sigh. when you reboot via terminal, be sure you are on the right machine.... 

i know better.. 

I think i tried that, and it went sideways. But it could have been me too.

Mon Nov 27 2023 17:40:44 EST from darknetuser

So, Citadel (and therefore ncensored) has nice IMAP support that allows to check on the rooms from an email client. So far, so good.

I wonder if integration is complete enough that you could use an email client for both reading AND replying to messages in rooms.

Its been a while, reading was fine.  When i replied i think it went to the person that posted and not the room.

I didnt try a 2nd time.  I meant to do more testing when i had mine running again, but never got around to it. And to be honest forgot all about that until just now.

Mon Nov 27 2023 18:42:46 EST from darknetuser

2023-11-27 18:40 from Nurb432
I think i tried that, and it went sideways. But it could have been me

too.

Do you mean you could read your IMAP rooms but not reply to them?

In theory you can set rooms to accept external email as messages to be posted.

So this afternoon i had watched a few minutes of an agents of shield video. ( mostly testing my new TV streamer i mentioned in another room and how it deals with chromecast end points ) just now i was messing with a new AI model. its response"  "Let me explain this to you like we're talking about your favorite superheroes." then talked about a magical place.

For those that dont know..  the magical place of Tahiti is a huge plot element in the show..   . . wtf. wtf.., 

Kill it with fire

And, for what its worth i was NOT asking it for a story. I have set of python programming questions i send to new models, to judge their output. ( my only real use case ) And other than talking about super heroes, and storing the data in a magical place, it did produce decent code. not perfect, but decent.  And while it responded a bit faster than some, i do have better models so wont be using that one.  So, in effect its dead now. But soooooooo odd.

May have mentioned it before, but my image AI server, one evening i was talking to someone ( in person ) about a subject, then out of the blue, all the AI did was related images..  ( and i have seen that sort of thing since, not as dramatic tho, where *everything* was..  )

They all operate off a certain level of randomness ( unless you carry the seed value over, for replication/continuation purposes ), so it seems that we have more influence over 'random' than we think we do. Quantum mechanics for the win? 

Sat Dec 02 2023 08:20:24 EST from nonservator

Kill it with fire

If it refused, you used the wrong model.  Most likely something like ChatGPT. Every model is different, training, scope, restrictions, intent, etc.   Few of us in that world as a hobby use the commercial fronted models. Or their servers.  Many in the business world use their own trained models too. 

Also, prompt engineering is critically important, regardless of what model you use.  its still part science, part art..

Sat Dec 02 2023 11:59:05 EST from IGnatius T Foobar

It isn't intelligent. It isn't thinking. And it has its own agenda.

I asked it to write a sort algorithm, just to see what it would do. And it refused because it assumed I was going to use it for nefarious purposes.
Not even kidding.

Correct me if I'm wrong but it seems "Artificial Intelligence" is not what the media (and alternative media) portray it as. From what I can tell, it's just a computer doing what it is programmed to do. Sure, it can do all kinds of stuff autonomously, (like generating pictures or whatever) but only because it's programmed to, no?

The difference is modern AI is programmed similar to the same way we learn and uses very similar techniques as humans for determining 'next steps' ( results ).. Including a bit of randomness we perceive as creativity.   And several are now being trained on the output of earlier AI, much like humans are trained by the previous generation of people.

But AI has far more stored than any one human can.. and can access it far faster than we can.

Its got enough information mapped and enough randomness now that even the true experts that invented the algorithms are not 100% sure how its doing some of the things it does.

While im not saying its 'sentient' and has a concept of self-existence, its not 'just a computer doing what its programmed to do' either. Not any more.. 

Sun Dec 03 2023 10:07:40 EST from msgrhys

Correct me if I'm wrong but it seems "Artificial Intelligence" is not what the media (and alternative media) portray it as. From what I can tell, it's just a computer doing what it is programmed to do. Sure, it can do all kinds of stuff autonomously, (like generating pictures or whatever) but only because it's programmed to, no?

Let's see if that image AI has any actual value.  Can it draw kitties?  Attach result images.

Sat Dec 02 2023 09:03:08 EST from Nurb432

May have mentioned it before, but my image AI server, one evening i was talking to someone ( in person ) about a subject, then out of the blue, all the AI did was related images..  ( and i have seen that sort of thing since, not as dramatic tho, where *everything* was..  )

i could have worked on it more to make it more realistic, but ya, it does cats.

Then unlike some other drawing AIs, yours has value.

Sun Dec 03 2023 14:24:47 EST from Nurb432

i could have worked on it more to make it more realistic, but ya, it does cats.

 

The media hype is diminishing what the true advances are.  I personally think its a bigger deal then you do :) 

So here, like in most areas now, reading comprehension of children is going down the tubes.

I guess here they are going to mandate a new (?) way of teaching reading skills, using 'more than 40 years of research into how people learn and comprehend speech'

The short version is they are going to teach kids context.. So in a sentence to look at each word and compare it to the others in the sentence to identify each word in roper context of the others, and be able  to predict future words in a sentence..

Ummm..  sounds familiar?  lol.