AI has been in popular technology discourse for a while now, and for good reason, as it theoretically can fulfill any role given enough time and resources. These implications are far-reaching and have already made waves in fields previously thought to be untouchable like art. For those in the learning industry, however, the question is how can AI affect training?

How Will AI Affect the Training Industry

While it might be tempting to believe that AI can never affect your specific field, it can, will, and some start-up is planning on building a business from it. That said, these possibilities will all be based off currently available AI (or the direction AI is going in). Like all technologies, AI will evolve until the point where it could be capable of handling far more than it can today.

Generating Written Content

AI has been used primarily to generate written content and it always hasn’t been the greatest at it. This has changed recently with AI’s like ChatGPT, which can reliably naturally respond to any number of prompts. It is very conceivable that soon you could feed an AI training topic to get a boilerplate text document. The fact that it’s a boilerplate is essential here, AI responses often must be edited for the small details, however it can understand and structure the logic.

Generating Image Content

Perhaps the second most popular use for AI, generating images from prompts has been exploding over the internet for both good and bad reasons. Regardless, can you use it for your training? Well, it depends. AI art is very good at making, for lack of a better word, creative images. Technical images, however, have yet to be seen or, at least, not worthy of note.

Making something like a flow chart with AI could be possible theoretically, but in practice it won’t be the most efficient. This is because of what was touched on in the above with written content, AI, currently, is good at the logic but not the details. If an AI were to get details wrong in an informational graphic, one would still need the technical skills edit the image. A feat made even harder by the unique style AI art can have. As of right now, AI art could be used to help stylize courses but not to generate informational content.

Provide and Suggest Different Analytics

If there’s one thing that AI is consistent at its collecting data and then using that to make decisions. As such, AI could be a very useful to for analyzing the information collected from a user and then making suggestions based off that. Perhaps AI could highlight potent data points for specific learners while tracking them over time and setting goals for them. You might read that and think, "well, I can do this now," and you can, but AI could do it for significantly more learners, significantly faster.

Sorting and Understanding Users

Sorting learners one of the biggest potential benefits, but also some of the most disastrous outcomes from AI. For the benefits, AI could help bolster individualized learning by analyzing when and how a learner performs best. Then the AI could map a path for said learner that is most conducive to them—theoretically. To fulling understand why all of these potential benefits are constantly cushioned by “theoretically,” we must understand that we can’t understand what’s in the black box.

What Is the Black Box

Whenever talking about AI, you’ll often hear of the “black box” an analogy of how decisions are made. It begins when AI is still “learning,” to oversimplify, the AI makes a bunch of different versions of itself then the version closest to the goal is chosen which is then used to make a new generation of versions. This process continues until the AI can reliably produce the “right” outcome. However, during each one of these generations, the AI adds more data to make its judgement, which quickly becomes incomprehensible to even the people who made it. That is the black box.

We know that within the black box the AI is making decisions, however, the how and why are way too complicated for people to fully understand. Imagine if you had to parse over a million different data points to decide anything, that’s what you would have to do to see within the black box.

Black Box Consequences

Consequently, AI can develop some unsavory data points that don’t influence the outcome but serve to reinforce preexisting prejudices. There are numerous examples of AI being taught prejudice, like the newest ChatGPT, which has been shown to give racist responses. It is a tricky situation because the AI is not itself racist, but somewhere, in that tangled black box, it is making punitive decisions about race. Even hardcoded barriers against this might not be effective as AI can mark so many different reasons for making decisions that it could circumvent those barriers.

This could happen with AI used specifically for training programs. What if the sorting AI decides that one race should be a lower path than other races? Learners of otherwise standard aptitude may be ranked lower only because of their superficial characteristics.

The Tangled AI Web

It’s easy to get caught up in waves of anticipation and AI is a really exciting new technology. However, despite the optimistic excitement, AI can still develop flaws. We must try to be cognizant of these flaws and not treat AI as perfectly logical gospel. After all, it is made by flawed people who might be asking it flawed questions.

In it’s current state, AI isn’t about to replace anyone’s jobs but it can serve as a great tool for inspiration. This technology can make so many different connections that you’ve might never have thought of, but it could also use that same ability to make unfair decisions that can impact people’s futures. AI's here, its not going anywhere, what matters the most right now is how we use it.

Get started with CoreAchieve for free.

Photo by Sophie Louisnard on Unsplash