ChatGPT Improves Its Summary of My Writing -- But Still Adds Serious Errors!

In May 2024, I was boggled by how badly ChatGPT summarized my writing, especially my definition of the Metaverse -- even when it's online right here -- so last week, I checked back to see if ChatGPT had improved.

Yes, this seems like a self-indulgent ego stroke, but I promise there's an important point here -- especially when a majority of people use an LLM like ChatGPT instead of Google.  

Anyway, I first got back the mishmash posted above. So in my next prompt, I wrote, "That's not accurate."

ChatGPT apologized... then gave me another version of the definition that was still wrong.

But the next day, when I asked for my definition again, ChatGPT's response was markedly improved. It even included my picture! Look:

        

Wagner James Au defines the Metaverse as:

“A vast, immersive virtual world simultaneously accessible by millions of people through highly customizable avatars and powerful experience creation tools integrated with the offline world through its virtual economy and external technology.” VentureBeat+3New World Notes+3Muck Rack+3

This definition draws from Neal Stephenson’s original concept in Snow Crash, emphasizing:turingchurch.com+5Voices of VR+5Muck Rack+5

Immersive Virtual World: A persistent, real-time environment.

Simultaneous Access: Millions of users interacting concurrently.

Customizable Avatars: Personalized digital representations.

Experience Creation Tools: Robust tools for users to build and modify the virtual space.

Integration with the Offline World: Connections to the physical world through virtual economies and technologies. X (formerly Twitter)VentureBeat+7valibrarian.net+7New World Notes+7

Au also critiques the notion of a fully interoperable "open metaverse," suggesting that such a model may not align with the inherently immersive and real-time nature of virtual worlds. Instead, he advocates for:VentureBeat

Interoperability of Community: Facilitating user communities to move across platforms, often through tools like Discord, rather than focusing solely on technical interoperability. VentureBeat

In essence, Au envisions the Metaverse as a dynamic, user-driven ecosystem where community and creativity are central, and where the virtual and physical worlds are interconnected.

Like a frantically embarrassed student at a strict boarding school, ChatGPT studied overnight to fix its answer, and by the next morning came pretty damn close. So in fairness, the LLM's self-correcting mechanism, apparently prompted by me saying it was wrong, led to it producing something, well, less wrong.

But still!

I never wrote opposing a fully interoperable open metaverse because it "may not align with the inherently immersive and real-time nature of virtual worlds", whatever that means. I wrote that different virtual worlds have different game/physics engines, user interfaces, user communities, even different demographics, all of which makes a fully interoperable open metaverse undesirable for most consumers, let alone impractical.

Also, a future where the "virtual and physical worlds are interconnected" is not something I wrote much about, beyond unique cases like connecting a profoundly disabled person to a virtual world where they can socially thrive. But that's not the majority of how most people will experience virtual worlds, at least for the near future.

Also, I write in the book about the essential limitations of virtual worlds recreating amazing real world experiences, giving my own childhood growing up in Hawai'i as a specific example. So not only is ChatGPT getting that part wrong, I'm somewhat offended personally!

And again, remember that a majority of people use an LLM like ChatGPT instead of Google, especially Gen Z. They're not likely to tell ChatGPT to fact check itself, nor know if its answers are 100% correct, or much less than 100%.

So imagine what we're getting collectively wrong, on so many topics vastly, vastly more important than how I define the Metaverse.

Please support posts like these by buying Making a Metaverse That Matters and joining my Patreon!

Become a member!

Comments

[HalfOfTheSLOfficeNeedsFiring]

ChatGPT and its other LLMs are pretty terrible even with code.
If I didn't have at least a good understanding of the code I am currently working in and refreshing knowledge on, i'd probably miss the terrible mistakes and suggestions ai spews out on this topic.
It gets things wrong, then when you point it out, and why, it just repeats your own code back to you and blows smoke up your butt with flattery. Terrible.
It's been so bad so far and especially on intermediate things i do not put any trust into its answers.
People learning should stay far away from AI, or be very skeptical about its responses, as it can waste a lot of your time past boilerplate, or in your case simple stories.
Even then it's just regurgitating what you typed in a new way and not always good.