Is AI Literacy Alone Enough to Deliver Value?

Many London financial services firms have invested heavily in AI literacy over the past two years.

Awareness sessions, e-learning modules and tool demonstrations have been rolled out across trading floors, operations teams and back-office functions, often driven by regulatory expectation and the government's Skills Compact.

Yet a common pattern emerges six or twelve months later. Attendance figures look strong, satisfaction scores are positive, and yet very little has actually changed in how work gets done.

This is not a sign that literacy training has failed. It is a sign that literacy was never designed to deliver value on its own.

Why the investment often stalls

Firms across London are under considerable pressure to demonstrate workforce AI capability.

Regulators expect firms to understand the tools they use. The Skills Compact sets a broader expectation that the financial services workforce will be equipped to work alongside AI. Competitors are moving, and boards are asking questions.

The natural response has been to commission literacy programmes: structured training that explains what AI is, how large language models work, and what generative tools can and cannot do.

This is a sensible first step. But operations leaders frequently report the same outcome. Staff can describe AI concepts confidently in a workshop, yet nothing changes in the reconciliation queue, the underwriting file review, or the client onboarding process the following Monday.

The gap is rarely the training content. It is the absence of everything that needs to surround that training for it to matter operationally.

Why training programmes alone were never designed for this

Traditional compliance and skills training follows a familiar pattern: define the required knowledge, deliver it, test understanding, record completion.

This works well for topics with a fixed, stable body of knowledge, such as financial crime regulations or market abuse rules. Once someone understands the rule, they can apply it consistently because the workflow around them already expects and enforces that application.

AI literacy is different in one important respect. The workflows that AI is meant to improve usually have not changed at all. Staff finish the training and return to exactly the same process, using exactly the same systems, with no expectation from their manager that anything should be done differently.

Awareness without a changed workflow rarely produces changed behaviour. This is not a flaw in the training. It is a limitation of treating literacy as a complete solution rather than a starting point.

Where AI-specific enablement genuinely helps

This is where the journey from literacy to value can accelerate, provided it is approached deliberately.

Embedding AI tools directly into existing workflows, such as a reconciliation dashboard or a policy review screen, gives staff a concrete, low-friction opportunity to apply what they learned. Coaching delivered in the flow of work, rather than in a classroom, reinforces literacy at the point of use rather than weeks before it.

AI can also help firms identify where literacy is and is not translating into behaviour, by surfacing patterns in how tools are actually being used across teams. This gives learning and development functions something literacy training alone cannot: a feedback loop.

None of this replaces the need for governance. If anything, embedding AI into live workflows makes governance more urgent, not less, because staff are now making judgement calls about when to rely on AI output and when to escalate.

What needs to be in place operationally

Converting literacy into value requires a small number of organisational conditions that are frequently missing.

Someone needs to own the outcome, not just the training completion rate. Without a named owner accountable for downstream impact, literacy programmes tend to be measured by attendance rather than by any change in operational performance.

Workflows and manager expectations need to change alongside individual skills. If a manager never asks whether AI was used appropriately in a piece of work, staff have little reason to change their behaviour regardless of what they learned.

Governance and oversight structures need to exist before staff are encouraged to apply AI tools independently. This does not need to be complex, but it does need to exist: a clear point of approval, a defined escalation path, and clarity on where human judgement remains final.

Firms that put these conditions in place alongside literacy tend to see the gap close quickly. Firms that treat literacy as the finish line tend to see engagement without impact indefinitely.

Example

A London-based commodities trading firm ran a firm-wide AI literacy programme covering large language models and generative AI tools for all operations staff.

Six months later, adoption in daily settlement reconciliation work remained negligible, despite high satisfaction scores from the training itself.

On review, the Head of Operations and Learning and Development Manager found that no workflow had been redesigned to incorporate AI-assisted checks, no manager had been briefed on how to supervise AI-assisted reconciliation, and no governance checkpoint existed to approve its use.

Once the firm assigned workflow owners, added a lightweight approval step, and set a measurable target of reducing reconciliation exception time, AI-assisted settlement checks were adopted within weeks and exception handling time fell measurably. Literacy became value only once embedding and governance were added.

FAQs

• Why doesn't AI literacy training alone improve business outcomes?

  Literacy builds individual awareness and confidence, but it does not by itself change workflows, manager expectations or oversight structures. Those are the elements that actually drive measurable outcomes, and without them staff typically return to unchanged processes after training ends.

• What should firms add to their literacy programmes to see real value?

  Firms need to embed AI use into everyday workflows, establish clear governance and oversight before independent use is encouraged, and measure programmes against business outcomes rather than attendance or completion rates.

• How does the Skills Compact relate to the literacy-to-value gap?

  The Skills Compact sets an expectation that the financial services workforce will be equipped to work alongside AI. Meeting that training expectation is not the same as achieving operational value. Firms need both workforce capability and the organisational conditions that convert it into measurable impact.