skirtcafe.org

Uncle Al wrote: ↑Sun Jun 11, 2023 7:06 pm I've never heard of LLM's before reading this so, I 'googled' what it meant.
What I just read was an eye opener
Much of the information goes back into 2018.

Yes - there's a great talk on them (it focuses mainly on the dangers of AI) and the presenters essentially state that LLM's - large language models - not only are useful for text, but can be applied to a broad variety of things (images, sound, medical data) and they work equally as well:

https://www.youtube.com/watch?v=xoVJKj8lcNQ

crfriend wrote: ↑Thu Mar 30, 2023 12:28 am ...
a core-dump on green-bar paper, and a dropped/shuffled card deck from the computing centre.
...

Just how old are you?
I once dropped my box of cards when walking from my car in the parking lot to the data center at Texas Tech. It was a windy spring afternoon and the 30-40mph wind did its thing on the couple hundred of cards that were on the ground. Some of those cards were last seen flying east at 30mph over east Lubbock.

So we had a chatbot troll, I'll keep an eye out for the sucker as the last time I looked a chatbot has no need for a skirt.

moonshadow wrote: ↑Sun Jun 11, 2023 6:52 pm

ScotL wrote: ↑Sun Jun 11, 2023 5:42 pm I’d happily meet anyone who visits South Florida. Problem is, I only know of one other Floridian and I think he’s up in the Tampa area.

Whereabouts? We've been that way recently.

Miami

Coder wrote: ↑Sun Jun 11, 2023 6:24 pm

Barleymower wrote: ↑Sun Jun 11, 2023 6:07 pm Darn I thought I had moonshadow spot on as the imposter. Then you have post up some photos of yourself. I don't think the bot would do that. Skurtcafe is turning into 'among us'

OMG - I love that game. Very apt comparison.

Barleymower wrote: ↑Sun Jun 11, 2023 6:14 pm I think we could narrow down who the imposter is:
1. Do the mods 'not' reply to anyone's posts?
2. Is the suspect a recent joiner of the cafe in the last 12 months?
3. Have they posted any pictures?
Unless I'm offcourse?

I've been using that criteria as well. Keep in mind LLM's / "AI" have been around - easily accessible to the general public - only since November 2022 - though researchers and geeks have had access to them for a few years now. I think LLM's have been able to achieve a higher-functioning chatbot, one that in small doses can be undetectable, so suspect that kind of technology would be a first pick.

Just an FYI. Posting pictures is not the easiest for some of us. But is that a rule that identifies one as “not a bot?”

ScotL wrote: ↑Sun Jun 11, 2023 9:14 pm Just an FYI. Posting pictures is not the easiest for some of us. But is that a rule that identifies one as “not a bot?”

No - it's just another data point. Plus, someone crafty enough could use generative AI to create a bunch of fictitious images, they would just have to remove extra arms or fingers

phathack wrote: ↑Sun Jun 11, 2023 7:50 pm

crfriend wrote: ↑Thu Mar 30, 2023 12:28 am ...
a core-dump on green-bar paper, and a dropped/shuffled card deck from the computing centre.
...

Just how old are you?

I'm old enough to remember the days of batch-processing and submitted decks of cards. Those sequence numbers are there for a reason, and the nice diagonal lines on the sides were there for a reason as well.

ScotL wrote: ↑Sun Jun 11, 2023 9:10 pmMiami

A data-centre at CloudFlare.

BouffantBelle wrote: ↑Sun Jun 11, 2023 7:19 pm Well I don't mean to be a belligerent bugger, but i'm kind of hoping there is no bot and you're all completely wrong about all of this...

If there isn't one, one of the members was so obnoxious towards me a while ago that I have kill-filed him (or it).

When I was a Nuclear Reactor Operator one of my jobs was to load the "Safety Computer" with data for the new run. This computer kept track of the temperature of all of the fuel and targets in the reactor. The targets were for making a product that was used in several different ways. We did not product steam to generate Electricity. the program came in two ways, one a stack of more than 500 punch cards all lined up in sequence and on paper punch tape. The cards contained the actual data and the tape had the program to run.
Oh what joy when the supervisor accidentally knocked the box with the cards on to the floor, It took a week to get a new box of cards. So we stayed shut down for an extra week because of that.

Fun and games!!

Fred

crfriend wrote: ↑Sun Jun 11, 2023 10:23 pm

ScotL wrote: ↑Sun Jun 11, 2023 9:10 pmMiami

A data-centre at CloudFlare.

Are you accusing me of being a bot because of where I live? There are over 2.6 million people who live in the Miami Dade county.

pelmut wrote: ↑Sun Jun 11, 2023 10:26 pm

BouffantBelle wrote: ↑Sun Jun 11, 2023 7:19 pm Well I don't mean to be a belligerent bugger, but i'm kind of hoping there is no bot and you're all completely wrong about all of this...

If there isn't one, one of the members was so obnoxious towards me a while ago that I have kill-filed him (or it).

What does kill-filed mean?

Coder wrote: ↑Sun Jun 11, 2023 9:18 pm

ScotL wrote: ↑Sun Jun 11, 2023 9:14 pm Just an FYI. Posting pictures is not the easiest for some of us. But is that a rule that identifies one as “not a bot?”

No - it's just another data point. Plus, someone crafty enough could use generative AI to create a bunch of fictitious images, they would just have to remove extra arms or fingers

But isn’t that just the point? Can’t AI beat all of the rules one would put up to prove someone is a human? Outside of meeting in person. But then I’m led to believe there is a human behind the bot. So couldn’t that human just meet someone and “prove they’re not a bot” but still continue to be a bot?

Fred in Skirts wrote: ↑Mon Jun 12, 2023 1:07 am ... The cards contained the actual data and the tape had the program to run.
Fred

I designed a system that worked the other way around: the program (ALGOL) was on cards and the data was automatically logged on paper tape.  We were at a remote field station and everything had to be sent to the central computer by internal post. The operators would run the program but the tape was stored separately on a shelf where it wasn't in view, so the cards would come back with a note saying "No data".  A few days later the operators would find the tape but, by then they had returned the program cards, so the tape would come back with a note saying "No program".

This went on for several weeks and it became very frustrating.  I eventually solved the problem by asking my brother-in-law to mould some transparent plastic boxes that held a stack of cards and a roll of tape, side-by-side in the same container, with a slide-out lid.  The operators then found it more convenient to store the data and program together, so they ran them simultaneously.

Many years later I heard that the paper tape reader had never been much of a success and only one programmer (which must have been me) had ever managed to get any use out of it.   ...hardly surprising.

ScotL wrote: ↑Mon Jun 12, 2023 2:28 am

Coder wrote: ↑Sun Jun 11, 2023 9:18 pm

ScotL wrote: ↑Sun Jun 11, 2023 9:14 pm Just an FYI. Posting pictures is not the easiest for some of us. But is that a rule that identifies one as “not a bot?”

No - it's just another data point. Plus, someone crafty enough could use generative AI to create a bunch of fictitious images, they would just have to remove extra arms or fingers

But isn’t that just the point? Can’t AI beat all of the rules one would put up to prove someone is a human? Outside of meeting in person. But then I’m led to believe there is a human behind the bot. So couldn’t that human just meet someone and “prove they’re not a bot” but still continue to be a bot?

It seems to me that the only solid way to "prove" that one is human is for moderators to vet members whose human identity they can guarantee by having seen them in the flesh and conversed with them; these proven humans thus become persons who in turn can guarantee "second-order" members by also meeting them in the flesh ... and so on.

Anyone whom Moon has met is thus a real human member since Moon has been approved by Carl.

I put my claim to being a real human by dint of having been "proven" by Stevie in Peterborough, who strikes me as being thoroughly human and reputable, even if I can't be sure that he has been vetted by persons higher in the chain.

If a member here thinks they might be the one whom everyone thinks is the bot, but you can't meet or phone for whatever reason...

Send a handwritten postcard with your username and a friendly message, preferably in CURSIVE to

448 Cummings Street Ste 182
Abingdon, VA 24210

Let's see ChatGPT do that! It may not be completely "bot proof", but it certainly lends at least a small amount of legitimacy.

Only cost 44 cents.