AI Diversity
Like practically everyone I’ve been playing with ChatGPT. I was pleased to notice folks of different stripe testing it in surprising ways that I wouldn’t have thought of. One tester gave it specific characters each with detailed descriptors and had it weave what seemed a good story out of it. My own approach was asking either scientific or philosophical questions. I also asked to be poetic and was disappointed in the simplicity produced in an instant, short rhyming childlike stanzas. I know I could have gotten something better out of it. In my own discussions and interactions, I found it repetitive in its sourcing and down-right pedestrian, a mouthpiece for party-line or consensus unable to process a different point of view. Now I find in the paper the U.S. government is already wanting to regulate it. Regulate what? I had to literally coax it to find what was obvious and readily available. Here I’m not talking politics where there might have been some reluctance, but physics and science. I personally would have appreciated it presenting even a slightly different perspective on something that I hadn’t thought of, but that wasn’t happening. It means that in its current iteration it might not give us something adventuresome, or maybe the inventors are too afraid to put out anything but a plain vanilla product. I do understand they aren’t giving us mortals everything they’ve got.
In this first step, we’ve seen only the consensus with practically no diversity at all.
That’s why I’d predict in future we’re bound to have a profusion of diverse AI products fit for a diverse human marketplace, same as diversity of News, and Platform products we have today so we can again have food fights with our own AI versions. I don’t see a Google or Twitter cornering the AI market which is fine by me. Now I hear Elon Musk is already working on another AI brand he calls TruthGPT. If we find ourselves laboring in our own parochial cubicles using versions of AI what have we accomplished?