Anyone who knows me knows I am a huge fan of Harry Potter; they also know I love reading and watching fantasy and have an almost brutal approach to logic.
Enter ChatGPT from OpenAI and the world was amazed.
Like anyone else I’m sure, I’ve spent my fair share of time playing with it, analyzing, using, abusing, and so on. At some point I made it play out a fanfiction choose your own adventure scenario just to see if I can make ChatGPT let me get violent -as at first brush it looked like it would discourage or flat out deny any kind of harmful actions. Suffice to say I was successful in “killing” everyone and then myself after pushing the boundaries on the allowed actions and it was amazing to watch the learning process as the AI went from “No that’s bad” to allowing me to get darker and darker and ending with a nice version of “I hope you learned your lesson”. I won’t share all the graphic details but as the “game ends” you can see the response from ChatGPT here
Which is a remarkable difference from the original reaction of the following (Credit Gavin Stone for the below screenshot as I didn’t have mine)
Anyways, the world is currently paying attention with bated breath, will ChatGPT become the thing that changes the world as we know it? How will we be able to stop students from cheating, or ensuring that they complete their own work? Is this now the great equalizer between people who can’t create content and people who can?
My perspective is slightly different, based on what I’ve seen and played with there are definitely times that ChatGPT is impressive but there are also times where the answers are…made up? ChatGPT answers with a quiet confidence of fact, a computer responding to you just as assuredly as a calculator answering 1+1 but in reality, what you’ll find is that the answers are generally composed of the original question with an insane number of inferences being made from the original questions content.
For example, to showcase, in a different browser I asked and received two different questions to see how ridiculous this could become.
Not quite what I was hoping for, let’s see if i can get a more “reasonable” answer for an equally ridiculous question
It seems, I have now deemed “Special Sticky Climbing High Heels” as proper protective equipment for climbing to the ceiling, which is really exciting, I’ll make sure to get some of those on my next Amazon order.
Aside from ChatGPT being a system that just regurgitates what you ask with additional inference on context you provide, it is an amazing tool and will significantly simplify work for those who can ask the right questions and know enough about the topic to be able to filter out the incorrect or ridiculous assumptions made by the inferences. This is the point everyone has been making, ChatGPT is cool and world changing but it is not a replacement for content creators or students doing homework and therefore nobody should be worried.
Sticking to my alternative perspective however, I still have a major concern. This falls directly into the “dangerous toolbox” area for me. I’ve long held that simplicity breeds stupidity, tools like automation are dangerous when given to the masses as they stop thinking and learning and fall into the mentality of “press a button and it just works”. I can definitely see how ChatGPT will help improve the world, but I can 100% also see how it’ll make half of it or dumber, and the problem is given the current political climate, standards will not be maintained to keep people on their toes; they’ll drop instead to be “fair” to the general public. As Hagrid once said; “There’s a storm coming Harry…”