Geoffrey Moore: Understanding ChatGPT: A Triumph of Rhetoric

[W]e need to stop thinking of ChatGPT as artificial intelligence. It creates the illusion of intelligence, but it has no semantic component. It is all form and no content. It is a like a spider that can spin an amazing web, but it has no knowledge of what it is doing.

You may have noticed that when I talk about things like GPT and Midjourney, I try as much as I can to call them text and image generators, not AI tools. In my opinion, calling these things “AI” is improperly selling what they’re doing, and it gives people who don’t like these tools an easy zinger they can bust out to show how stupid these “supposedly intelligent” tools are (and are therefore useless).

Busting out the “I’m old” card, I was in college in the mid to late 2000s, and Wikipedia existed, but it wasn’t universally respected as a source for any form of truth. I had professors who were adamant that using Wikipedia for research was just an exercise in reading bullshit, and real information could be found in scholarly reports. If I cited Wikipedia in my bibliography, it would be roundly rejected.

But over time, Wikipedia’s value has been made pretty darn clear to the world, and there have been systems and standards put in place to make sure it’s more reliable than it was 20 years ago, it’s still kinda the same deal: you should not implicitly trust everything you read on Wikipedia, and you should never, ever cite it as a source in an official essay/report/paper/whatever. But only someone with their head firmly in the sand would tell you that Wikipedia is useless because it can not be completely trusted.

• Here’s a 2005 article in the New York Times where someone looks themself up on Wikipedia, only to find things that aren’t true. Something I’ve seen many people do with ChatGPT recently.
• Here’s an article from CNN in 2007 warning that Wikipedia can be useful, but will also tell you lies
• Here’s a podcast also from 2007 talking about conservatives making their own Wikipedia, just like conservatives want to make their own chatbots today. The quest for bending reality to your political preference never ends.
• Here’s a 2008 article from MIT talking about the new definition of “truth” in the age of Wikipedia.