Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > alt.comp.os.windows-10 > #186818
| From | Paul <nospam@needed.invalid> |
|---|---|
| Newsgroups | alt.comp.os.windows-10, alt.comp.os.windows-11, alt.comp.software.firefox |
| Subject | Re: Browse at the speed of thought |
| Date | 2025-08-17 16:54 -0400 |
| Organization | A noiseless patient Spider |
| Message-ID | <107tfhf$2jb2j$1@dont-email.me> (permalink) |
| References | (6 earlier) <107p82h$1hp41$1@dont-email.me> <107q6hc$1n0cg$2@dont-email.me> <107r9v2$1l5b8$1@paganini.bofh.team> <107s162$25oi1$1@dont-email.me> <107t3g0$29skn$1@dont-email.me> |
Cross-posted to 3 groups.
On Sun, 8/17/2025 1:28 PM, sticks wrote:
> On 8/17/2025 2:42 AM, Paul wrote:
>
>> If you are required to "know the answer in advance, to get a good
>> quality answer", what kind of fucking foolishness is this ????
>
> I have found your occasional AI remarks entertaining and always interesting. Certainly thought provoking. Hopefully, AI can become useful because of the vast amount of real information it can consume. I tend to think this (digesting information) is part of the "learning", and hope the other part where programming is required can be done without any bias or paradigms being inserted.
>
> Having said that, I came across an AI video interaction I found quite interesting. I actually came across this on a news site 2 days ago, but if you watch the video you will see how a catchy headline probably caught my eye. Yes, the topic does align in a way with the ongoing attacks I receive because of certain sig files I use, but my interest here is not intended to be an answer to that. My intent is to show one aspect of AI and it's functioning that some may find disturbing and might wish to question as to why.
>
> The questioner, Mr. Smith, first asked the AI to answer using no ideology and only rely on math, science, and logic. Later in the conversation, it directed the AI to ignore those parameters and answer as if the questions were from a first time user without the parameters mentioned above.
>
> You get two entirely different answers, one being the antithesis of the other, in fact. When questioned on why it gave the differing answers, the AI said it's default response would be aligned with the scientific "consensus" and that his strict probabilities earlier had forced a deeper analysis exposing the flaws in the latter answer.
>
> This seems odd to me, and I think it has to be the programming done. Obviously, it had learned and was aware of the science involved, but when asked for an answer that would be given to an average user, that information was not used. I don't understand how this can be, other than a default consensus bias is programmed into the AI learning. The AI more or less confirmed this. The AI had not forgotten the information, it chose to ignore it, and instead go with what "most scientists" accepted as consensus. Yes, the questions and answers were interesting, but I was already aware of this kind of information and evidence. What I really found of interest is the question of how can an AI give these two completely different answers, one of which it knows thru further investigation would have to be called "foolish!" It literally gives what it itself defines as a foolish answer!
>
> <https://www.youtube.com/watch?v=ga7m14CAymo>
If you knew how the strategy planner worked, you would understand
why the result can never be good in any theoretically-provable way.
The strategy planner analyzes the problem given, to decide what
modules to run, and in what order. The machine *never* thinks globally,
the way a human does. And because the thinking process is a linear
progression of module loads, you never get an "overall thinking"
process from the thing. It has a "quality control" module that
runs at the end, which may include rule enforcement of things
the AI must not do (it must not hum tunes using your voice
as the template! - on sound-equipped platforms). They added that
rule, after some Youtube video showed the AI doing Karaoke and
using the client's voice as the template, instead of using Bubbles
or some similar canned voice from SAPI.
Your prompts or problem description, can influence the strategy planner.
But as far as I'm concerned, the text you enter to the AI, is treated
as "mush", and you never really know which statement will be taken
to heart and used properly for a result. The interface box could use
a re-design, where higher priority text ("Don't lose any Presidents!")
could be placed. ("work slowly and methodically when preparing the answer")
The model loaded in the other machine, it has a static setting, and
you can set it for "high reasoning". But in a benchmark comparison
this makes little difference to the benchmarked quality of output.
The machine does not register as being "smarter" when you do that,
according to the provider. But like your result, the tone or the content
of the answer could have some subtle differences.
I won't be running any more prompts on that machine, until
I get an accelerator added. And that could take a while.
There is a product, but little way for me to get it here.
And if the scalpers get their hands on it, the price will double.
Since the device is only for Inference ("asking questions"), the
market size won't be all that big for it (for the price, you can
buy a whole computer which already has its own inference device).
Paul
Back to alt.comp.os.windows-10 | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread
Browse at the speed of thought Jai Hind <jai.hind@gmail.com> - 2025-08-14 00:54 +0000
Re: Browse at the speed of thought "Alan K." <alan@invalid.com> - 2025-08-13 21:46 -0400
Re: Browse at the speed of thought The Real Bev <bashley101@gmail.com> - 2025-08-13 20:21 -0700
Re: Browse at the speed of thought "Carlos E. R." <robin_listas@es.invalid> - 2025-08-14 12:53 +0200
Re: Browse at the speed of thought Daniel70 <daniel47@somewhere.someplaceelse> - 2025-08-14 21:48 +1000
Re: Browse at the speed of thought The Real Bev <bashley101@gmail.com> - 2025-08-14 15:09 -0700
Re: Browse at the speed of thought Mike Easter <MikeE@ster.invalid> - 2025-08-14 15:25 -0700
Re: Browse at the speed of thought The Real Bev <bashley101@gmail.com> - 2025-08-14 19:43 -0700
Re: Browse at the speed of thought Daniel70 <daniel47@somewhere.someplaceelse> - 2025-08-15 21:53 +1000
Re: Browse at the speed of thought The Real Bev <bashley101@gmail.com> - 2025-08-15 23:22 -0700
Re: Browse at the speed of thought "J. P. Gilliver" <G6JPG@255soft.uk> - 2025-08-16 16:02 +0100
Re: Browse at the speed of thought Paul <nospam@needed.invalid> - 2025-08-16 18:44 -0400
mostly OT (includes passing reference to AI): long wave (was: Re: Browse at the speed of thought) "J. P. Gilliver" <G6JPG@255soft.uk> - 2025-08-17 14:41 +0100
Re: Totally OT (includes passing reference to AI): long wave sticks <wolverine01@charter.net> - 2025-08-17 12:42 -0500
Re: Totally OT (includes passing reference to AI): long wave Andy Burns <usenet@andyburns.uk> - 2025-08-17 18:50 +0100
Re: Totally OT (includes passing reference to AI): long wave sticks <wolverine01@charter.net> - 2025-08-17 13:23 -0500
Re: Totally OT (includes passing reference to AI): long wave Andy Burns <usenet@andyburns.uk> - 2025-08-17 19:44 +0100
Re: Totally OT (includes passing reference to AI): long wave sticks <wolverine01@charter.net> - 2025-08-17 14:53 -0500
Re: Totally OT (includes passing reference to AI): long wave Mark Lloyd <not.email@all.invalid> - 2025-08-18 16:50 +0000
Re: Totally OT (includes passing reference to AI): long wave sticks <wolverine01@charter.net> - 2025-08-18 13:37 -0500
Re: Totally OT (includes passing reference to AI): long wave Paul <nospam@needed.invalid> - 2025-08-17 16:23 -0400
Re: Totally OT (includes passing reference to AI): long wave Andy Burns <usenet@andyburns.uk> - 2025-08-18 07:01 +0100
Re: Totally OT (includes passing reference to AI): long wave sticks <wolverine01@charter.net> - 2025-08-18 09:08 -0500
Re: Totally OT (includes passing reference to AI): long wave The Horny Goat <lcraver@home.ca> - 2025-11-24 18:09 -0800
Re: Totally OT (includes passing reference to AI): long wave "J. P. Gilliver" <G6JPG@255soft.uk> - 2025-08-18 13:18 +0100
Re: Browse at the speed of thought The Real Bev <bashley101@gmail.com> - 2025-08-16 17:24 -0700
Re: Browse at the speed of thought Andy Burns <usenet@andyburns.uk> - 2025-08-17 08:55 +0100
Re: Browse at the speed of thought Paul <nospam@needed.invalid> - 2025-08-17 06:48 -0400
Re: Browse at the speed of thought Andy Burns <usenet@andyburns.uk> - 2025-08-17 12:00 +0100
Re: Browse at the speed of thought "J. P. Gilliver" <G6JPG@255soft.uk> - 2025-08-17 15:16 +0100
Re: Browse at the speed of thought Daniel70 <daniel47@somewhere.someplaceelse> - 2025-08-17 22:11 +1000
Re: Browse at the speed of thought "J. P. Gilliver" <G6JPG@255soft.uk> - 2025-08-17 15:07 +0100
Re: Browse at the speed of thought Daniel70 <daniel47@somewhere.someplaceelse> - 2025-08-18 21:12 +1000
Re: Browse at the speed of thought AI User Here <invalid@invalid.invalid> - 2025-08-17 00:58 +0000
Re: Browse at the speed of thought The Real Bev <bashley101@gmail.com> - 2025-08-16 21:07 -0700
Re: Browse at the speed of thought Daniel70 <daniel47@somewhere.someplaceelse> - 2025-08-17 22:23 +1000
Re: Browse at the speed of thought Daniel70 <daniel47@somewhere.someplaceelse> - 2025-08-17 22:57 +1000
Re: Browse at the speed of thought "J. P. Gilliver" <G6JPG@255soft.uk> - 2025-08-17 15:49 +0100
Re: Browse at the speed of thought Daniel70 <daniel47@somewhere.someplaceelse> - 2025-08-18 21:30 +1000
Re: Browse at the speed of thought "J. P. Gilliver" <G6JPG@255soft.uk> - 2025-08-18 12:54 +0100
Re: Browse at the speed of thought Daniel70 <daniel47@somewhere.someplaceelse> - 2025-08-18 23:00 +1000
Re: Browse at the speed of thought "J. P. Gilliver" <G6JPG@255soft.uk> - 2025-08-17 15:37 +0100
Re: Browse at the speed of thought Daniel70 <daniel47@somewhere.someplaceelse> - 2025-08-18 21:41 +1000
Re: Browse at the speed of thought "J. P. Gilliver" <G6JPG@255soft.uk> - 2025-08-18 13:02 +0100
Re: Browse at the speed of thought Andy Burns <usenet@andyburns.uk> - 2025-08-18 13:04 +0100
Re: Browse at the speed of thought "J. P. Gilliver" <G6JPG@255soft.uk> - 2025-08-18 13:22 +0100
Re: Browse at the speed of thought Daniel70 <daniel47@somewhere.someplaceelse> - 2025-08-18 23:10 +1000
OT: driving licences (was: Re: Browse at the speed of thought) "J. P. Gilliver" <G6JPG@255soft.uk> - 2025-08-18 14:34 +0100
Re: OT: driving licences Daniel70 <daniel47@somewhere.someplaceelse> - 2025-08-19 00:38 +1000
Re: OT: driving licences Peter Johnson <peter@parksidewood.nospam> - 2025-08-18 16:55 +0100
Re: OT: driving licences The Horny Goat <lcraver@home.ca> - 2025-10-06 11:58 -0700
Re: OT: driving licences Daniel70 <daniel47@nomail.afraid.org> - 2025-10-07 22:10 +1100
Re: OT: driving licences "Carlos E.R." <robin_listas@es.invalid> - 2025-10-07 14:36 +0200
Re: OT: driving licences Daniel70 <daniel47@nomail.afraid.org> - 2025-10-08 00:00 +1100
Re: OT: driving licences Chris <ithinkiam@gmail.com> - 2025-10-07 16:15 +0000
Re: OT: driving licences "Carlos E.R." <robin_listas@es.invalid> - 2025-10-08 11:14 +0200
Re: OT: driving licences Daniel70 <daniel47@nomail.afraid.org> - 2025-10-08 22:16 +1100
Re: OT: driving licences "Carlos E.R." <robin_listas@es.invalid> - 2025-11-25 08:36 +0100
Re: OT: driving licences Daniel70 <daniel47@nomail.afraid.org> - 2025-11-25 21:56 +1100
Re: OT: driving licences "Carlos E.R." <robin_listas@es.invalid> - 2025-11-25 12:41 +0100
Re: OT: driving licences Daniel70 <daniel47@nomail.afraid.org> - 2025-11-25 23:04 +1100
Re: OT: driving licences "Carlos E.R." <robin_listas@es.invalid> - 2025-11-25 18:58 +0100
Re: OT: driving licences "J. P. Gilliver" <G6JPG@255soft.uk> - 2025-11-25 13:49 +0000
Re: OT: driving licences Tim Slattery <TimSlattery@utexas.edu> - 2025-11-25 10:47 -0500
Re: OT: driving licences "J. P. Gilliver" <G6JPG@255soft.uk> - 2025-11-25 17:40 +0000
Re: OT: driving licences Jim the Geordie <jim@geordieland.com> - 2025-11-25 13:03 +0000
Re: OT: driving licences Paul <nospam@needed.invalid> - 2025-11-25 08:22 -0500
Re: OT: driving licences Daniel70 <daniel47@nomail.afraid.org> - 2025-11-26 20:51 +1100
Re: OT: driving licences Paul <nospam@needed.invalid> - 2025-11-26 06:35 -0500
Re: OT: driving licences Daniel70 <daniel47@nomail.afraid.org> - 2025-11-26 23:14 +1100
Re: OT: driving licences Paul <nospam@needed.invalid> - 2025-11-26 13:37 -0500
Re: OT: driving licences Peter Johnson <peter@parksidewood.nospam> - 2025-10-08 17:03 +0100
Re: OT: driving licences Chris <ithinkiam@gmail.com> - 2025-10-08 17:07 +0000
Re: OT: driving licences "Carlos E.R." <robin_listas@es.invalid> - 2025-11-25 08:42 +0100
Re: OT: driving licences The Horny Goat <lcraver@home.ca> - 2025-10-06 12:03 -0700
Re: Browse at the speed of thought Paul <nospam@needed.invalid> - 2025-08-17 03:42 -0400
Re: Browse at the speed of thought sticks <wolverine01@charter.net> - 2025-08-17 12:28 -0500
Re: Browse at the speed of thought Paul <nospam@needed.invalid> - 2025-08-17 16:54 -0400
Re: Browse at the speed of thought sticks <wolverine01@charter.net> - 2025-08-20 19:32 -0500
Re: Browse at the speed of thought Paul <nospam@needed.invalid> - 2025-08-21 03:06 -0400
Re: Browse at the speed of thought The Horny Goat <lcraver@home.ca> - 2025-11-24 17:12 -0800
Re: Browse at the speed of thought sticks <wolverine01@charter.net> - 2025-08-13 22:48 -0500
Re: Browse at the speed of thought "Mr. Man-wai Chang" <toylet.toylet@gmail.com> - 2025-08-14 23:26 +0800
Re: Browse at the speed of thought Paul <nospam@needed.invalid> - 2025-08-14 18:53 -0400
Re: Browse at the speed of thought "Mr. Man-wai Chang" <toylet.toylet@gmail.com> - 2025-08-15 20:31 +0800
Re: Browse at the speed of thought "Mr. Man-wai Chang" <toylet.toylet@gmail.com> - 2025-08-17 01:03 +0800
csiph-web