Do not treat conference research as a strategy reset for L&D

A Fosway presentation on AI, skills and capability demand is worth noting because it shows where an influential analyst group is focusing attention. On its own, though, it is too thin to support major L&D investment shifts.

The factual base is narrow. David Wilson presented Fosway research at the Learning Technologies Autumn Forum, and the presentation covered AI, skills and capability demands shaping L&D strategy. That shows which topics are getting airtime. It does not show which interventions improve performance, where capability bottlenecks sit, or whether organizations using these approaches get better outcomes.

For a CHRO, CLO or transformation lead, that gap is where strategy drift begins. AI pressure can push teams to buy tools, rename priorities and expand skills taxonomies before they have identified the work that is changing, the decisions that are moving and the manager load that follows. When task mix changes, the issue is often larger than what people need to learn. Work design, decision rights and approval routes may no longer fit the new capability level.

I would begin with one business problem where time to proficiency, quality, throughput or manager burden is under pressure. Then separate task change from training demand by looking at what is still human led, what is AI-assisted, what is AI led with oversight and what is automated. From there, check the constraint around the employee as well as inside the employee. If the bottleneck is workflow, governance or role design, more content will not fix it. If learning is part of the answer, set the smallest evidence threshold that could change the funding decision.

The mistake is to treat a conference research summary as proof that the function should pivot wholesale to data driven L&D, AI enabled learning or skills first operating models. Those may be sensible directions for organizations that adopt them with clear business use cases, but this source does not establish adoption quality, business impact or repeatable return.

The limitation is straightforward: this is a vendor published account of a presentation, with no independent case evidence in the record. I would treat it as attention data, not outcome data. Speed without diagnosis scales waste.