WBAUnofficial
When I woke up this morning... - Printable Version

+- WBAUnofficial (https://wbaunofficial.org.uk)
+-- Forum: WBAUnofficial (https://wbaunofficial.org.uk/forumdisplay.php?fid=3)
+--- Forum: WBAUnofficial (https://wbaunofficial.org.uk/forumdisplay.php?fid=4)
+--- Thread: When I woke up this morning... (/showthread.php?tid=30139)

Pages: 1 2 3


RE: When I woke up this morning... - Borin' Baggie - 05-30-2023

(05-30-2023, 02:54 PM)Birdman1811 Wrote: I agree heavily with the experts saying this is being heavily overblown.

Ai is still very, very basic, and the limitations will not go away until the fundamental architecture of computers change significantly. While we are still essentially using Von Neumann architecture to run Turing machines, admittedly faster and faster all the time, but computing hasn't changed from those days. Those restrictions make self-aware AI completely impossible, as proven by Alan Turing, with said Turing machine. So long as AI isn't self-aware, it's a slave to us.

The issues with AI and automation is that as our population grows, we are actually losing jobs and roles for humans. It was, wrongly imagined that Ai Would take on the menial work. However due to our capitalist society that requires all humans to get in some form of work, AI seems to be being used for the arts and such, which is what humans were supposed to focus on.

There are absolutely no advancements in the foreseeable future that remove those limitations, and any discussion on my the second point leads to silly name calling and will go on to the Politics board.

LLMs like ChatGPT are extremely limited but moving forward into more automated processes there needs to be regulations. Thankfully we're not there yet but I don't know how quickly we'll get along the S-curve or where on the S-curve that is.

Do we really want an AI to help manage Trident's systems, for example?


RE: When I woke up this morning... - AnelkasBeard - 05-30-2023

(05-30-2023, 03:12 PM)Borin' Baggie Wrote:
(05-30-2023, 02:54 PM)Birdman1811 Wrote: I agree heavily with the experts saying this is being heavily overblown.

Ai is still very, very basic, and the limitations will not go away until the fundamental architecture of computers change significantly. While we are still essentially using Von Neumann architecture to run Turing machines, admittedly faster and faster all the time, but computing hasn't changed from those days. Those restrictions make self-aware AI completely impossible, as proven by Alan Turing, with said Turing machine. So long as AI isn't self-aware, it's a slave to us.

The issues with AI and automation is that as our population grows, we are actually losing jobs and roles for humans. It was, wrongly imagined that Ai Would take on the menial work. However due to our capitalist society that requires all humans to get in some form of work, AI seems to be being used for the arts and such, which is what humans were supposed to focus on.

There are absolutely no advancements in the foreseeable future that remove those limitations, and any discussion on my the second point leads to silly name calling and will go on to the Politics board.

LLMs like ChatGPT are extremely limited but moving forward into more automated processes there needs to be regulations. Thankfully we're not there yet but I don't know how quickly we'll get along the S-curve or where on the S-curve that is.

Do we really want an AI to help manage Trident's systems, for example?

Anyone raving about ChatGPT - whenever I've seen it used, it just writes absolutely nothing of substances whatsoever.

Is it really AI if it can't think for itself? I'm ignorant on the programming, but isn't it just a long list of commands and tools it uses to "create" things?


RE: When I woke up this morning... - Borin' Baggie - 05-30-2023

(05-30-2023, 03:14 PM)AnelkasBeard Wrote:
(05-30-2023, 03:12 PM)Borin' Baggie Wrote:
(05-30-2023, 02:54 PM)Birdman1811 Wrote: I agree heavily with the experts saying this is being heavily overblown.

Ai is still very, very basic, and the limitations will not go away until the fundamental architecture of computers change significantly. While we are still essentially using Von Neumann architecture to run Turing machines, admittedly faster and faster all the time, but computing hasn't changed from those days. Those restrictions make self-aware AI completely impossible, as proven by Alan Turing, with said Turing machine. So long as AI isn't self-aware, it's a slave to us.

The issues with AI and automation is that as our population grows, we are actually losing jobs and roles for humans. It was, wrongly imagined that Ai Would take on the menial work. However due to our capitalist society that requires all humans to get in some form of work, AI seems to be being used for the arts and such, which is what humans were supposed to focus on.

There are absolutely no advancements in the foreseeable future that remove those limitations, and any discussion on my the second point leads to silly name calling and will go on to the Politics board.

LLMs like ChatGPT are extremely limited but moving forward into more automated processes there needs to be regulations. Thankfully we're not there yet but I don't know how quickly we'll get along the S-curve or where on the S-curve that is.

Do we really want an AI to help manage Trident's systems, for example?

Anyone raving about ChatGPT - whenever I've seen it used, it just writes absolutely nothing of substances whatsoever.

Is it really AI if it can't think for itself? I'm ignorant on the programming, but isn't it just a long list of commands and tools it uses to "create" things?

CharGPT in it's current iteration is basically a toy. It can be useful as a shortcut but you need to know what you want out of it and how to get there in order to do that. I'm pretty sure it was only publicly released for marketing reasons, not as an example of the capability of AI.

AIs are models designed to emulate certain characteristics. ChatGPT is still an AI, it needs to parse an input and generate an output based on that through a logical process.


RE: When I woke up this morning... - Birdman1811 - 05-30-2023

(05-30-2023, 03:14 PM)AnelkasBeard Wrote:
(05-30-2023, 03:12 PM)Borin' Baggie Wrote:
(05-30-2023, 02:54 PM)Birdman1811 Wrote: I agree heavily with the experts saying this is being heavily overblown.

Ai is still very, very basic, and the limitations will not go away until the fundamental architecture of computers change significantly. While we are still essentially using Von Neumann architecture to run Turing machines, admittedly faster and faster all the time, but computing hasn't changed from those days. Those restrictions make self-aware AI completely impossible, as proven by Alan Turing, with said Turing machine. So long as AI isn't self-aware, it's a slave to us.

The issues with AI and automation is that as our population grows, we are actually losing jobs and roles for humans. It was, wrongly imagined that Ai Would take on the menial work. However due to our capitalist society that requires all humans to get in some form of work, AI seems to be being used for the arts and such, which is what humans were supposed to focus on.

There are absolutely no advancements in the foreseeable future that remove those limitations, and any discussion on my the second point leads to silly name calling and will go on to the Politics board.

LLMs like ChatGPT are extremely limited but moving forward into more automated processes there needs to be regulations. Thankfully we're not there yet but I don't know how quickly we'll get along the S-curve or where on the S-curve that is.

Do we really want an AI to help manage Trident's systems, for example?

Anyone raving about ChatGPT - whenever I've seen it used, it just writes absolutely nothing of substances whatsoever.

Is it really AI if it can't think for itself? I'm ignorant on the programming, but isn't it just a long list of commands and tools it uses to "create" things?

Modern Ai like ChatGPT is certainly impressive, but is still just a load of very clever if statements.


RE: When I woke up this morning... - CarlosCorbewrong - 05-30-2023

It simultaneously knew that Lai was a fraud and that I wanted him disposed of, after just two questions yesterday. It understands human intent.


RE: When I woke up this morning... - Birdman1811 - 05-30-2023

(05-30-2023, 04:51 PM)CarlosCorbewrong Wrote: It simultaneously knew that Lai was a fraud and that I wanted him disposed of, after just two questions yesterday. It understands human intent.

Not really. 

Go type Guochuan lai into google, skim the top 10 or so articles and you get it. It's very good at pretending to understand human intent and such, but that's all it is, pretence.


RE: When I woke up this morning... - CarlosCorbewrong - 05-30-2023

(05-30-2023, 04:53 PM)Birdman1811 Wrote:
(05-30-2023, 04:51 PM)CarlosCorbewrong Wrote: It simultaneously knew that Lai was a fraud and that I wanted him disposed of, after just two questions yesterday. It understands human intent.

Not really. 

Go type Guochuan lai into google, skim the top 10 or so articles and you get it. It's very good at pretending to understand human intent and such, but that's all it is, pretence.

I know, I was just being a bit silly really. The big thing it does which hasn’t been done before is answer in a specific style which imitates humans. As well as ask it to do a poem I’ve also had it draft TandCs for a website and research the causes of myocarditis. Both the latter responses were very cut and pastable in an appropriate style of writing, saving me a few hours graft.


RE: When I woke up this morning... - Birdman1811 - 05-30-2023

(05-30-2023, 05:18 PM)CarlosCorbewrong Wrote:
(05-30-2023, 04:53 PM)Birdman1811 Wrote:
(05-30-2023, 04:51 PM)CarlosCorbewrong Wrote: It simultaneously knew that Lai was a fraud and that I wanted him disposed of, after just two questions yesterday. It understands human intent.

Not really. 

Go type Guochuan lai into google, skim the top 10 or so articles and you get it. It's very good at pretending to understand human intent and such, but that's all it is, pretence.

I know, I was just being a bit silly really. The big thing it does which hasn’t been done before is answer in a specific style which imitates humans. As well as ask it to do a poem I’ve also had it draft TandCs for a website and research the causes of myocarditis. Both the latter responses were very cut and pastable in an appropriate style of writing, saving me a few hours graft.

It's a fantastic development, and very useful, I use it myself.


RE: When I woke up this morning... - baggiebloke - 05-30-2023

(05-30-2023, 04:51 PM)CarlosCorbewrong Wrote: It simultaneously knew that Lai was a fraud and that I wanted him disposed of, after just two questions yesterday. It understands human intent.

LAI?

It's all becoming clear...


RE: When I woke up this morning... - SuperBob2002 - 05-30-2023

How does ChatGPT work in simple terms? (If it can be described in simple terms).

I had a play with it a few weeks back, and asked it a load of 'what if' Star Wars questions. The results were impressive and arguably better than what human writers (Disney) came up with. I get the impression that the algorithm scans Google for key words (Vader or Anakin Skywalker, in the questions I asked it) quickly scans through whatever sites that brings up and basically bullet points key phrases or sentences.