Yes, ChatGPT5 is a dud, in comparison to its predecessors.
Yet, there is an interesting aspect of this flop that I find encouraging. To wit:
Using an AI LLM is a multi-step process, not a single-step task. Like many of my acquaintences, I early on made the mistake of judging the potential of AI on the basis of hullucinatory responses to single questions. "Aha!", I thought, "See, AI doesn't know everything."
Only when I realized how many times I've asked a human a question, only to be dragged into a multi-faceted discussion about context, hidden assumptions, and overt, but inaccurate assumptions, and similar factors did I begin to make progress with ChatGPT, or my now preferred Assisted Intelligence tool, Claude. It's a process, and either you or your AI has to be in control of that process. You can't just go along for the ride and hope to get anywhere close to where you want to be.
Spoiler alert. That's exactly the way many, many discussions here and at other Access forums proceed, is it not?
So, consider the progression of ChatGPT in that light. They got along famously until ChatGPT5 didn't. Bulding an LLM is a process, and there are always hidden assumptions and overt, but inaccurate assumptions and miscalculations along the way. This, apparently, is one of those missteps.
In my opinion, the disappointment of ChatGPT5 is no more an indication of a hype bubble itself bursting than its predecessor was a harbinger of eutopia dawning for us all.
Stuff happens. It has since the beginning of recorded human history. We learn from it, or we don't. But the lesson of the history of humanity is that more often than not, somebody learns spomething from every experience, and somebody builds on that learning. Not everybody gets it on day one. But somebody does and they reap the benefits.
Perhaps we can hope that ChatGPT5 means there will be a little less hubris and a little more practicality in the next phase of AI development. I think there are going to be some more hard thumps up side the head for the most zealous AI'ers and those least willing to learn from their own mistakes. However, there will be enough hard-nosed pragmatics who do get the point and who keep progress moving.
www.latimes.com
