Here's a conversation I had with Bard just now about something I asked
it to write. After composing the text it cited its source:

Bard: Sources:
www.xyzzy.com/Sustainability/Policies/Pages/default.aspx

Me: Bard, be careful about sourcing material from www.xyzzy.com.
Nearly all the English text there is from us.

Bard: Understood. I will be careful about sourcing material from
www.xyzzy.com. I will avoid using any text that is identical to or
closely based on text that I have translated for you. I will also try
to find other sources of information to corroborate the information
that I find on the Xyzzy website.

I appreciate you letting me know about this. I want to make sure that
I am providing you with accurate and original information.

--
Bob
Sidera errantia quibus procella tenebrarum in aeternum servata est

The original article was posted to:
1. alt.english.usage, and
2. allt.usage.english
and I have taken the liberty of correcting the latter to
alt.usage.english

Opinicus:

> Here's a conversation I had with Bard just now about
> something I asked it to write. After composing the text it
> cited its source:

Do you use it professionally? Have you the craftsman's
pride and satisfaction of creating something entirely by
yourself? For, unlike a thesaurus, AI is not "just a
tool"...

> Bard: Sources:
> www.xyzzy.com/Sustainability/Policies/Pages/default.aspx

I see no such page, but the domain reminded me of
interactive fiction, and I checked the site with a
forefeeling that another great meme is .com-merically
abused. I was glad that it actually is about IF, though.
The title quotation is wrong:

You are in a twisty maze of passages, all alike.

The original is:

You are in a maze of twisty little passages, all alike.

I wonder what is the purpose of that website, with great and
huge and numerous collections of adventure games out there,
I myself owning the MS-DOS versions of nearly all classic
Infocom games, although I have not finished them all.

Bard:

> I appreciate you letting me know about this.

He certainly is certainly picking up teenspeak. Adults say
"your letting me know."

--
() ascii ribbon campaign -- against html e-mail
/\ www.asciiribbon.org -- against proprietary attachments

On Fri, 20 Oct 2023 20:12:46 +0300, Anton Shepelev
<anton.txt@gmail.moc> wrote:

First off, thanks for correcting "allt.usage.english";
"www.xyzzy.com/Sustainability/Policies/Pages/default.aspx" was my
anonymization of a real address that linked to one of my past
translations.

To answer your questions: Yes, I do use all three LLMs (Bard, Bing,
ChatGPT) professionally and I do take pride in my work. As for the
reason why, here's a (redacted) email that I sent a potential client
(a ready-to-wear manufacturer) a couple of weeks ago. They'd got my
name from someone and asked me if I'd be interested in doing a
translation of their catalog. It was all pictures of their offerings
and fairly brief captions describing them. They also sent a sample of
one caption that someone had translated to English and asked my
opinion. (Apparently it had been criticized.)

<email>
Thank you for asking about my services. I'm afraid this isn't a
project I'd be interested in. However I can say that the English in
the sample you sent was both correct and informative. There's no
reason why you shouldn't use it.

Because catalogs don't deal with complex subjects or use complex
language, their translation is something that AI has become quite good
at doing. I can recommend two for your project: Google Bard and
Microsoft Bing.

I gave one of the two-sentence Turkish captions you sent me to both
programs. Here are the results;

<Bard's translation>

<Bing's translation>

To be brutally honest, I couldn't have done much better myself.
Neither program will make any mistakes in English spelling or grammar,
you can be sure of that, and both will almost always get the
terminology right. However sometimes one LLM is better than the other.
For instance in this example, I liked Bard's first sentence but not
the second and Bing's second sentence but not the first. So I picked
the better of each and combined them into this result:

<"My" translation>

This is the crucial role that only a human translator can and must
play when using AI tools.

Best regards,
Bob
</email>

When a retired college professor friend of mine looked somewhat
askance at my using LLMs I asked her if she had graduate assistants
when she worked at university and she said of course she did. So I
asked "And what did they do?" She explained that they did research,
gathered material, analyzed results, wrote reports etc. "And what did
they do with these reports?" I asked. "They gave them to me." "And
what did you do with them? "I read them, corrected them, compared
them, combined… Oh, I see."

The moral: Bard, Bing, and ChatGPT are like "graduate assistants".
They do all the donkey-work and the professor (user) gets all the
credit.

It's to Parker & Stone's credit that in the end-credits to "Deep
Learning", they cited ChatGPT as one of the episode's authors.

--
Bob
The people your parents warned you about

> The original article was posted to:
> 1. alt.english.usage, and
> 2. allt.usage.english
> and I have taken the liberty of correcting the latter to
> alt.usage.english
>
> Opinicus:
>
> > Here's a conversation I had with Bard just now about
> > something I asked it to write. After composing the text it
> > cited its source:
>
> Do you use it professionally? Have you the craftsman's
> pride and satisfaction of creating something entirely by
> yourself? For, unlike a thesaurus, AI is not "just a
> tool"...
>
> > Bard: Sources:
> > www.xyzzy.com/Sustainability/Policies/Pages/default.aspx
>
> I see no such page, but the domain reminded me of
> interactive fiction, and I checked the site with a
> forefeeling that another great meme is .com-merically
> abused. I was glad that it actually is about IF, though.
> The title quotation is wrong:
>
> You are in a twisty maze of passages, all alike.
>
> The original is:
>
> You are in a maze of twisty little passages, all alike.
>
> I wonder what is the purpose of that website, with great and
> huge and numerous collections of adventure games out there,
> I myself owning the MS-DOS versions of nearly all classic
> Infocom games, although I have not finished them all.
>
> Bard:
>
> > I appreciate you letting me know about this.
>
> He certainly is certainly picking up teenspeak. Adults say
> "your letting me know."

Opinicus:

> "www.xyzzy.com/Sustainability/Policies/Pages/default.aspx"
> was my anonymization of a real address that linked to one
> of my past translations.

www.xyzzy.com is real website about a real and very good
computer game that, being a pioneering example of
interactive fiction, is very much on-topic in this group.

> To answer your questions: Yes, I do use all three LLMs
> (Bard, Bing, ChatGPT) professionally and I do take pride
> in my work.

Once you have employed an LLM, the work is no longer
entirely yours, because the LLMs were trained on the works
of other authors, and often without their permission. The
LLM is your co-author.

> As for the reason why, here's a (redacted) email that I
> sent a potential client (a ready-to-wear manufacturer) a
> couple of weeks ago.

I like modern English, where some manufacturers are ready-
to-wear!

> [...]

I liked your e-mail. Working with you must a pleasant
experience.

> For instance in this example, I liked Bard's first
> sentence but not the second and Bing's second sentence but
> not the first. So I picked the better of each and combined
> them into this result

You delegated the creative part to the LLM, but retained the
critical part of the work.

> When a retired college professor friend of mine looked
> somewhat askance at my using LLMs I asked her if she had
> graduate assistants when she worked at university and she
> said of course she did. So I asked "And what did they do?"
> She explained that they did research, gathered material,
> analyzed results, wrote reports etc. "And what did they do
> with these reports?" I asked. "They gave them to me." "And
> what did you do with them? "I read them, corrected them,
> compared them, combinedЕ Oh, I see."
>
> The moral: Bard, Bing, and ChatGPT are like "graduate
> assistants". They do all the donkey-work and the
> professor (user) gets all the credit.

I see several problems with your interpretation. First,
modern AI in general and LLMs in particular are capable of
way more that Donkey work. Second, the credit is largely
not due the human user, because AI is not a tool (which the
user understands and precisely controls), but a co-creator
trained on a corpus of previous art (often stolen).

> It's to Parker & Stone's credit that in the end-credits to
> "Deep Learning", they cited ChatGPT as one of the
> episode's authors.

Your refernece to this de-jure credits seems incompatible
with your stated pride in your work. Do you pride youself
on having prompted the LLM and appreciated its output? That
is the menial work of a prompt engineer, not that of a
creator, and the formal attribution of the credit to the
user of LLM is dubious.

P.S.: Ouch! -- why did you switch to top-posting?

--
() ascii ribbon campaign -- against html e-mail
/\ www.asciiribbon.org -- against proprietary attachments

Opinicus:

> "www.xyzzy.com/Sustainability/Policies/Pages/default.aspx"
> was my anonymization of a real address that linked to one
> of my past translations.

www.xyzzy.com is real website about a real and very good
computer game that, being a pioneering example of
interactive fiction, is very much on-topic in this group.

> To answer your questions: Yes, I do use all three LLMs
> (Bard, Bing, ChatGPT) professionally and I do take pride
> in my work.

Once you have employed an LLM, the work is no longer
entirely yours, because the LLMs were trained on the works
of other authors, and often without their permission. The
LLM is your co-author.

> As for the reason why, here's a (redacted) email that I
> sent a potential client (a ready-to-wear manufacturer) a
> couple of weeks ago.

I like modern English, where some manufacturers are ready-
to-wear!

> [...]

I liked your e-mail. Working with you must a pleasant
experience.

> For instance in this example, I liked Bard's first
> sentence but not the second and Bing's second sentence but
> not the first. So I picked the better of each and combined
> them into this result

You delegated the creative part to the LLM, but retained the
critical part of the work.

> When a retired college professor friend of mine looked
> somewhat askance at my using LLMs I asked her if she had
> graduate assistants when she worked at university and she
> said of course she did. So I asked "And what did they do?"
> She explained that they did research, gathered material,
> analyzed results, wrote reports etc. "And what did they do
> with these reports?" I asked. "They gave them to me." "And
> what did you do with them? "I read them, corrected them,
> compared them, combinedЕ Oh, I see."
>
> The moral: Bard, Bing, and ChatGPT are like "graduate
> assistants". They do all the donkey-work and the
> professor (user) gets all the credit.

I see several problems with your interpretation. First,
modern AI in general and LLMs in particular are capable of
way more that Donkey work. Second, the credit is largely
not due the human user, because AI is not a tool (which the
user understands and precisely controls), but a co-creator
trained on a corpus of previous art (often stolen).

> It's to Parker & Stone's credit that in the end-credits to
> "Deep Learning", they cited ChatGPT as one of the
> episode's authors.

Your refernece to this de-jure credits seems incompatible
with your stated pride in your work. Do you pride youself
on having prompted the LLM and appreciated its output? That
is the menial work of a prompt engineer, not that of a
creator, and the formal attribution of the credit to the
user of LLM is dubious.

P.S.: Ouch! -- why did you switch to top-posting?

--
() ascii ribbon campaign -- against html e-mail
/\ www.asciiribbon.org -- against proprietary attachments

[Google Groups dropped the cross-post, reinstating]

Lionel Edwards to Anton Shepelev:

> > He certainly is certainly picking up teenspeak. Adults
> > say "your letting me know."
>
> Both are good English.

Lately I have been overinduliging myself in black-and-white
drama and noir movies of 1940s, 50s, and the first half of
the 60s. Your remark above having put me on the alert for
fused participles, I have observed that educated characters
invaribly use the genitive form (`your') and uneducated ones
the accusative (`you').

I was not watching with a pen and paper at hand (it would
have taken effort, and not have been indulgence :-), so you
have to take my word for it. The last one was "The Reckless
Moment" -- compare the speech of the Harpers against that of
their black servant.

This shows that the accusative fused participle is a recent
invention of the illitirate, a mere distortion due to a lack
reading and a formal study of grammar. The same black
servant says something like "so that you would not forget it
again, /like/ the last time," instead of /as you did/.

By declaring both usages good English you are degrading the
language to the level of its worst speakers, or
"democratising" it to use a stinking euphemism. Many
relaxations of modern English syntax are of the same base
origin, such as misusing "due to" instaead of "because" (the
former is an adverbial modifier, something /is/ due to
something). This regress towards more primitive form. Stop
it, and save English.

--
() ascii ribbon campaign -- against html e-mail
/\ www.asciiribbon.org -- against proprietary attachments

I wrote:

> This shows that the accusative fused participle is a
> recent invention of the illitirate,

And even today someone wrote to me:

> This is an excellent distillation of my lengthy
> deconstruction of that guy's own undisciplined rant. I
> appreciate you expressing it so economically.

and he the maintainer of a major typographical package,
apparently a native, and certainly a good, speaker and
writer in English, but sloppiness is like charos, enthropy,
and evil: it is enough if good men remain idle for it to
prevail...

--
() ascii ribbon campaign -- against html e-mail
/\ www.asciiribbon.org -- against proprietary attachments