Beyond the Hype: What Actually Works When You Bring AI Into Learning & Development
If you had to describe your relationship with AI right now using one word, one emoji, or one GIF, what would you pick?
That's exactly how Tracy King, founder of Inspired Ed, opened last week's session with Forj, Beyond the Hype: Practical Use of AI in Learning and Development. And the answers said it all. Some attendees dropped in "excited." Others said "overwhelmed." One admitted "fast" was the only word that fit. A few just sent an emoji and let it speak for itself: 🤯 😨 🤠 🤪
The most common answer, by far, was some version of excited. The second most common feeling running through the chat: overwhelmed. If that's where you land too, you were in good company. And that tension between excitement and overwhelm turned out to be the whole point of the hour.
Here's what L&D teams need to know.
Start with the foundation, not the tool
Tracy's biggest warning: the organizations struggling with AI right now usually skipped the groundwork. Before any tool gets onboarded, teams need clear answers in three areas:
- Governance and risk — who's accountable for ethical use, and who's watching for it
- Operations and policy — what's okay to use AI for, what isn't, and how that gets enforced day to day
- Resources — the time and budget to actually train people, not just hand them a login
Skip this step, and you end up with what Tracy called shadow AI use: staff quietly experimenting with public tools, sometimes with sensitive content, because no one gave them a real alternative. She pointed to a well-known case where an engineer pasted proprietary code into a public GPT window, not realizing it was now part of a public model. It's a good reminder that most AI risk isn't malicious. It's just untrained.
Design and production are not the same job
This was the framework that seemed to land hardest with the group: AI can speed up production, but it can't replace design.
Design work is yours. Understanding your learners, sequencing a pathway, deciding what "good" looks like for your audience: that judgment doesn't come from a prompt. Production is where AI genuinely shines: drafting, formatting, generating first-pass assessment items, building media.
The catch? When teams blur the two, they either expect AI to do more than it can, or they undervalue the design work that makes any of it useful. Tracy's advice: draft fast with AI, but budget real time for humans to review for accuracy, voice, and whether the content is still serving the actual learning objective.
Faster drafts don't mean faster projects
One of the more honest moments of the session: AI can produce five times the content your team used to make in a week. That doesn't shrink your review workload. It grows it. Every draft still needs a human checking for accuracy, bias, and accessibility gaps, things like color contrast, alt text, and screen reader compatibility that can look fine on the surface and still fail in practice.
The fix isn't slowing down. It's building validation into the workflow from the start, and being upfront with stakeholders that "AI made it faster" and "AI made it effortless" are two very different claims.
The skills worth building next
Tracy closed with five areas L&D professionals should be investing in as AI takes on more of the production layer:
- Prompt engineering — better inputs, better outputs
- Output evaluation — knowing what to check for, and building the habit of checking it
- Workflow design — deciding deliberately where AI fits in your process, rather than bolting it on
- Data fluency — reading learner analytics ethically and drawing real conclusions from them
- Custom GPT design — packaging your organization's own content into tools members can query directly
Her point wasn't that any of this replaces instructional design expertise. It's that it sits on top of it.
Missed the live session?
The full recording is available now. If you're anywhere between "messy middle" and "all in," it's worth the watch.
