AI written books

[gmw]

Originally Posted by andyh2000

[...] I don't know why you're so sure that artificial neural networks have nothing to do with natural ones. As far as I know that's explicitly where they got the idea from.

Andrew

On every topic I've seen discussed here on MR, the manner of Quoth's posts would have you believe they are an expert, privy to insider knowledge but apparently unable to share links to sources confirming it. (I don't count the quotes from 1967 as particularly relevant to modern attempts at AI.)

The truth, I assume, is the more prosaic explanation that this is a mannerism. I don't imagine that Quoth is involved in recent (the last decade or so) work on AI and its related fields and so their strongly expressed opinion is based on much the same information that you or I (or ChatGPT) have access to. But pattern matching machines fed by different data are very likely to arrive at a different result, so declaring just one of them as absolute fact is problematic. Something we see demonstrated here on MR most days.

While it probably sounds like it, this isn't meant to be picking only on Quoth. There are many here that share this mannerism to a greater or lesser extent - and that probably includes me, on some days at least. Taking as fact anything that any of us post here, without independent substantiation, is about as smart as accepting anything that ChatGPT tells you. Most of our posts (should) carry an implicit "IMO whether it sounds like it or not".

[andyh2000]

Originally Posted by Quoth

No, it's nonsense.
The name is inspired by the biology. They are nothing like at all in operation. It's AI industry propaganda.
We don't really know how vision exactly works. It's nothing like a camera feeding a computer. A computer neural network is a kind of database structure. It doesn't work at all like biological systems which are still poorly understood. It's just "AI industry" jargon.

Go do a real course in programming rather than reading industry sponsored articles. The Register https://www.theregister.com/ is more reliable than many other sources and it's only a tech news site.

Originally Posted by SteveEisenberg

Reading the ChatGPT passage in #25 slowly, several times, I see that.

But on first reading, not so much. I'd like to think that if ChatGPT wrote a mystery novel in the style of one of my favorite authors, I'd soon realize the plot and characterizations were only superficially plausible -- but I'm not sure. Human author plot errors get by me, so why not machine mistakes? And, getting back to #25, if I was an English composition teacher, I think it would at least get a B.

Can either of you point out the bits that were nonsense please, for my edification?

Andrew

[Quoth]

AI written books might not be protected by Copyright in the USA.
https://www.theregister.com/2023/02/...ght_ai_art_us/

[SteveEisenberg]

Originally Posted by andyh2000

Can either of you point out the bits that were nonsense please, for my edification?

Take this from ChatGPT:

The assertion that "we do know exactly how computers work" is also not entirely true. While we have a solid understanding of the underlying hardware and software mechanisms of computers, there are still many areas of computer science that are not fully understood, such as the theoretical limits of computation and the development of algorithms for certain types of problems.

The "theoretical limits of computation" has to do with what computers might do in the distant future, not how they work now.

The "development of algorithms for certain types of problems" means nothing concrete and implies that ChatGPT has some deeper understanding than what we have been given, and we should trust it there.

As for the discussion of whether there is a "great range of AI systems," this was a rhetorical device on Quoth's part that didn't mean anything concrete (#31 is relevant here). If I cared about that sentence, and Quoth and I were posting back and forth, I would have either asked for clarification, or, more likley, ignored the claim on grounds that it didn't advance Quoth's claims enough for me to have to refute it. ChatGPT doesn't know when to stop digging.

[gmw]

Originally Posted by SteveEisenberg

[...] The "development of algorithms for certain types of problems" means nothing concrete and implies that ChatGPT has some deeper understanding than what we have been given, and we should trust it there.

As for the discussion of whether there is a "great range of AI systems," this was a rhetorical device on Quoth's part that didn't mean anything concrete (#31 is relevant here). If I cared about that sentence, and Quoth and I were posting back and forth, I would have either asked for clarification, or, more likley, ignored the claim on grounds that it didn't advance Quoth's claims enough for me to have to refute it. ChatGPT doesn't know when to stop digging.

Hmm... so ChatGPT is sounding more and more like a regular MR participant. Do you so suppose we might all be AIs?

[ZodWallop]

Originally Posted by gmw

...would have you believe they are an expert, privy to insider knowledge but apparently unable to share links to sources confirming it...

There are many here that share this mannerism to a greater or lesser extent - and that probably includes me, on some days at least.

I never do that! And I know it for a fact

[jbjb]

Originally Posted by Quoth

* A relay can easily implement Not, Or and And, the building blocks of all digital computers, though Nand (And with inverted output) and Nor (Or with inverted output) are the usual building blocks. Absolutely nothing more is needed. A memory cell can be interconnected gates to make a Flip-Flop, though now a stored charge is often used, a solid state version of magnetic core stores.

Minor clarification - Nand *or* Nor is all that's needed. Either will do, you don't need both.

[Quoth]

Originally Posted by jbjb

Minor clarification - Nand *or* Nor is all that's needed. Either will do, you don't need both.

True, but simpler to have both at the component level. Similarly while And and Or are basic operations real devices use Nand and Nor (Valves, RTL, TTL, CMOS etc). It doesn't matter to relay logic if the contacts are the change over type.

https://en.wikipedia.org/wiki/De_Morgan%27s_laws

https://en.wikipedia.org/wiki/Logic_optimization

[ownedbycats]

Originally Posted by gmw

Hmm... so ChatGPT is sounding more and more like a regular MR participant. Do you so suppose we might all be AIs?

Beep. Boop.

[Quoth]

I'm only the messenger:
https://www.newstatesman.com/the-wee...l-intelligence

Written by "an insider".

May need to disable cookies and reload.

"The programming web forum StackOverflow banned ChatGPT generated answers in 2022, – shortly after the chatbot was released – saying that its answers are often simultaneously incorrect and plausible, making them difficult to moderate"

Summary of another part of the article:
It's designed to be plausible, but there is no indication to reader which souces were scraped, truthiness or copyright.

[jbjb]

Originally Posted by Quoth

True, but simpler to have both at the component level. Similarly while And and Or are basic operations real devices use Nand and Nor (Valves, RTL, TTL, CMOS etc).

In practice, for hand-designed gate level logic, one or the other is generally used, not both. The logic optimisation typically results in a sum-of-products expression (AND-OR, which de-morganises to NAND-NAND), or a product-of-sums (OR-AND, which de-morganises to NOR-NOR). In practice it's usually NAND-NAND that is chosen - at least in CMOS designs, which covers the vast majority of digital logic today (CMOS NAND gates have better symmetry of rise and fall times than NOR gates, when same-size MOSFETs are used). Most hand designs will involve a mixture of NAND gates and edge-triggered D-type flip-flops.

Of course, the vast majority of digital design these days isn't done by hand at gate level - it's usually done at RTL level in an HDL such as system verilog or VHDL and synthesised. Depending on the target, the synthesised output may be NAND gates and d-types, but it often isn't - e.g. if it's to FPGA the basic cell from which it's all built is usually a look-up-table for the combinatorial logic, a flip-flop, and a 2 input mux. Other target types like PLAs have their own structures. (Not to mention that in practice much of the logic is just done in software these days, as embedded microcontrollers are so easy to include in custom digital designs.)

[Quoth]

It's a black box now.
Nor gates last popular with logic cards using discrete transistors and resistors? I bought a load of logic cards when a teen so as to strip the resistors, diodes and transistors for my own analogue projects.

I've got four obsolete Xilinx FPGA dev kit boards going free to anyone picking up or paying postage. Maybe a load of PIC gear too.

[Quoth]

Probably ChatGPT hasn't got a Ravelry account.

https://www.theguardian.com/technolo...attern-results

[Roos]

Originally Posted by gmw

Hmm... so ChatGPT is sounding more and more like a regular MR participant. Do you so suppose we might all be AIs?

I think ChatGPT sounds like a (or better yet, any) politician, fluffing up lots of words to confuse the listener, sounding plausible, while hiding lack of any necessary knowledge, completely filled with empty words, hot air and poppycock, we should fire all those useless sacks of politicians shits and replace them with individually slightly altered versions of chatgtp and let them talk each others ears off while we get on with our life.

[ownedbycats]

Originally Posted by Roos

I think ChatGPT sounds like a (or better yet, any) politician, fluffing up lots of words to confuse the listener, sounding plausible, while hiding lack of any necessary knowledge, completely filled with empty words, hot air and poppycock, we should fire all those useless sacks of politicians shits and replace them with individually slightly altered versions of chatgtp and let them talk each others ears off while we get on with our life.

I saw a screenshot of someone linking up two Bing chatbots and they got stuck in a loop of appreciating each other. So... politicians?