Replies

I spent an hour or so toying with ChatGPT a couple of weeks ago. I have also watched a number of videos of people using it to repair things, write music and recipes. ChatGPT gave fairly miserable results and show zero creativity in my estimation. In terms of repairing things, it did nothing more than search for all the potential solutions to troubleshoot a problem. I had to do a similar Internet search yesterday to troubleshoot a problem with my phone’s hotspot. No big deal. ChatGPT sort of automates that search. Neat, but leave it at that. Remove the marketing fluff, lower your expectations and you have a useful automation for some problems.

As a retired software engineer I have 35+ years of experience of seeing claims of all sorts to never materialize. Way back in ‘86 I worked on a failed project, an ‘expert system’ written in Prolog. The technical press at the time extolled AI claims of Prolog. Why do I write this? There are a couple of patterns to be recognized. Tools, even useful tools, are given significant hype by the press. Then reality sets in. Gartner calls that reality the Trough of Disillusionment.

Another pattern is there are two cycles of hype. One in the technical media and then followed by the mainstream media. Additionally, there is the resurrection of hype, sometimes with a twist. I previously mentioned Prolog. It wasn’t new in ‘86. It dates back to the early 70’s. It’s still used today. IBM Watson has part of its code base written in Prolog. It isn’t magic, in fact multiple technologies and techniques are being used, each with their own strengths and weaknesses. What’s the twist I mentioned above? It’s the opaque use and rebranding of underlying implementation. Watson uses Prolog, Java, and C++ as some of its languages along with various programming models and techniques. That’s the underlying stale package combined with tremendous compute power, all of which is branded by the mainstream media as something sort of amazing and nebulously AI. While, I don’t know much about the underlying implementation of ChatGPT, I have high confidence that the same applies to it.

Word of caution: When the mainstream media gets a hold of things a large percentage of the population believes it is fact. In the case of ChatGPT, that means the results it spews are deemed by many as fact because of media hype surrounding AI. ChatGPT has a leftist bias in many things from history to current events and politics. I’m not sure if that is intentional or not, but it at least seems to be a by product of leftist developers and ChatGPT’s inability to discern the validity of inputs.

It makes ChatGPT a leftist propaganda tool that is accepted by many as fact just because of most people’s limited cognitive abilities and ignorance. If anything, that’s the real magic of ChatGPT.