Watch Reviews by ChatGPT

The unfortunate thing here is that I may take a person with prior knowledge of the product to pick up that error. I’m guessing “AI” will be used for a lot of product descriptions and fake reviews.  I could see sellers spamming Amazon with AI reviews to make their product look like it’s a popular quality item regardless of the truth. Other than tricking people or cheating on homework, I’m having trouble figuring out a good use for this type of software. 

Isn't there enough artificial intelligence in the watch world already?

The first generation 58 had an off-the-shelf 2824 on it, so it's not too far off. A far cry from the Kinessi movement. 

I think just the Black Bay. The in house must've been introduced with the 58.

I see it as being useful for disseminating bulk specifications and spitting it out onto a spreadsheet. 

From what I understand it's more just compiling stuff from its database into something that appears reasonably coherent - but it's probably all just snippets of text from actual human reviews and forums. It has no understanding of the material.

To me it seems like the chat bot is just regurgitating information that is already available on the web, from various other reviews and tech sheets. Probably the most impressive thing here is how it turns all the information into coherent paragraphs, although having any prolonged and nuanced discussions about watches is still well out of reach at the moment. 

Watching 2001: A Space Odyssey the other night makes me wonder that both a wrist watch and ChatGPT are both tools somewhere between a bone and HAL?

Folks, we should keep in mind that ChatGPT is operating on a small data set.  It currently does not have access to the internet.  That likely limits is ability to search out latest information.  So what likely happened is that the data set only included mention of Tudor using ETA based movements.  I speculate that with search capability, it would have found the info on Tudor's in-house caliber. 

The other point is that you have to keep refining your requests and teaching it how to respond. I tried correcting it on a fact, and it responded by learning and keeping that in mind for all subsequent chats on that thread. 

I'm reading that GPT-4 is really powerful. Hopefully, they release it soon and we can give it a test run.  

I asked it to write a song about the Omega moon watch in the style of Taylor Swift.  Here's what it produced.  Not great, but impressive for a language processing model.  I'm pretty sure this is original, cuz I can't imagine Taylor Swift writing this song: 

Verse 1: I was just a little girl, dreaming of the stars But then I saw it, shining bright on TV The Omega Speedmaster, on the moon with Neil Armstrong and Buzz Aldrin A symbol of humanity's endless pursuit

Chorus: Oh, the Omega Speedmaster moon watch A piece of history that will never fade A testament to courage and innovation A true cultural icon

Verse 2: I fell in love with its sleek design The precision and accuracy that it held within I had to have it, I just had to own The watch that conquered space, and the moon it roamed

Chorus: Oh, the Omega Speedmaster moon watch A piece of history that will never fade A testament to courage and innovation A true cultural icon

Bridge: It's been on wrists of astronauts and presidents Celebrities and every day men and women It's a symbol of strength and determination A reminder to always reach for the moon

Chorus: Oh, the Omega Speedmaster moon watch A piece of history that will never fade A testament to courage and innovation A true cultural icon

I'm sure this AI can pass the Turing test but what it output was the blandest review possible. If the reviewer was a human it would be one of those that can put you asleep just by telling you their name.

More like a boring ad unfortunately. 

Still impressive what AI can do, though.

I suppose we'll have to wait for SingGPT before we hear that AI-generated lyrical masterpiece.