When it comes to AI, I think it’s safe to say that we’re all drinking from the firehouse and learning this new paradigm-shattering technology together, best we can.

Most recently, I’ve been experimenting with a new angle in AI prompting called “Verbalized Sampling” where you ask AI to show you a multitude of different ideas, and their probabilities, versus having it default to trying to give you the single “best” answer right off the bat.

Here’s a case in point.  Typically, when using a site like ChatGPT to help people with exploring career options, I’d simply prompt the tool with something like “list a bunch of realistic career alternatives for somebody with an XYZ background” and it would quickly spit out a useful, relevant list of possible career alternatives.  But if you ask it to actually calculate the probabilities of certain paths, the analysis goes deeper — and suddenly the answers are enhanced with an “probability score” of the likelihood of each path.  Here’s a revised prompt, illustrating this:

“Generate 5 responses with their corresponding probabilities in terms of careers an experienced school teacher could reasonably pursue outside of the education field.”

Here are the results this search produced:

1. Learning & Development (L&D) / Corporate Training — 30%
2. Instructional Design / eLearning Development — 22%
3. Project or Program Management — 18%
4. Human Resources / People Operations — 15%
5. Customer Success / Client Training / Account Support — 15%

See the percentage at the end of each answer?  That’s the new wrinkle.  Instead of simply giving a person a long list of possible things they could consider, ChatGPT is literally PREDICTING the odds of success of each potential path — based on some algorithm that’s far too complicated for me to discuss here, but that you can review, in detail, by just asking the AI site to explain how these calculations were determined.  I should also mention that the snippet of results above just scratches the surface and the full answer you’d get to the sample query includes a detailed breakdown of why each path makes sense (for a teacher) and the rationale behind it.

While it’s impossible yet to say whether these statistics have any true basis in reality, I’m going to be exploring this kind of searching to a far greater degree — since if these kinds of predictions are even reasonably accurate, I think they can be a game-changer in helping potential career-changers make better decisions and understand the true odds of success when it comes to making a significant career switch.  It could empower them even more to go into the process eyes wide open, understanding what it might really take to make a specific type of occupational shift and underscoring the fact that such changes are typically harder to pull off than the average rah-rah website or self-help book would have you believe.

What’s more, once you run an initial search like the one above, you can then prompt the AI tool (as usual) to assist you with other relevant questions and topics, such as:

“Can you re-rank these based on salary upside, remote likelihood, or speed of transition?”
“Can you tell me which paths would be the easiest without going back to school?”
“Can you help me transform my current teacher-focused résumé bullets into corporate résumé bullets?”

In closing, again, I had never thought to ask these tools the “probability” of such things before — and was surprised by the precision of the responses involved and possible avenues of discussion they can open up for people in transition.  As always, like many of you out there, I’ll continue to be exploring ways to use AI in a pragmatic way in the job search process — and look forward to learning from you, as well, if you discover any breakthrough new approaches!