How Can AI Help (or Hinder) Learning Content Development?

How can AI help with learning content development?

Most people agree that learning is important and necessary. And that it's always number six in the list of the top five things to do! Learning and development is always the first to be cut - never gone - just cut to a nub.

Most learning teams don’t have all the resources they wish they had, which is why AI quickly got their attention as a practical support tool. Even in the earliest days of ChatGPT, chat models were surprisingly good at creating multiple-choice questions, especially plausible but incorrect answer options that sounded believable without being right.

Another thing AI is amazing at is taking existing content and changing it, making it longer, making it shorter, creating an outline, or reapplying something to a slightly different audience. For sure, the human learning experts still need to be part of that whole process, but AI is good at advising on how to tweak it slightly. 

How can AI hinder learning content development?

Most of us who have used AI have seen some seemingly impressive stuff created from a prompt. But the saying, “Garbage in, garbage out” holds especially true here. 

Think of a woodworker or carpenter who needs to make a whole bunch of drawers or shelves or cabinetry. They will build a jig for the thing they need to create over and over again. But jigs wear out. And for those of us who spend any time using chat models know, as the context window gets longer, the tool can begin wearing out.

Ultimately it can break. And if you're not careful and just blindly copying and pasting or trusting what your chat is creating for you, that’s incredibly risky. So, if the tool begins telling you something that sounds a little off, you’ll have to go back several prompts and make sure that you're not dealing in nonsense without having noticed it.

You need to be an expert craftsperson to know that a problem has arisen. This is why human judgment remains essential in the learning process. AI can support the work, but it cannot replace the expertise needed to know whether the work is actually good.