What AI in education will look like in five years

Five-year predictions about technology in education have a poor track record, usually because they either overstate how fast change happens or underestimate how much of the sector resists it for good reason. With that caveat firmly in place, some directions already look clear enough for FE leaders to start planning around now.

From novelty to infrastructure

The current phase, where AI use is patchy, staff-led and often informal, is unlikely to last five years. Expect AI features to become embedded quietly into the platforms providers already use: VLEs, MIS systems and awarding body portals, rather than staff going out to separate standalone tools. This matters for planning because it shifts AI governance from "should we allow this tool" to "how do we configure the AI features already built into our existing systems", which is a different and arguably harder conversation.

Adaptive learning becomes more mainstream

Adaptive platforms that adjust content difficulty and pacing to an individual learner's responses are already used in some functional skills and GCSE resit provision. Expect this to expand into more vocational areas over the next five years, particularly for knowledge-based content that supports practical learning, freeing tutor time for the parts of teaching that genuinely need a human: workshops, observed practice and pastoral support.

Assessment design shifts, not just assessment technology

Rather than better AI detection tools solving the authenticity problem, the more likely path is that assessment design itself shifts toward methods that are naturally harder to complete with undeclared AI help: more in-person elements, oral questioning, staged submissions and workplace-based evidence. This is already the direction JCQ and several awarding bodies are signalling, and it lines up with what actually works today, as covered in Is AI marking safe? Assessment integrity in the age of generative AI.

Regulation catches up, unevenly

Expect more specific regulatory guidance from Ofsted, Ofqual and awarding bodies on AI use in teaching and assessment, but do not expect it to arrive all at once or stay stable. Providers who wait for perfect regulatory clarity before writing their own policy will be waiting a long time; better to have a living policy that gets reviewed regularly, as covered in Writing an AI policy for your college or training provider, than no policy while waiting for certainty that is unlikely to arrive on any fixed timeline.

Accessibility gains that stick

Some of the most durable, uncontroversial AI applications in FE are accessibility ones: real-time captioning, text to speech, reading level adaptation and translation. These face far less ethical and integrity debate than assessment-related uses, and are likely to become standard expectations rather than innovative extras within a few years, similar to how spellcheck moved from novel to invisible.

Staff roles shift, they do not disappear

Concerns about AI replacing teaching roles in FE are largely misplaced for the medium term. What is more likely is a shift in where staff time goes: less time on first-draft content creation and routine admin, more time on the judgement-heavy, relationship-heavy parts of the role that AI genuinely cannot do, questioning technique, pastoral support, workplace assessment and quality assurance decisions. Providers that plan staff development around this shift now will be better positioned than those hoping the change does not affect their provision.

Data and procurement scrutiny increases

As more AI tools embed into core systems, procurement teams and DPOs will need to scrutinise data handling far more closely than in the early, ad hoc phase of individual staff trying free tools. Expect more formal AI risk assessment as part of procurement processes, similar to existing data protection impact assessments, becoming standard practice rather than best practice.

What this means for leadership now

Leaders do not need to predict every detail correctly to prepare well. Three things are worth doing regardless of exactly how the next five years unfold:

• Build a policy and review cycle that can absorb change, rather than a static document, as detailed in the policy piece linked above
• Invest in staff CPD on AI literacy now, since the skills gap between confident and hesitant staff is already wide and will only grow
• Treat data protection and procurement scrutiny of AI tools as a permanent, ongoing task rather than a one-off check

Building capability ahead of the curve

Providers who invest in structured staff development now, rather than waiting for the picture to become clearer, will be better placed to adapt as tools and regulation shift; our qualifications and CPD courses increasingly reflect this by building digital and AI literacy into leadership and teaching qualifications rather than leaving it as an optional extra.

A realistic close

The next five years will likely bring AI further into the infrastructure of FE without radically changing what good teaching looks like. The providers who do well out of this transition will be the ones who treat it as a gradual, manageable shift requiring ongoing attention, not a single decision to make once and forget.