Including AI into Emacs

Including AI into Emacs

[Jean Louis]

Today, I’ve noticed that there are many free software projects that
use artificial intelligence, or AI for short. These projects are
created by developers all around the world who want to make useful
tools available to everyone. AI can help solve complex problems,
automate tasks, and improve the way we interact with technology. It’s
exciting to see so many people contributing to these projects and
making AI more accessible to everyone.

I believe it would be really beneficial to include AI into GNU
Emacs. Emacs is already a powerful tool, but adding AI features could
make it even better. For example, AI could help with writing by
suggesting improvements or catching mistakes. It could also assist
with coding by providing smart autocomplete options or debugging
help. These features would make Emacs more efficient and
user-friendly, helping users get their work done faster and with fewer
errors.

I am using AI every day, and will soon provide references on how to
use it self-contained, as every user can download it and run it on
their own computers.

Jean Louis

Including AI into Emacs

[Jean Louis]

My previous e-mail about AI was written only to test the following
function:

(defun rcd-monitor-directory-for-mail-files (directory)
  "Monitor DIRECTORY for creation of *.mail files, and call `msmtp-count-remaining` when one is created.

DIRECTORY should be the path to the directory you want to monitor.

This function sets up a file notification watch on DIRECTORY.
When a new file with the `.mail` extension is created within DIRECTORY,
the `msmtp-count-remaining` function is called."
  (file-notify-add-watch
   directory
   '(change)
   (lambda (event)
     ;; Structure of EVENT:
     ;; (ACTION EVENT-TYPE FILE)
     (let* ((event-type (cadr event))
            (file-path (caddr event))
            (file-name (file-name-nondirectory file-path)))
       (when (and (eq event-type 'created)
                  (string-suffix-p ".mail" file-name))
         (msmtp-count-remaining))))))

As I was invoking the function always manually with: M-x msmtp-count-remaining

(defun msmtp-count-remaining ()
  "Count and send any remaining MSMTP messages in the queue"
  (interactive)
  (let ((default-directory (rcd-my-home))
	(msmtp-runqueue (executable-find "msmtp-runqueue.sh")))
    (rcd-general-log "Function `msmtp-count-remaining' invoked" nil 1 nil nil nil 6)
    (let ((count (length (directory-files "~/.msmtpqueue" nil "\\.mail"))))
      (if (> count 0)
	  (progn
	    (start-process msmtp-runqueue "RCD MSMTP" msmtp-runqueue)
	    (rcd-message "MSMTP: There is %s e-mails in queue." count))
	(rcd-message "MSMTP: No emails.")))))

Now I can just put:

(rcd-monitor-directory-for-mail-files "~/.msmtpqueue")

in the Emacs configuration and as soon as e-mail is sent, it is also
dispatched with msmtp:

msmtp is an SMTP client that can be used to send mails from Mutt and
probably other MUAs (mail user agents). It forwards mails to an SMTP
server (for example at a free mail provider), which takes care of the
final delivery.  Using profiles, it can be easily configured to use
different SMTP servers with different configurations, which makes it
ideal for mobile clients.

Jean Louis

Re: Including AI into Emacs

[Bruno Barbier]

Hi Jean-Louis,

Jean Louis <bugs@gnu.support> writes:
> ...
> I am using AI every day, and will soon provide references on how to
> use it self-contained, as every user can download it and run it on
> their own computers.
>

Are you aware of the gptel package ?
   https://github.com/karthink/gptel

It allows to use AI pretty much everywhere in Emacs.  And it provides
backends for many existing AI.

Cheers,
Bruno

Re: Including AI into Emacs

[Eli Zaretskii]

> Date: Fri, 06 Dec 2024 20:19:47 +0300
> From: Jean Louis <bugs@gnu.support>
> 
> Today, I’ve noticed that there are many free software projects that
> use artificial intelligence, or AI for short. These projects are
> created by developers all around the world who want to make useful
> tools available to everyone. AI can help solve complex problems,
> automate tasks, and improve the way we interact with technology. It’s
> exciting to see so many people contributing to these projects and
> making AI more accessible to everyone.
> 
> I believe it would be really beneficial to include AI into GNU
> Emacs. Emacs is already a powerful tool, but adding AI features could
> make it even better. For example, AI could help with writing by
> suggesting improvements or catching mistakes. It could also assist
> with coding by providing smart autocomplete options or debugging
> help. These features would make Emacs more efficient and
> user-friendly, helping users get their work done faster and with fewer
> errors.
> 
> I am using AI every day, and will soon provide references on how to
> use it self-contained, as every user can download it and run it on
> their own computers.

There are many 3rd-party packages that add "AI" support to Emacs.

Re: Including AI into Emacs

[Eli Zaretskii]

> Date: Fri, 06 Dec 2024 20:22:23 +0300
> From: Jean Louis <bugs@gnu.support>
> 
> My previous e-mail about AI was written only to test the following
> function:

Please avoid using Emacs public mailing lists for testing your code.

Re: Including AI into Emacs

[John Yates]

On Fri, Dec 6, 2024 at 1:26 PM Eli Zaretskii <eliz@gnu.org> wrote:

> > Date: Fri, 06 Dec 2024 20:22:23 +0300
> > From: Jean Louis <bugs@gnu.support>
> >
> > My previous e-mail about AI was written only to test the following
> > function:
>
> Please avoid using Emacs public mailing lists for testing your code.


Thank you, Eli, for your many examples of polite, gentle, respectful
project leadership.

Re: Including AI into Emacs

[Jean Louis]

On December 6, 2024 9:25:34 PM GMT+03:00, Eli Zaretskii <eliz@gnu.org> wrote:
>> Date: Fri, 06 Dec 2024 20:22:23 +0300
>> From: Jean Louis <bugs@gnu.support>
>> 
>> My previous e-mail about AI was written only to test the following
>> function:
>
>Please avoid using Emacs public mailing lists for testing your code.
>

I am not expressing myself well as usual. I should have avoid to say "only" as it wasn't.

Mailing list wasn't used for testing my code, it can be seen in the function. 

The fact that email was written on my personal computer was used to run the code, as when specific file appears in directory the function is invoked.

I hope it's now clear.


Jean

Re: Including AI into Emacs

[Basile Starynkevitch]

On Fri, 2024-12-06 at 20:19 +0300, Jean Louis wrote:
> Today, I’ve noticed that there are many free software projects that
> use artificial intelligence, or AI for short. These projects are
> created by developers all around the world who want to make useful
> tools available to everyone. AI can help solve complex problems,
> automate tasks, and improve the way we interact with technology. It’s
> exciting to see so many people contributing to these projects and
> making AI more accessible to everyone.
> 
> I believe it would be really beneficial to include AI into GNU
> Emacs. Emacs is already a powerful tool, but adding AI features could
> make it even better. For example, AI could help with writing by
> suggesting improvements or catching mistakes. It could also assist
> with coding by providing smart autocomplete options or debugging
> help. These features would make Emacs more efficient and
> user-friendly, helping users get their work done faster and with
> fewer
> errors.
> 
> I am using AI every day, and will soon provide references on how to
> use it self-contained, as every user can download it and run it on
> their own computers.


On of the issues is to define what is an open source AI system.

Is https://clipsrules.net/ an AI system? (On Linux you could run it
from GNU emacs)?

Can https://github.com/RefPerSys/RefPerSys be extended (or the basis
of) to become an AI system? It is GPLv3+ licensed?

Both software are open source... See also http://refpersys.org/ 

Regards from near Paris in France....

-- 
Basile STARYNKEVITCH <basile@starynkevitch.net>
8 rue de la Faïencerie
92340 Bourg-la-Reine, France
http://starynkevitch.net/Basile & https://github.com/bstarynk 

Re: Including AI into Emacs

[Jean Louis]

* Basile Starynkevitch <basile@starynkevitch.net> [2024-12-06 22:14]:
> On of the issues is to define what is an open source AI system.

I like to use "Free Software" instead of "Open Source" as many times Open Source is not really free, so there are differences. But one has to distinguish it with licenses and research to verify if anything open source is really free.

There are Large Language Models that are allegedly free software, it requires research. Here is one list of it:

eugeneyan/open-llms: 📋 A list of open LLMs available for commercial use.
https://github.com/eugeneyan/open-llms

> Is https://clipsrules.net/ an AI system? (On Linux you could run it
> from GNU emacs)?

That sounds like US government financed software and such is always
released to public domain if I am not mistaken.

There is MIT No Attribution license in the documentation.

> Can https://github.com/RefPerSys/RefPerSys be extended (or the basis
> of) to become an AI system? It is GPLv3+ licensed?

I think yes, why not.

> Both software are open source... See also http://refpersys.org/ 

I wish to see use for me personally, managing documents, understanding my workflows, implementing them in code, like that.

-- 
Jean Louis

Re: Including AI into Emacs

[Jean Louis]

* Bruno Barbier <brubar.cs@gmail.com> [2024-12-06 21:16]:
> 
> Hi Jean-Louis,
> 
> Jean Louis <bugs@gnu.support> writes:
> > ...
> > I am using AI every day, and will soon provide references on how to
> > use it self-contained, as every user can download it and run it on
> > their own computers.
> >
> 
> Are you aware of the gptel package ?
>    https://github.com/karthink/gptel

I was using in invalidated manner the emacs-chatgpt but in general I
can do myself Curl and http requests. My usage is totally specific, I
have memories, various memories that I select from Dynamic Knowledge
Repository, then I can replace region with the information, or add new
information into buffer.

I could not try gptel, it doesn't work. But will try in future.

> It allows to use AI pretty much everywhere in Emacs.  And it provides
> backends for many existing AI.

Soon I will buy Nvidia RTX 4xxx and then I will be using it offline
exclusively.

-- 
Jean Louis

Re: Including AI into Emacs

[Jean Louis]

* Basile Starynkevitch <basile@starynkevitch.net> [2024-12-06 22:12]:
> On of the issues is to define what is an open source AI system.

Without referring to common definitions, all of computing can be
considered artificial intelligence as it extends the mind functions
and memory of humans.

It is just a matter of development level to determine how far it may go.

At first, it might handle tasks like invoices. It could ask the user
every day, "Is there any invoice to enter?" If the user answers "Yes,"
the computer could then request the details.

On the next level, the computer may not need to ask the user but
could, through observation, understand that an invoice needs to be
entered. For example, it could recognize a customer's face through a
camera, identify the customer, and automatically generate an invoice.

All Emacs programs and essentially all programs in the world are forms
of artificial intelligence.

But now there are fantastic new developments where computers can
recognize pictures, generate new images through imagination, create
videos, correct text, and more easily understand what we think. This
is what we have been striving for.

And common users now can use it without tedious programming, and that
is huge advance.

It is like jumping from shell usage of Internet to graphics
browser. That happened in past, and is now happening with the AI
accessibility for all.

Of course, the notorious Emacs psychotherapist is predating all of the
modern chatbots. 

-- 
Jean Louis

Re: Including AI into Emacs

[Christopher Howard]

Jean Louis <bugs@gnu.support> writes:

> I believe it would be really beneficial to include AI into GNU
> Emacs. Emacs is already a powerful tool, but adding AI features could
> make it even better. For example, AI could help with writing by
> suggesting improvements or catching mistakes. It could also assist
> with coding by providing smart autocomplete options or debugging
> help. These features would make Emacs more efficient and
> user-friendly, helping users get their work done faster and with fewer
> errors.
>

Could you clarify what you mean exactly but "including AI into GNU Emacs"? I know for sure I don't want Emacs sending information about my system, or questions that I have about Emacs or my project, off to some company's LLM chat system, or however that works exactly. I do not want to become dependent on some remote computer program or AI in order to be able to write code or figure out how Emacs works.

I'm certain interested in running tools locally (same computer, or my network) that help me with "suggesting improvements, catching mistakes", etc. Do you need something massive like ChatGPT to accomplish that, or just some Emacs-centric expert systems? Or maybe something like MycroftAI, running locally?

-- 
Christopher Howard

Re: Including AI into Emacs

[Jean Louis]

* Christopher Howard <christopher@librehacker.com> [2024-12-07 01:59]:
> Jean Louis <bugs@gnu.support> writes:
> 
> > I believe it would be really beneficial to include AI into GNU
> > Emacs. Emacs is already a powerful tool, but adding AI features could
> > make it even better. For example, AI could help with writing by
> > suggesting improvements or catching mistakes. It could also assist
> > with coding by providing smart autocomplete options or debugging
> > help. These features would make Emacs more efficient and
> > user-friendly, helping users get their work done faster and with fewer
> > errors.
> >
> 
> Could you clarify what you mean exactly but "including AI into GNU Emacs"? I know for sure I don't want Emacs sending information about my system, or questions that I have about Emacs or my project, off to some company's LLM chat system, or however that works exactly. I do not want to become dependent on some remote computer program or AI in order to be able to write code or figure out how Emacs works.

There are large language models (LLMs) that may be downloaded and used
in local network, Ethernet or on the local computer. Emacs itself is
huge contribution to general artificial intelligence. LLMs help in
correction of text, answering the questions, correcting the text,
making headings, emphasizing words.

It is not question of releasing private data, asking remote computer
what, those are other issues, I don't agree to it.

There are local models and free software that shall be accessible
through Emacs. 

Examples: https://github.com/Mozilla-Ocho/llamafile/

> I'm certain interested in running tools locally (same computer, or
> my network) that help me with "suggesting improvements, catching
> mistakes", etc. Do you need something massive like ChatGPT to
> accomplish that, or just some Emacs-centric expert systems? Or maybe
> something like MycroftAI, running locally?

There is huge knowledge built-in, it can't be compared to our Emacs only
based artificial intelligence.

When offline, lot of knowledge is accessible on fingertips by using
large language models. Since 11 months 8 days I have logged 3623
requests and interactions with the large language models, and created
so much, it greatly helped me enhance my common Emacs experience, and
it is great extension to editor.

Simple run:

$ ./llava-v1.5-7b-q4.llamafile

Louis: What is GNU Emacs?

Llama: GNU Emacs is a powerful text editor that's been around since 1984. It was designed by Richard Stallman and is part of the GNU project which promotes free software development.

What makes it stand out from other editors is its extensibility, thanks to an extensive set of packages or plugins known as "emacs modes." These modes allow you to customize Emacs' features based on your needs, making it a versatile and powerful tool for various programming languages, web browsing, organizing files, emailing, and more.

Another notable feature is the use of key bindings called "keys" in place of traditional menus or shortcuts. This unique approach to navigation allows users to personalize their editing experience further by creating customized mappings for specific tasks within Emacs.

In summary, GNU Emacs is a highly versatile and customizable text editor that has been widely used for decades due to its extensibility, robustness, and the extensive community support it enjoys.

-- 
Jean Louis

Re: Including AI into Emacs

[Bruno Barbier]

Jean Louis <bugs@gnu.support> writes:

> * Bruno Barbier <brubar.cs@gmail.com> [2024-12-06 21:16]:
>> 
>> Jean Louis <bugs@gnu.support> writes:
>> 
>> Are you aware of the gptel package ?
>>    https://github.com/karthink/gptel
>
> I was using in invalidated manner the emacs-chatgpt but in general I
> can do myself Curl and http requests. My usage is totally specific, I
> have memories, various memories that I select from Dynamic Knowledge
> Repository, then I can replace region with the information, or add new
> information into buffer.
>
> I could not try gptel, it doesn't work. But will try in future.

I'm not sure why it doesn't work for you: when I tried it, it just
worked.  As far as I understand, it does exactly what you need: it uses
curl to make asynchronous http requests and to replace or add
information in any buffer, even the minibuffer.


> Soon I will buy Nvidia RTX 4xxx and then I will be using it offline
> exclusively.

gptel probably already has a backend for this too ;)

Best,

Bruno

Re: Including AI into Emacs

[Jean Louis]

* Bruno Barbier <brubar.cs@gmail.com> [2024-12-07 12:32]:
> > I was using in invalidated manner the emacs-chatgpt but in general I
> > can do myself Curl and http requests. My usage is totally specific, I
> > have memories, various memories that I select from Dynamic Knowledge
> > Repository, then I can replace region with the information, or add new
> > information into buffer.
> >
> > I could not try gptel, it doesn't work. But will try in future.

I found out why, tried it out and it works. But programming, inside,
is for me too difficult as I am not used to that type of
programming. I am used for functions which always return something,
but gptel is more than that. Actually I just see too many
complications for my uses.

I need to be able to get result without complications. Synchronous
request is fine and better, rather than asynchronous, where it is
vague what is going to happen with the buffer information.

This is how I prefer it:

(defun rcd-chatgpt-shell (prompt memory model)
  (chatgpt-shell-post :context (list (cons memory nil) (cons prompt nil) ) :version model))

or

(defun rcd-llamafile (prompt memory model)
  "Return answer by using llamafile."
  (let ((model)
	(buffer (let ((url-request-method "POST")
		      (url-request-extra-headers
		       '(("Content-Type" . "application/json")
			 ("Authorization" . "Bearer no-key")))
		      (url-request-data
		       (json-encode
			`((model . "LLaMA_CPP")
			  (messages . [
				       ((role . "system")
					(content . "You are my AI assistant."))
				       ((role . "user") 
				       (content . ,prompt))
				    ])))))
		  (url-retrieve-synchronously
		   "http://localhost:8080/v1/chat/completions"))))
    (when buffer
      (let* ((json-response (setq my-json (rcd-parse-http-json-string (buffer-to-string buffer))))
	     (content (cdr (assoc 'content (cdr (assoc 'message (aref (cdr (assq 'choices my-json)) 0))))))
	     (content (string-replace "</s>" "\n" content)))
	content))))

and 3-4 functions replace all of the catch-all functions in gptel

I have tried understanding gptel code, but I don't. And it doesn't look like I should be going into it.

Yes, it works, but I do not know how to get clear, easy to capture
response, to be returned for easy database logging.

Another issue, I don't see why I need too many files, whole package with many models, not needed.

Like for llamafile, I can easily change the model and same function can work, as above.

-- 
Jean Louis

Re: Including AI into Emacs

[Bruno Barbier]

Jean Louis <bugs@gnu.support> writes:

> * Bruno Barbier <brubar.cs@gmail.com> [2024-12-07 12:32]:
>> > I was using in invalidated manner the emacs-chatgpt but in general I
>> > can do myself Curl and http requests. My usage is totally specific, I
>> > have memories, various memories that I select from Dynamic Knowledge
>> > Repository, then I can replace region with the information, or add new
>> > information into buffer.
>> >
>> > I could not try gptel, it doesn't work. But will try in future.
>
> I found out why, tried it out and it works. But programming, inside,
> is for me too difficult as I am not used to that type of
> programming. I am used for functions which always return something,
> but gptel is more than that. Actually I just see too many
> complications for my uses.
>
> I need to be able to get result without complications. Synchronous
> request is fine and better, rather than asynchronous, where it is
> vague what is going to happen with the buffer information.

gptel works asynchronously indeed (which IMHO is a good thing), but I
understand that it might not fit in your workflow.

> ...
> and 3-4 functions replace all of the catch-all functions in gptel
>
> I have tried understanding gptel code, but I don't. And it doesn't look like I should be going into it.
>
> Yes, it works, but I do not know how to get clear, easy to capture
> response, to be returned for easy database logging.
>
> Another issue, I don't see why I need too many files, whole package with many models, not needed.
>
> Like for llamafile, I can easily change the model and same function
> can work, as above.

IIUC, all you would need is to use the documented API ('gptel-request')
and use its callback.

But, if you don't need integration with org, conversations, multiple
backends, etc. , or just for fun, it makes sense to roll your own
solution.

I just wanted to mention gptel in case you weren't aware of it.

Best regards,
Bruno

Re: Including AI into Emacs

[Jean Louis]

* Bruno Barbier <brubar.cs@gmail.com> [2024-12-07 14:29]:
> IIUC, all you would need is to use the documented API ('gptel-request')
> and use its callback.

I am still using my own function, it works well for me. My workflow is
that I press C-F5 to invoke the prompt, and prompt takes whatever is
in the region and it is added to it. 

It is of course possible to interpolate the response in any text, like
in Org, I can of course enter it, just that I don't use Org too
often. From meta level I may generate Org or just use snippets, I am
using all kinds of markups which can be converted to each other.

But I am curious how do you use AI with integration in Org?

How do you use it in conversations?

I can mark region of what customer said, I can say use 123 memory and
answer to customer, and I get pretty good answer based on my previous
AI memory. Is that what you mean?

Or do you mean some kind of automatic conversation?

> But, if you don't need integration with org, conversations, multiple
> backends, etc. , or just for fun, it makes sense to roll your own
> solution.

Multiple backends I have and it works well.

But which other uses do you have?

-- 
Jean Louis

Re: Including AI into Emacs

[Bruno Barbier]

Jean Louis <bugs@gnu.support> writes:

> * Bruno Barbier <brubar.cs@gmail.com> [2024-12-07 14:29]:
>> IIUC, all you would need is to use the documented API ('gptel-request')
>> and use its callback.
>
> I am still using my own function, it works well for me. My workflow is
> that I press C-F5 to invoke the prompt, and prompt takes whatever is
> in the region and it is added to it. 
>
> It is of course possible to interpolate the response in any text, like
> in Org, I can of course enter it, just that I don't use Org too
> often. From meta level I may generate Org or just use snippets, I am
> using all kinds of markups which can be converted to each other.
>
> But I am curious how do you use AI with integration in Org?
>
> How do you use it in conversations?
>
> I can mark region of what customer said, I can say use 123 memory and
> answer to customer, and I get pretty good answer based on my previous
> AI memory. Is that what you mean?
>
> Or do you mean some kind of automatic conversation?

The package repository has a nice video that shows this package in
action.  I think it will answer your questions better than I could
(I discovered this package recently; I just tried it out of curiosity,
no real needs).

>> But, if you don't need integration with org, conversations, multiple
>> backends, etc. , or just for fun, it makes sense to roll your own
>> solution.
>
> Multiple backends I have and it works well.
>
> But which other uses do you have?

I'm not sure I understand your question.  I hope the video did answer
it for me ;)

Cheers,

Bruno

Re: Including AI into Emacs

[Jean Louis]

[-- Attachment #1: Type: text/plain, Size: 2429 bytes --]

* Bruno Barbier <brubar.cs@gmail.com> [2024-12-10 01:57]:
> The package repository has a nice video that shows this package in
> action.  I think it will answer your questions better than I could
> (I discovered this package recently; I just tried it out of curiosity,
> no real needs).

I have so much real need. It is indispensable, and I can't wait buying
strong GPU to run it locally and fast. It already works locally, just
it is so slow as I don't have GPU. I will upgrade memory, GPU so that
I can run free LLMs locally.

(defun rcd-pie-llm-usage-per-day ()
  "Generate an R PNG image showing LLM usage per day."
  (interactive)
  (rcd-sql-pie-chart "LLM usage per day" 
		     "SELECT extract(day from log_datecreated)::text AS day, 
                             count(log_id) AS count 
                        FROM log 
                    WHERE log_logtypes IN (8, 11, 12, 14)
                      AND log_datecreated > (current_date - 30)
                    GROUP BY day 
                    ORDER BY day ASC"))

The above higher level function uses GNU R Langugage to generate the
attached file `llm-usage-per-day.png'

(defun rcd-llm-usage-by-day ()
  (interactive)
  (cf-chart-bar-quickie "SELECT date_part('day', log_datecreated)::int AS day, 
                                count(log_name) 
                           FROM log 
                          WHERE log_logtypes IN (8, 11, 12, 14)
                       GROUP BY date_part('day', log_datecreated)::int
                       ORDER BY day DESC" 
			"LLM usage by day" 
			"LLM usage by day" 
			"Totals"))

The above higher level function uses Emacs Lisp function
`chart-bar-quickie' to generate the attached file
`LLM-usage-per-day2.png'

I can't agree to asynchronous injection of text, it sounds dangerous
to me. When there is any lag in computer, my text anyway doesn't get
well written due to whatever problems and often my speed writing.

And I am often using temporary buffers for that reason that I do not
disturb my other buffers with whatever changes.

My LLM usage is such that I just do C-u F5 and then text will be added
to the previous text, or I do F5 and text will be inserted, if region
was marked it is used as part of the prompt.

Everything is logged in the database, so I can convert each response
later to other objects.

Basically I get fundamentally same result as gptel with just few functions in background.

-- 
Jean Louis

[-- Attachment #2: LLM-usage-per-day2.png --]
[-- Type: image/png, Size: 30742 bytes --]

[-- Attachment #3: llm-usage-per-day.png --]
[-- Type: image/png, Size: 51068 bytes --]

Re: Including AI into Emacs

[Bruno Barbier]

Jean Louis <bugs@gnu.support> writes:

> * Bruno Barbier <brubar.cs@gmail.com> [2024-12-10 01:57]:
>> The package repository has a nice video that shows this package in
>> action.  I think it will answer your questions better than I could
>> (I discovered this package recently; I just tried it out of curiosity,
>> no real needs).
>
> I have so much real need. It is indispensable, and I can't wait buying
> strong GPU to run it locally and fast. It already works locally, just
> it is so slow as I don't have GPU. I will upgrade memory, GPU so that
> I can run free LLMs locally.

That's why I pick asynchronous calls when possible: I don't have to care
about how fast I will get the result.

> ...
> I can't agree to asynchronous injection of text, it sounds dangerous
> to me. When there is any lag in computer, my text anyway doesn't get
> well written due to whatever problems and often my speed writing.

Emacs does this all the time when running processes asynchronously.

> And I am often using temporary buffers for that reason that I do not
> disturb my other buffers with whatever changes.

Then asynchronous injection should be 100% safe.

> My LLM usage is such that I just do C-u F5 and then text will be added
> to the previous text, or I do F5 and text will be inserted, if region
> was marked it is used as part of the prompt.
>
> Everything is logged in the database, so I can convert each response
> later to other objects.
>
> Basically I get fundamentally same result as gptel with just few
> functions in background.

That's all that matters!

Best regards,

Bruno

Re: Including AI into Emacs

[Basile Starynkevitch]

On Fri, 2024-12-06 at 13:59 -0900, Christopher Howard wrote:
> Jean Louis <bugs@gnu.support> writes:
> 
> > I believe it would be really beneficial to include AI into GNU
> > Emacs. Emacs is already a powerful tool, but adding AI features
> > could
> > make it even better. For example, AI could help with writing by
> > suggesting improvements or catching mistakes. It could also assist
> > with coding by providing smart autocomplete options or debugging
> > help. These features would make Emacs more efficient and
> > user-friendly, helping users get their work done faster and with
> > fewer
> > errors.
> > 
> 
> Could you clarify what you mean exactly but "including AI into GNU
> Emacs"? I know for sure I don't want Emacs sending information about
> my system, or questions that I have about Emacs or my project, off to
> some company's LLM chat system, or however that works exactly. I do
> not want to become dependent on some remote computer program or AI in
> order to be able to write code or figure out how Emacs works.
> 
> I'm certain interested in running tools locally (same computer, or my
> network) that help me with "suggesting improvements, catching
> mistakes", etc. Do you need something massive like ChatGPT to
> accomplish that, or just some Emacs-centric expert systems? Or maybe
> something like MycroftAI, running locally?
> 

Without needing a remote supercomputer, you could run
https://clipsrules.net/ on your Linux desktop (supplying it a rules
source file), or extend https://github.com/RefPerSys/RefPerSys (it is
GPLv3+ work-in-progress inference engine) to run locally on it and
suggest some improvement (or contextual autocompletion) to some EMACS
edited source file.

ChatGPT is certainly not the only possible open source symbolic AI
software, and they don't require a supercomputer or datacenter. For
example GNU prolog is also an open source AI software.

As a concrete example GNU chess is some open source AI program and you
don't need a datacenter to run it.

Very probably, both CLIPSRULES and RefPerSys could be extended (in a
few months of work) for simple tasks like English grammar checking or
English spellchecking.

Regards.
-- 
Basile STARYNKEVITCH           <basile@starynkevitch.net>
8 rue de la Faïencerie
92340 Bourg-la-Reine,          France
http://starynkevitch.net/Basile & https://github.com/bstarynk 

Re: Including AI into Emacs

[Jean Louis]

* Bruno Barbier <brubar.cs@gmail.com> [2024-12-10 13:38]:
> That's why I pick asynchronous calls when possible: I don't have to care
> about how fast I will get the result.

I see, I can understand that, just I can't be writing and expecting
inconclusive results. I am also not in country with so good
Internet. I can understand you are there, and all is good and fine,
but me no. Location where I am is not reliable for Internet.

Too many times I have lost connection, cannot connect, etc. That is because of Internet.

> > I can't agree to asynchronous injection of text, it sounds dangerous
> > to me. When there is any lag in computer, my text anyway doesn't get
> > well written due to whatever problems and often my speed writing.
> 
> Emacs does this all the time when running processes asynchronously.

Running local process asynchronously is fine and I do it myself many
times. But asking anything from Internet servers is quite different.

I could maybe adopt that approach, just I must be on reliable Internet
connection.

-- 
Jean Louis

Re: Including AI into Emacs

[Jean Louis]

* Basile Starynkevitch <basile@starynkevitch.net> [2024-12-10 13:46]:
> Without needing a remote supercomputer, you could run
> https://clipsrules.net/ on your Linux desktop (supplying it a rules
> source file), or extend https://github.com/RefPerSys/RefPerSys (it is
> GPLv3+ work-in-progress inference engine) to run locally on it and
> suggest some improvement (or contextual autocompletion) to some EMACS
> edited source file.

I asked you how, do you have example how you use it? I have clips
installe

Need example.

I have been reading PDF for users on CLIPS, and my impression is that
it is as much complicated as writing rules in Emacs Lisp. I have to
see some benefit. I have large database and I could generate various
rules and let computer figure out things. 

For example, I would like that computer finds out some keywords, and
when keywords in the e-mail quoted message are found, to prepare the
answer or to find out the proper snippet to be used as answer.

I would need just 5 minutes to write rules in Emacs Lisp to find
proper snippet as answer to particular clients, as clients have
pattern of how they ask questions.

Or spam handling, here is some insight:

(defparameter *spam-sets*
  '(("gratitude" "website" "magnificent" "investigation")
    ("identifying" "initiating" "developing" "partnership")
    ("export" "social" "media" "marketing")
    ("Google" "Ads")
    ("Bing" "Ads")
    ("Facebook" "Ads")
    ("Google" "Bing" "Facebook")
    ("PPC")
    ("web" "design" "professional")
    ("fortnite" "bucks")

Then it is handled like this:

    (dolist (set *spam-sets*)
      (when (subsetp set text-list :test #'equalp)
	(setq spam t)))

Right now I have only vague idea that it would be more tiresome to
make it in CLIPS, then in Common Lisp.

Give me examples please.

> ChatGPT is certainly not the only possible open source symbolic AI
> software, and they don't require a supercomputer or datacenter. For
> example GNU prolog is also an open source AI software.

Yes, sure, I agree,

ALL COMPUTER PROGRAMS EMBODY ASPECTS OF ARTIFICIAL INTELLIGENCE:
https://gnu.support/articles/ALL-COMPUTER-PROGRAMS-EMBODY-ASPECTS-OF-ARTIFICIAL-INTELLIGENCE-92631.html

All software is artificial intelligence. 

And LLM or Large Language Models run on local computers, as I said,
soon I will run it that way. It works now, just terribly slow.

> As a concrete example GNU chess is some open source AI program and
> you don't need a datacenter to run it.

Sure. But every game, and every software and Emacs itself is
artificial intelligence. It is extended mind. But now the term AI is
used in marketing to make it easier accessible to common people.

> Very probably, both CLIPSRULES and RefPerSys could be extended (in a
> few months of work) for simple tasks like English grammar checking or
> English spellchecking.

I need examples. How do you use it?

-- 
Jean Louis

---
via emacs-tangents mailing list (https://lists.gnu.org/mailman/listinfo/emacs-tangents)

Re: Including AI into Emacs

[Christopher Howard]

Jean Louis <bugs@gnu.support> writes:

> Sure. But every game, and every software and Emacs itself is
> artificial intelligence. It is extended mind. But now the term AI is
> used in marketing to make it easier accessible to common people.

It seems to me that some important distinctions are being blurred throughout this thread. I am seeing the term AI used to refer to three things:

(1) generally, any kind of computation or problem solving that involves computer programming;
(2) computation that involves inferences and rules (e.g., a prolog program)
(3) using LLM, i.e., "the use of large neural networks for language modeling" (wikipedia definition).

Activities (1) and (2) are things that I can do on my own computer, maybe even without having to leave Elisp or the running, single Emacs thread. For activity (3), even I can do it without the help of remote compute cluster, it is going to require a large model database, plus intense computing resources, like a separate computer, or an expensive GPU requiring proprietary drivers.

I'm open minded to integrations of (3), if they can be done cost-effectively, if they are truly useful, and if I don't have to give up my computing freedoms, but that has to be proven to me. And I don't want that approach confused with (1) and (2).

-- 
Christopher Howard

---
via emacs-tangents mailing list (https://lists.gnu.org/mailman/listinfo/emacs-tangents)

Re: Including AI into Emacs

[Jean Louis]

* Christopher Howard <christopher@librehacker.com> [2024-12-10 20:02]:
> Jean Louis <bugs@gnu.support> writes:
> 
> > Sure. But every game, and every software and Emacs itself is
> > artificial intelligence. It is extended mind. But now the term AI is
> > used in marketing to make it easier accessible to common people.

> It seems to me that some important distinctions are being blurred
> throughout this thread. I am seeing the term AI used to refer to three
> things:

> (1) generally, any kind of computation or problem solving that involves computer programming;
> (2) computation that involves inferences and rules (e.g., a prolog program)
> (3) using LLM, i.e., "the use of large neural networks for language modeling" (wikipedia definition).

You are right, I cannot personally say AI only for LLMs just because that is getting popular.

Like Basile Starynkevitch explained, there are systems like CLIPS and
RefPerSys, Prolog, etc., there are many ways how computer represents
AI. LLMs are not the only AI, that is like degradation of all of the
previous work on which that LLM was based.

LLM represent to me, the enhanced workflow taught to computer to
recognize the needs and provide results

Those human needs we have been recognizing all over place, like in
Emacs or any other software. User is moving the arrow and trying to
shoot the spaceship, but spaceship can see him, react, and fight
against the user. Every game is type of artificial intelligence.

> Activities (1) and (2) are things that I can do on my own computer, maybe even without having to leave Elisp or the running, single Emacs thread. 

That is right, and many such we already all use.

But we do not integrate enough!

Integration, if that is the right work, is enhancing the human workflow to minimize efforts and provide optimum results. That is what I mean.

Programmers are not necessarily scientists, and so they think in terms
of typing. But it is possible to control light with brainwaves, with
special hat, or typing on computer with the eyeball movements.

Makers of LLMs now provided "trained" models  that
can type text, translate text more accurately then common translators. 

> For activity (3), even I can do it without the help of remote
> compute cluster, it is going to require a large model database, plus
> intense computing resources, like a separate computer, or an expensive
> GPU requiring proprietary drivers.

Here is example that works without GPU:
https://github.com/Mozilla-Ocho/llamafile/

and other examples on same page. 

> I'm open minded to integrations of (3), if they can be done
> cost-effectively, if they are truly useful, and if I don't have to
> give up my computing freedoms, but that has to be proven to me. And I
> don't want that approach confused with (1) and (2).

Just as usual, you have got the computing cost, electricity and
computer wearing cost. 

It seems that those files are free software, Apache 2.0 License, but I
did not inspect everything, and some models may be free, some not,
choose what is free.

To not confuse it, we shall simply talk about the LLMs when it is the
subject.

-- 
Jean Louis

---
via emacs-tangents mailing list (https://lists.gnu.org/mailman/listinfo/emacs-tangents)

Re: Including AI into Emacs

[Christopher Howard]

Jean Louis <bugs@gnu.support> writes:

> Integration, if that is the right work, is enhancing the human workflow to minimize efforts and provide optimum results. That is what I mean.

That is not integration, that is optimization or efficiency. Integration may lead to better optimization or efficiency but it might have the opposite effect.

>
> Programmers are not necessarily scientists, and so they think in terms
> of typing. But it is possible to control light with brainwaves, with
> special hat, or typing on computer with the eyeball movements.
>

None of those interface have any appeal to me at all. Well, okay, controlling light with brainwaves sounds interesting, at least. But even so I don't see how the input interface has anything to do with whether or not LLMs (or other AI approaches) should be integrated it our workflow. Unless an input interface is so compute intensive that it requires some kind of cluster-based neural network just to work at all.

> Makers of LLMs now provided "trained" models  that
> can type text, translate text more accurately then common translators. 
>

This sounds like an argument for using LLMs to do language translation, which I suppose must be acknowledged. Regarding prose: I've read the mind-numbing, generic prose output on the Internet that is now being spit out by LLMs, and I hope that goes away. The artwork generated is also terrible, which has been showing up on some of the cheap furnishing products we buy from China.

>> For activity (3), even I can do it without the help of remote
>> compute cluster, it is going to require a large model database, plus
>> intense computing resources, like a separate computer, or an expensive
>> GPU requiring proprietary drivers.
>
> Here is example that works without GPU:
> https://github.com/Mozilla-Ocho/llamafile/
>
> and other examples on same page. 
>

I don't see how a llama driven chat interface or an image generator is going to be useful to me, or worth the computing costs. But I suppose if something like that could be specialized to have expert knowledge of the libraries on my computer or my work flow, it might be worth playing around with.

> Just as usual, you have got the computing cost, electricity and
> computer wearing cost.

My understanding was, for LLMs, the difference involves orders of magnitude. That is what I hear others saying, at least.

Regarding inference engines, I recall with Prolog there is a lot of backtracking going on, so the essence of figuring out a workably efficient program was (1) coming up with intelligent rules, and (2) figuring out when to cut off the backtracking. I have a old Prolog book on my book shelf, but I haven't played around with Prolog at all for years

-- 
Christopher Howard

---
via emacs-tangents mailing list (https://lists.gnu.org/mailman/listinfo/emacs-tangents)

Re: Including AI into Emacs

[Jean Louis]

* Christopher Howard <christopher@librehacker.com> [2024-12-10 21:15]:
> Jean Louis <bugs@gnu.support> writes:
> 
> > Integration, if that is the right work, is enhancing the human
> > workflow to minimize efforts and provide optimum results. That is
> > what I mean.
> 
> That is not integration, that is optimization or
> efficiency. Integration may lead to better optimization or
> efficiency but it might have the opposite effect.

Sure optimization. Though I didn't express me well enough. I mean
connecting human methods of interactions with computer methods.

Integrating means making part of the whole. It is of course
optimization as well.

Examples of integration:

- program monitoring and making statistics of events in the house,
  acting upon events logically; if human turned on lights at specific
  time, lights will be turned on in future automatically; learning
  patterns; acting upon triggers; sensors detecting events such as
  movements, cleaning toilet, mopping the floor when nobody is at
  home, feeding pets, cutting grass, recognizing strangers at the
  gate;

- learning patterns of communication, rejecting nicely patterns not
  relevant, accepting relevant (higher level spam detection);
  answering common questions;

- understanding the agenda, the plan, reviewing automatically what was
  done, what not, making new agenda and plan based on previous one;
  printing it morning early in few copies, making it ready for human;

> > Programmers are not necessarily scientists, and so they think in terms
> > of typing. But it is possible to control light with brainwaves, with
> > special hat, or typing on computer with the eyeball movements.
> 
> None of those interface have any appeal to me at all. Well, okay,
> controlling light with brainwaves sounds interesting, at least. But
> even so I don't see how the input interface has anything to do with
> whether or not LLMs (or other AI approaches) should be integrated it
> our workflow. Unless an input interface is so compute intensive that
> it requires some kind of cluster-based neural network just to work
> at all.

We are already integrating, just it moves slow. The new LLM revolution
is making it possible for common man to create it easier, in much more
easier way than programming, it is higher level programming, the goal
we wanted to achieve back in time.

Look here what people are doing: https://websim.ai/

Computer development now moves inevitably into direction of complex
integrations.

> > Makers of LLMs now provided "trained" models  that
> > can type text, translate text more accurately then common translators. 
> 
> This sounds like an argument for using LLMs to do language
> translation, which I suppose must be acknowledged. Regarding prose:
> I've read the mind-numbing, generic prose output on the Internet
> that is now being spit out by LLMs, and I hope that goes away. The
> artwork generated is also terrible, which has been showing up on
> some of the cheap furnishing products we buy from China.

Yes, it works very well for translations.

- **Spanish:** Sí, funciona muy bien para traducciones.
- **French:** Oui, cela fonctionne très bien pour les traductions.
- **German:** Ja, es funktioniert sehr gut für Übersetzungen.
- **Chinese (Simplified):** 是的，它在翻译方面效果非常好。
- **Arabic:** نعم، إنه يعمل بشكل جيد جدًا للترجمات.
- **Russian:** Да, это очень хорошо работает для переводов.
- **Japanese:** はい、翻訳にとてもよく機能します。
- **Portuguese:** Sim, funciona muito bem para traduções.
- **Hindi:** हां, यह अनुवाद के लिए बहुत अच्छा काम करता है।
- **Italian:** Sì, funziona molto bene per le traduzioni.

Imagine. And I just said: translate in bullet list in 10 different
languages.

You can also say: make Emacs Lisp structure out of it:

Sure! Here is the translation of "Yes, it works very well for translations." in 10 different languages, formatted as an Emacs Lisp structure:

```emacs-lisp
'(("Spanish" . "Sí, funciona muy bien para traducciones.")
  ("French" . "Oui, cela fonctionne très bien pour les traductions.")
  ("German" . "Ja, es funktioniert sehr gut für Übersetzungen.")
  ("Chinese (Simplified)" . "是的，它在翻译方面效果非常好。")
  ("Arabic" . "نعم، إنه يعمل بشكل جيد جدًا للترجمات.")
  ("Russian" . "Да, это очень хорошо работает для переводов.")
  ("Japanese" . "はい、翻訳にとてもよく機能します。")
  ("Portuguese" . "Sim, funciona muito bem para traduções.")
  ("Hindi" . "हां, यह अनुवाद के लिए बहुत अच्छा काम करता है।")
  ("Italian" . "Sì, funziona molto bene per le traduzioni."))
```

Isn't that great? 

I have been wasting hours and hours looking into various structures
and making new Lisp structures to import data from other programs, now
I just insert sample data and get the Emacs Lisp program ready, often
almost ready for production.

> >> For activity (3), even I can do it without the help of remote
> >> compute cluster, it is going to require a large model database, plus
> >> intense computing resources, like a separate computer, or an expensive
> >> GPU requiring proprietary drivers.
> >
> > Here is example that works without GPU:
> > https://github.com/Mozilla-Ocho/llamafile/
> >
> > and other examples on same page. 
> 
> I don't see how a llama driven chat interface or an image generator
> is going to be useful to me, or worth the computing costs. But I
> suppose if something like that could be specialized to have expert
> knowledge of the libraries on my computer or my work flow, it might
> be worth playing around with.

- Website Revision System, sales and marketing:
 
  - Open Graph image related to page can be automatically generated,
    very useful in Internet marketing;

  - Correct titles, make them more appealing, describe the article,
    generate slugs;

  - By using list of links as memory, automatically link words in the
    article with relevant links;

  - Answer customers' questions, point out to articles, products,
    provide support;

- Family:

  - Generate daily routines for children; 

  - Generate planning, fun, entertainment;

- Programming:

  - Create templates, improve CSS, programming code;

  - Quickly find answers, debug; becoming rapid;

There is infinite list of uses.

> > Just as usual, you have got the computing cost, electricity and
> > computer wearing cost.
> 
> My understanding was, for LLMs, the difference involves orders of
> magnitude. That is what I hear others saying, at least.

As I said, there are LLMs working on computer without GPU:

GitHub - Mozilla-Ocho/llamafile: Distribute and run LLMs with a single file.:
https://github.com/Mozilla-Ocho/llamafile

It works on mine i5 CPU, though slow. We will see soon when I insert
Nvidia GPU how it will work. I am very satisfied with results.

It can describe the picture on my computer, which is a fantastic
feature! 📸 I can already envision indexing all my images. It's not
just about my personal life; I also have numerous pictures related to
courses and teaching others.

> Regarding inference engines, I recall with Prolog there is a lot of
> backtracking going on, so the essence of figuring out a workably
> efficient program was (1) coming up with intelligent rules, and (2)
> figuring out when to cut off the backtracking. I have a old Prolog
> book on my book shelf, but I haven't played around with Prolog at
> all for years

SWI-Prolog:
https://www.swi-prolog.org/

-- 
Jean Louis

---
via emacs-tangents mailing list (https://lists.gnu.org/mailman/listinfo/emacs-tangents)