses that remain on the sidelines will find that gap harder to close over time.

The scale of non-adoption in science and technology

New data reported by Managed IT Magazine places the non-adoption rate at around two-thirds of science and technology businesses. That is a striking figure for a sector that is, by definition, comfortable with complex tools and technical methods.

The same data identifies the leading barriers. Around 7% of respondents cited lack of expertise. A further 7% cited difficulty identifying practical business use cases. Security concerns ranked lower than either of those two factors.

Those numbers deserve a closer read. The barrier is not fear of AI. It is a capability and clarity problem. Teams cannot see where AI would actually help them, and they do not have the internal knowledge to work that out.

What the barriers actually mean

When a decision-maker says their team lacks expertise, they rarely mean they cannot use software. They mean they cannot evaluate which AI applications are worth pursuing, how to assess risk, or how to measure whether something is working.

When an operations director says they cannot identify practical use cases, they mean the generic AI platforms they have explored do not obviously connect to their procurement process, their lab reporting workflow, or their regulatory documentation cycle.

These are not technology problems. They are process and knowledge problems. That distinction matters enormously when choosing how to address them.

Why off-the-shelf tools fall short for SMEs

Generic AI platforms are designed for the broadest possible audience. They offer a blank canvas and expect users to define their own applications. For a large business with a dedicated technology team, that flexibility is useful.

For a science SME with 30 to 150 staff, it is an obstacle. The operations manager, finance lead, or HR professional looking at a general-purpose AI tool faces an immediate question: what exactly should I ask this to do? Without a structured answer to that question, adoption stalls at the experimentation phase.

ONS data on AI in UK businesses confirms that smaller firms consistently report lower adoption rates than larger ones, with capability gaps and unclear ROI cited as recurring factors across sectors.

The pattern is consistent. The tools exist. The willingness to use them often exists. What is missing is the translation layer between a generic AI capability and a specific business process.

The approach that moves businesses forward

Businesses that have successfully introduced AI in operational settings tend to share a common characteristic. They did not start with a platform. They started with a workflow.

The approach works in four broad stages.

First, a business maps its highest-volume, lowest-complexity tasks across departments. Operations, finance, HR, and customer service each contain repetitive processes that consume time without requiring deep expert judgement. These are the natural candidates for initial AI application.

Second, decision-makers define what a good outcome looks like for each candidate process. Not a vague improvement, but a measurable one: time per task, error rate, volume handled, or response speed. This step is often skipped, and its absence is why many AI pilots produce no clear conclusion.

Third, an AI approach is matched to each process. This does not mean choosing a platform. It means defining what the AI needs to know, what it is being asked to produce, and what quality standard applies. A well-structured AI Assistant built around a specific workflow will outperform a general-purpose tool used without structure.

Fourth, the business tests, measures, and adjusts. The first version of any AI-assisted process will surface edge cases and gaps. A short review cycle closes those gaps before they become habits.

This is not a complex programme. For most SMEs, the highest-value use cases are identifiable within a focused half-day workshop. The implementation of the first assistant follows within days, not months.

Why adoption is 80% people and 20% technology

At gecco, we see the same pattern repeatedly. A business spends weeks evaluating AI platforms and arrives at a decision. Then the tool sits largely unused for three months.

The technology was not the problem. The missing element was a clear answer to the question every team member asks silently: what does this mean for how I do my job?

This is why the science sector's reported barriers are so instructive. Lack of expertise and difficulty identifying use cases are both people-and-process problems. They are resolved by structured guidance, not by a better software licence.

Operational teams adopt AI tools when two conditions are met. First, they understand specifically how the tool fits their daily tasks. Second, they have enough familiarity with the tool to trust its outputs in low-stakes situations before applying it to higher-stakes ones.

Both conditions require deliberate onboarding. A new platform dropped into a team's workflow without context will be used minimally, regardless of its capability.

Implementation safeguards worth knowing

UK SMEs planning AI adoption need to address two areas before going live with any AI-assisted process.

The first is UK GDPR compliance. Any AI tool that processes personal data, whether employee records, customer information, or supplier contacts, must operate within a documented data handling framework. This is not an obstacle to adoption. It is a straightforward scoping exercise that most SMEs can complete in a day with appropriate guidance.

The second is governance for responsible use. This means defining which decisions AI can inform and which decisions must remain with a qualified human. In science and technology settings, this is particularly relevant where AI outputs might inform regulatory submissions, quality assurance records, or client-facing reports. Clear rules here protect both the business and its clients.

Sector-specific regulations vary. Science SMEs working in life sciences, clinical research, or environmental monitoring will have additional compliance considerations beyond general UK GDPR. These should be scoped during the use case identification phase, not retrofitted after deployment.

None of these considerations make AI adoption impractical. They make it manageable. Businesses that address governance at the start avoid the costly rework that comes from discovering compliance gaps after a tool is embedded in a process.

Making this work for your business

For decision-makers reading this, the practical starting point is narrower than most expect. You do not need to assess AI across your entire business. You need to identify two or three high-volume processes where a structured AI Assistant could reduce manual effort, and then define what success looks like for each one.

gecco's AI Assistants are built around each client's actual workflows using the GRAFT methodology, with specialist assistants covering operations, finance, HR, and customer service. The starting point is identifying which processes are ready for AI support and structuring the assistant around how your team already works, so adoption happens naturally rather than requiring a change management programme.

Operational teams with clear, process-specific AI tools move from experimentation to consistent use within weeks. The gap between science SMEs that have adopted AI and those that have not is, in most cases, a structured starting point rather than a technology gap.

Start with what your business actually needs

If your team is weighing up where AI genuinely fits in a science or technology business, that is exactly what the AI Readiness survey is designed to surface.

Take the free AI Readiness survey. You will receive 65+ free resources and a custom AI Readiness report based on your answers. From there you can book a free 45-minute AI Readiness call to walk through the results with a gecco advisor.

Website · LinkedIn · Case Studies · Newsletter