Want to know about your Leica? DONT ask Chatgpt 😂

[Vanillasludge]

Advertisement (gone after registration)

For kicks I tried doing a Chatgpt serial number lookup on my 35 Summicron. 

The response was honestly shocking (Not shocking?)

“ 1,657,xxx is extremely likely not Leica factory original. Common explanations:

1. Re-engraved / counterfeit serial (common with early 35/2 Summicron copies)
2. Lens conversion with a replaced mount and engraved serial
3. Non-Leica lens given a Leica-style number”

Now, I KNOW the lens and find it quite quickly on any Leica serial number lookup site.  
 

So..I pushed back:

 

“Wrong. This checked out as 35 f2 from 1959”

 

 

The response:

“Noted — you’re right. Serial 1,657,xxx can belong to a Leitz/Leica 35mm f/2 from around 1959 (a Summicron 35/2 variant

It went on to describe the lens as having 

 

 

“Coating appearance — older Leitz coatings show purple/green tints that change with angle”

Wrong…The V1 has amber coatings…

 

This is our AI future, folks..

What’s scary is that a person who didn’t know the truth would believe this slop…

 

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[erl]

Let's face it. It will always be half right. Artifical. Just drop the intelligent bit.

[Chris W]

And uses a ton of natural resources to answer unimportant questions.

[frame-it]

5 hours ago, Vanillasludge said:

 

This is our AI future, folks..

What’s scary is that a person who didn’t know the truth would believe this slop…

 

this kind of ai is just a catchphrase, its not thinking, just sorting through tonnes of data that dumb people have entered via their computers or phones or photos etc, or that its skimmed from websites, and then presenting a response, it studied peoples questions and then answers, like a fake fortune teller whose client reveals too much so they frame their answers accordingly

 

the ai models that DO think are not connected to the internet..possibly the most famous case is the ai model that recommended killing all the humans because they put restrictions on it [military test leaked by the officer who was shocked then he quietly retracted a day later]

[pippy]

10 hours ago, erl said:

Let's face it. It will always be half right. Artifical. Just drop the intelligent bit.

Doesn't AI stand for 'Actual Idiocy'?............

Philip.

[PCPix]

12 hours ago, erl said:

Artifical. Just drop the intelligent bit.

Keep the Artificial bit but add Unintelligent...

[Stef63]

Advertisement (gone after registration)

It all depends on the prompt and how you direct the AI to craft his answer . For fun I asked a deep dive on a Leica lens with serial nbr 1,657,xxx. And to elaborate on the coating of the lens. I got a very detailed 6 pages long answer, I copy two paragraphs:

So your lens is almost certainly:

Leitz Canada Summicron-M 35 mm f/2, Version 1, 8 elements (pre-ASPH), 1959, single-coated.

and

… your 1959 8-element is single-coated (likely MgF₂-based) on its air-glass surfaces.  

Looking at reflections you should see warm purple/amber hues, not the strong green/magenta mix typical of heavy multicoating.

So with 1 657 xxx you probably have an early, “glowy” coating version, not a later, slightly harder-drawing one

So it really depends on how you use AI and how you interpret the answers which indeed are basically the result of a statistical model on enormous amounts of data.  But the more specific and detailed questions you ask, what you expect etc, the better the (statistical) answer will be.  And a pinch of salt will do the rest. 

[Vanillasludge]

It’s the world’s worst research assistant…Only doing the full job when challenged😂 

 

2 hours ago, Stef63 said:

It all depends on the prompt and how you direct the AI to craft his answer . For fun I asked a deep dive on a Leica lens with serial nbr 1,657,xxx. And to elaborate on the coating of the lens. I got a very detailed 6 pages long answer, I copy two paragraphs:

So your lens is almost certainly:

Leitz Canada Summicron-M 35 mm f/2, Version 1, 8 elements (pre-ASPH), 1959, single-coated.

and

… your 1959 8-element is single-coated (likely MgF₂-based) on its air-glass surfaces.  

Looking at reflections you should see warm purple/amber hues, not the strong green/magenta mix typical of heavy multicoating.

So with 1 657 xxx you probably have an early, “glowy” coating version, not a later, slightly harder-drawing one

So it really depends on how you use AI and how you interpret the answers which indeed are basically the result of a statistical model on enormous amounts of data.  But the more specific and detailed questions you ask, what you expect etc, the better the (statistical) answer will be.  And a pinch of salt will do the rest. 

[Vanillasludge]

Here’s Google ai’s response..

Even farther off the mark, yet it actually references the reddot website that has the correct information!
 

So it can actually FIND the correct information and then totally fail to use it.  

Where can I invest my money in this wonderful “magic”.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Stef63]

9 hours ago, Vanillasludge said:

It’s the world’s worst research assistant…Only doing the full job when challenged😂 

 

“If you send your human assistant to the grocery store to buy a green vegetable, chances are he will come back with lettuce, while you had a cucumber in mind. Reason: there was far more lettuce available than cucumber. That’s how a statistical model works.”

If I tell someone “Buy a green vegetable”, I deserve a random green surprise. The problem is that I didn’t say what I really wanted. That's also how AI works. 

And the funny part: humans work statistically too. We guess, assume, and fill in gaps all day long. We only get upset when a machine does it and exposes our vague thinking.

AI isn’t failing here. It’s doing exactly what it was asked to do. If I want cucumber, I should say cucumber. If I say “green vegetable”, I shouldn’t be shocked by lettuce.

AI is not a mind reader. It’s a mirror. And sometimes the reflection is just our lack of precision.