Author: Ross Garner
‘The pace of change is accelerating.’ I’ve heard this phrase a hundred times in conference sessions, podcasts, and in conversation. But it drives me nuts.
It’s rarely connected to anything interesting and is usually spat out as a mechanism to inspire panic and a sense of falling behind. Then followed by a pitch to sell me something.
So over the last few months, I’ve been speaking to some of our long-term custom learning customers to ask this question: How are the times changing for you?
In interviews with 11 customers across technology, finance, professional services, consumer goods, media, energy and pharma sectors, I wanted to find out how organizations are actually thinking about AI. How are they using it, what are the barriers they’re experiencing, and what kind of results are they seeing.
The headline finding: we’re all still figuring this out.
Attitudes to AI for content production vary
Most customers I spoke to are pragmatic about AI. They see the benefits for producing content, but usually with a degree of scepticism.
The expectation is that AI-generated learning content from a vendor be at least as good as human-generated content, and preferably better. Most expect AI to have little impact on pricing. A small number expected costs to fall.
There were outliers. Two of the organizations I spoke to are experimenting with vibe-coding their own learning platform, but see these as experiments with real governance concerns. One was resistant to AI in any form.
Governance is a live issue
Almost every organization I spoke to has some mechanism for assessing how AI is used, usually sitting with IT or a governance committee. In almost every case, it’s more concerned with risk than productivity. These structures exist to protect the organization, not champion adoption.
What customers want most is clarity: what data is being fed into AI systems, and what safeguards are in place. Absent documented reassurances, the fear factor will slow everything down.
AI for learning is rare
Most L&D use of AI remains focused on faster or cheaper production. Using AI for personalization, evaluation, summarisation, or tutoring within a course was the exception, not the rule.
When I shared examples of our own work in this area, customers were excited but wary. They wanted to explore further using their own company API keys, and with governance sign-off in place.
Compliance learning adds another wrinkle. Generative AI’s greatest strength, tailored content for each user, is also its greatest liability in a regulated environment. If every user experiences something different, the compliance team can’t say with certainty what anyone has read. In a legal dispute, that ambiguity is a serious risk.
There’s a useful distinction emerging though. An AI tutor that answers employee questions, constrained to approved policy documents, is seen as genuinely useful. A course that ticks a regulatory box still needs a fixed script that legal has signed off.
What stayed constant?
Most reassuring to me was each customer’s focus on learning impact. Across the board, customers care about the quality of the learning experience and its alignment with real performance goals. Performance consulting and measurement remain high on agendas.
There’s no doubt that the times are a-changin’. But based on these interviews, most organizations are in a similar place: partial tooling, low risk appetite, and a strong desire for guidance.
What this means for you
The learning profession has three keys jobs: diagnose the need, deploy a solution, measure the outcome.
The problem, historically, is that the second job takes too long.
AI is changing that, and that creates space. The question is whether you use that space to do more of the same, faster, or to do more of what has always mattered most: understanding the performance problem, and demonstrating that the solution worked.
The tools have changed. The job, at its core, has not.
If one thing came through clearly in these conversations, it’s that most organizations want guidance as much as technology. That’s one of the reasons we created the AI Skills Academy: to help people build practical AI skills with confidence.
Deep dive
As AI-generated videos become the norm, it’s worth asking if training can help users spot this.
In 2021 (prior to the advent of generative AI), researchers investigated this question with a group of adults who had limited exposure to the technology.
First, they assessed what people actually look at when judging whether a video is fake, and found that they spent too much time looking at eyes, hair and background, and not enough time looking at background details.
They then built a 10-minute training module, pointing out specific telltale signs.
A group of 46 participants were then split into two groups: one who received the training, and one who didn’t. Both groups scored about the same on a pre-assessment (55-58% accuracy), but the group who received training jumped to 88% in the post-assessment.
Interestingly, the group who received training actually scored lower on identifying real videos in the post-assessment. This indicates that while training can help users spot deepfakes, it also increases their suspicion of videos in general. Read the full article
Tahir, R., Batool, B., Jamshed, H., Jameel, M., Anwar, M., Ahmed, F., Zaffar, M.A., Zaffar, M.F., 2021. Seeing is Believing: Exploring Perceptual Differences in DeepFake Videos, in: Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems. Presented at the CHI ’21: CHI Conference on Human Factors in Computing Systems, ACM, Yokohama Japan, pp. 1–16.
This blog was originally published in our L&D Dispatch newsletter