6 tips to use generative AI for learning and development

Using generative AI for learning and development can help you streamline administrative tasks, deliver more engaging learning experiences, and close skills gaps faster. 

But if you want to truly integrate AI into your learning strategy and use it to drive ROI, it will take more than adopting AI tools on an ad hoc basis.

In this article, experts explain how to overcome challenges and strategically implement AI for learning and development.

Learn how AI will change the future of learning.

From AI assistants to greater personalization, generative AI can undoubtedly benefit the learning process. But it also comes with unique challenges that you and your people will need to overcome to use the tech effectively.

AI isn’t perfect—far from it. It’s only as good as the data it was trained on, and even then, it can make mistakes.

“Even with agentic AI agents (that are grounded and specialized), AI can get it wrong. And that wrong-ness is not usually in the form of an obvious hallucination,” explains Amy Coughlin, Principal Cloud Author at Pluralsight.

“One of the most common errors I've seen is the response to questions such as, ‘Does the Acme Widget support break dancing?’ To which the LLM says, ‘No. It supports waltzing but not break dancing.’

“By pointing out what it does support, the LLM sounds more authoritative, when, in fact, the Acme Widget does not support break dancing.”

Human oversight is critical to catching these mistakes and ensuring accuracy. Asking the models to cite their sources can also help reduce AI hallucinations.

It may be tempting to use AI to measure your people’s skills, but AI-generated assessments can’t fully capture your people’s knowledge or willingness to grow. 

Instead of relying on AI, use validated skill assessments to accurately understand their skills gaps, strengths, and areas of opportunity.

“With personalized genAI training can come predictive assessments, which are likely wildly inaccurate and harmful. Sure, measure candidates' current knowledge, but ignore any artificial metrics on their capacity for knowledge,” says Ian Marshall, Senior Software Development Author at Pluralsight.

When people overrely on genAI, their skills can suffer—which is the exact opposite of what you’re trying to achieve with AI upskilling.

“The biggest downside is that it's so attractive to human brains to have something that can do all the work for you—and that's a really bad thing for learning. If I ask generative AI to summarize meeting notes, and then take that summary and produce a message with some action items or answers to send to another team member, what have I really done?” says Jon Friskics, Principal Software Development Author for Pluralsight.

“If I already understand all the moving parts, and it's just helping me reduce the amount of typing I need to do, then that's great. But it's way too easy to remove the thinking part and jump right to having some output.”

To combat the resulting skill decay, understand where AI shines as a learning assistant and where it falls short. While it’s helpful for synthesizing information, answering questions, and personalizing learning content, it doesn’t replace hands-on application.

“AI is not a substitute for actually learning a topic. Most technical topics can only really be learned by actually doing them, by gaining hands-on practice in configuring, implementing, testing, and troubleshooting. AI is not a shortcut to gaining real-world experience,” emphasizes Faye Ellis, AWS Hero and Pluralsight Fellow.

To overcome AI challenges and fully reap the benefits of generative AI for learning and development, experts recommend starting small, implementing AI governance controls, and prioritizing your people throughout the transition.

Consider this your AI implementation guide for L&D:

When you’re ready to implement generative AI as part of your learning strategy, start small and plan carefully. 

“We've all experienced hallucinations and even extreme fails when it comes to AI output, but that doesn't mean it should be ignored as a learning tool. I do think it has a place; for instance it can really shine in helping learners explore established concepts,” explains Faye. 

“Where it can fall down sometimes is when dealing with cutting edge technologies, new SDKs, APIs, and programming languages or technologies that are constantly being updated. Generic models that have not been customized with up-to-date, relevant, domain specific data are going to struggle to keep pace with the latest innovations, resulting in unreliable responses that mislead and frustrate learners.”

Avoid using default or uncustomized LLMs. 

“Using vanilla ChatGPT (or Gemini or Copilot etc.) increases the possibility of getting wrong or limited information. It's better for the content to be grounded or find other ways of providing guidance and context to the AI model,” says Amy.

Providing enterprise-wide access to an LLM can also help cut down on shadow AI use and unauthorized data sharing.

“I think many employees will just buy their own plan if it's not provided to them. IP governance is an issue—if employees are pasting documents or code into generative AI tools, then that's potentially a problem for that company's data and IP security,” says Jon.

When you’re using generative AI for learning and development (or anything else!), governance and human oversight are key to safe, responsible use in your organization.

“Do not minimize the human component for authenticity, impact, and judgement,” says Peter Barrett, Learning Solutions Architect at Pluralsight. “Be aware of, and implement structure around, bias, ethics, and both domestic and international regulations concerning the use of genAI.”

The good news is that you don’t have to create everything from scratch. 

“Most genAI solutions now have governance controls in place. It's time to start adopting these tools to help protect company data including intellectual property,” advises Wayne Hoggett, Principal Author, Cloud, at Pluralsight.

Learn how to prepare your organization for EU AI Act readiness.

Adopting tech alone isn’t enough to drive ROI. You also need to help employees adjust to change, get buy-in from executive leaders, and prove the value of generative AI before allocating more resources to it.

“Leaders need to define how genAI will be implemented, including the tasks of change management for AI adoption and team and stakeholder education,” says Peter.

“Solid results, data, and examples will be needed to win the support of the executive team. Focus on small wins with low-risk pilots, such as creating a job ad, inventing a scenario, or building programming workflows for a lab, that can demonstrate confidence in and effectiveness of genAI, before tackling larger initiatives.”

Upskilling initiatives need to come from the top. But employees also look to their peers for guidance. AI champions can generate excitement and help bring more learners on board.

“In a perfect world, a large-enough organization could appoint oracles, or employees that embody the best-in-case human expertise in a domain, and empower them to help plan and evaluate the information that more novice employees are gathering from generative AI tools. Even just pointing learners in the right direction would be helpful—like starter prompts including guardrails to help eliminate tangents,” says Jon.

As generative AI automates entry-level tasks, junior tech roles are starting to disappear. It’s important that you continue to invest in and develop entry-level talent.

“We still need senior devs and tech experts to oversee, correct, and remove risk from code written by genAI. So, my advice is to invest in expanding the knowledge base of your junior devs and tech experts. After all, tomorrow they will become your senior-level team, and without them, all that risk is there to stay,” says Ian.

Despite AI challenges, building it into your learning strategy can help you engage learners and deliver upskilling value faster.

So don’t be afraid to use it! With thoughtful implementation, it can augment—not replace—your existing initiatives and human capabilities. 

“There is a lot of lived experience, call it wisdom if you want, that you can get when learning from a human with that experience that you just can't get from genAI,” says Wayne.

Bring AI governance, skills, and outcomes together. Build your team’s AI readiness with Pluralsight AI Academy, our end-to-end skill building program. Learn more.