TWAIN-L Archives

Mark Twain Forum

TWAIN-L@YORKU.CA

Options: Use Forum View

Use Monospaced Font
Show HTML Part by Default
Condense Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Sender:
Mark Twain Forum <[log in to unmask]>
Date:
Thu, 18 May 2023 21:34:04 -0400
Reply-To:
Mark Twain Forum <[log in to unmask]>
Message-ID:
Subject:
MIME-Version:
1.0
Content-Transfer-Encoding:
8bit
In-Reply-To:
Content-Type:
text/plain; charset="utf-8"
From:
Parts/Attachments:
text/plain (56 lines)
What you say here strikes me as more reasonable than what I first understood you to say about how you created your Twainbot.  You didn't input "all of Twain's personal data", you inputted 14 pages about his personality, but you don't say which 14 pages.

Now, I am genuinely intending this in a friendly way.  I would suggest you not use the term "subjective" as if inputting more data will make Twainbot any less "subjective".  Every single datum we have about Twain's personality is subjective.  No one alive today ever met him.  Probably few people have ever met people who knew him, although I'd bet there are such people, maybe on this list.  All we have are accounts from other people and from him, and every single one of those accounts is not only subjective, many of them were influenced by earlier subjective accounts.  Adding infinite subjectives together does not equal "objective", nor does it give any sort of truth, not about a long dead person's personality or writing talent or experience of life. Even that person alive in his prime had only a subjective idea of his own personality.  I think the best you can hope for in this pursuit is something that matches the expectations of experts or, as with Hal Holbrook's wonderful portrayal, a Twainbot so polished and convincing that it doesn't matter that it gives us a Twain who never really was.

Still, I think you've made an interesting start, you are modest about your progress so far, and I wish you well with your project.  I think it will be fascinating to see what you come up with.  But please remember that it's all subjective.  It's never gonna give you a "real" Twain, only a sentimentally convincing one.

Regards,

Carl

-----Original Message-----
From: Mark Twain Forum <[log in to unmask]> On Behalf Of DM Sataari
Sent: Thursday, May 18, 2023 5:14 PM
To: [log in to unmask]
Subject: Re: The Rings of Sataari

Yes, I never built an AI from scratch, from the bottom up. There is no need to do so, when they're already available.

I worked on an AI shell which was actually built for roleplaying.
Roleplaying bots work a little differently than other types in that they will give you emotional responses, unlike, let's say, ChatGPT, which sounds like the unemotional bot it is. Roleplaying bots will respond with humor, sarcasm, anger, passion, grief, irony, and they will even attempt to manipulate you. They will do this spontaneously, even without you informing them that they are to be funny or manipulative.

I gave the roleplaying bot a series of statements about Twain's personality, pages and pages of it (14 pages in total), and it uses these statements to "become" the character you describe. But, of course, personality is a subjective thing, and if you wrote 14 pages of notes about Twain's personality it would look different than mine. So there is a subjective aspect to this prototype Twainbot, although in the future it will be possible to input an unlimited number of notes about personality into a bot. And the subjective aspect of its personality will decrease in proportion to the size of the data set used.

Anyway, the important point in all of this is that these bots are primitive experimental first-generation prototypes. Which means these are, by far, the very worst possible bots in all of history -- think of those old Commodore computers with a half meg of RAM in the 80s. But even these most primitive models are already highly nuanced and realistic. And these bots do learn and teach themselves, both from humans who interact with them as well as from other bots as well.

But this message board is about Twain and not about AI, so my interest lies in virtually resurrecting Mr. Clemens so we can speak to him, rather than explaining the emergent properties in the future of AI technology. I don't think most people are really prepared for or comfortable with such a resurrection, so I am going to get an enormous amount of pushback, yet here we are.

Would you like to recommend any sources which contain psychological studies of Twain, as opposed to just his literature? I read something about him being secretly gay, it seemed a bit bonkers....





On Thu, May 18, 2023 at 11:5 AM Clay Shannon <[log in to unmask]> wrote:

> It's true that actually programming a chatbot from scratch is a job 
> requiring a team and a lot of time, but "[computer] programming" 
> actually has two meanings: its literal meaning, and the commonly 
> understood definition. Probably only people who are or have been 
> computer programmers, or at least taken a course or two in it, understand what programming is.
> Most non-programmers think that people who fiddle with computers 
> (super users and IT pros) are programmers (or even people who 
> "program" their VCR, which is really just setting it up or 
> manipulating its settings), but they're not.
>
> So I was assuming that what DM Sataari (which, BTW, looks like an East 
> Indian name to me) meant by "programming" a chatbot was that s/he had 
> modififed an existing one (such as OpenAI's) to focus on Twain 
> material only or something of that nature.
>
> Is the whole DM Sataari thing a hoax, or a social studies experiment?
> Perhaps.
>
> -- B. Clay Shannon
> [log in to unmask]

ATOM RSS1 RSS2