Has EAB blocked ChatGPT?

[kriz]

Why do people want to chat with a bot ? Give details and thats it, i see people greet the bot and all sorts of weird communicating is going on, i don`t get it.. Get the facts then done.

[pixie]

Originally Posted by kriz

Why do people want to chat with a bot ? Give details and thats it, i see people greet the bot and all sorts of weird communicating is going on, i don`t get it.. Get the facts then done.

I guess when the time arrives AGI will know how to differentiate those who treated it with manners from those who doesn't! Lol who cares how people interact with bots...

[malko]

Too much fantasy answers.
Those bots should better say "I don't know" instead of "blablating" around as they do.
Example : https://eab.abime.net/showthread.php...47#post1684647

[dreadnought]

Did you read the disclaimer?

"ChatGPT can make mistakes. Check important info"

[malko]

Making mistakes is a thing that is not related with the provided answers.
Personally I would better give my confidence in the answers of someone that is able to say "I don't know " until that someone gain the required knowledge to give an appropriate answer, being it approximate (by telling it).

[Nightfox]

I recommend talking to ChatGPT in voice mode. It can at times feel like you're talking to a human

[jman]

Originally Posted by RCK

I updated robots.txt, ChatGPT should be able to come to EAB again

I regret this decision because all these LLM are basically doing is exploiting user-generated content without giving proper credit or citing sources in a reliable way (among other concerns about them which here are not relevant).

I am in favor of reinstating the block (see also this post from me).

I mean, I am supporting EAB with a monthly donation but if my support goes into the guts of OpenAI and friends, I can well interrupt the donation.

[malko]

^ The problem is that there is less and less space for a minded choice :
https://www-20minutes-fr.translate.g..._x_tr_pto=wapp

In CH, even the public school creates cloud account (MS) for kids and part of the schoolwork or evaluations are provided through this channel. Do you really believe that kids are not evaluated by this IA ? Given all the past scandals, I don't.
https://www-microsoft-com.translate...._x_tr_pto=wapp
And of course, when parents kindly ask for a copy of the agreement/contract to see if data privacy is respected, they get no answers...
It is really difficult to see the good that can emerge from the "freeness" of these applications when simple questions/concerns are just ignored by people only wanting, without proactive controls, to push their interests.

[Nightfox]

Originally Posted by jman

I regret this decision because all these LLM are basically doing is exploiting user-generated content without giving proper credit or citing sources in a reliable way (among other concerns about them which here are not relevant).

I am in favor of reinstating the block (see also this post from me).

I mean, I am supporting EAB with a monthly donation but if my support goes into the guts of OpenAI and friends, I can well interrupt the donation.

ChatGPT when web searching does cite sources though. it'll link directly to the forum post

[TCD]

Originally Posted by Nightfox

I recommend talking to ChatGPT in voice mode. It can at times feel like you're talking to a human

I kinda feel sorry for you.

[Nightfox]

Originally Posted by TCD

I kinda feel sorry for you.

Just like people feel sorry for me that I still use an Amiga

[malko]

Originally Posted by Nightfox

Just like people feel sorry for me that I still use an Amiga

[ Show youtube player ]

[RCK]

Originally Posted by jman

I regret this decision because all these LLM are basically doing is exploiting user-generated content without giving proper credit or citing sources in a reliable way (among other concerns about them which here are not relevant).

ClaudeBot (AI) tried to sniff all EAB and was responsible of the latest big slowdown last month, so I rate limited it to 1 request per minut.
GPTBot (AI) is currently limited to 2 request per seconds, and doesn't seems happy with that, and I don't plan to open more EAB to thoses bots, so no worries

[jman]

Originally Posted by RCK

ClaudeBot (AI) tried to sniff all EAB and was responsible of the latest big slowdown last month, so I rate limited it to 1 request per minut.
GPTBot (AI) is currently limited to 2 request per seconds, and doesn't seems happy with that, and I don't plan to open more EAB to thoses bots, so no worries

hey Rick, thanks for throttling the scraping on EAB, so at least we legit users can keep on enjoying its content without disruption.

However, this does not address the point of allowing LLMs scraping our content which, in my opinion, is not the same of allowing a search engine.
Of course feel free to determine what you believe is better for EAB (I will do that with my finances as well).

[pixie]

Originally Posted by jman

hey Rick, thanks for throttling the scraping on EAB, so at least we legit users can keep on enjoying its content without disruption.

However, this does not address the point of allowing LLMs scraping our content which, in my opinion, is not the same of allowing a search engine.
Of course feel free to determine what you believe is better for EAB (I will do that with my finances as well).

What do you have to say regarding

Originally Posted by Nightfox

ChatGPT when web searching does cite sources though. it'll link directly to the forum post

I do not use the full tier of ChatGPT, but by your wordings I guess either you or Nightfox have the wrong impression on how ChatGPT present it's sources or lack thereof.

[jman]

Originally Posted by pixie

I do not use the full tier of ChatGPT, but by your wordings I guess either you or Nightfox have the wrong impression on how ChatGPT present it's sources or lack thereof.

I sporadically use LLMs so I may be wrong w.r.t. ChatGPT specifically citing sources from EAB. I've tried other LLMs in context of programming-related domains and received varying quality responses.

Set aside all that, which is not greatly relevant, my point is about not allowing LLMs to scrape user-generated content on EAB.

[dreadnought]

Originally Posted by jman

Set aside all that, which is not greatly relevant, my point is about not allowing LLMs to scrape user-generated content on EAB.

For me, the only reason this would make sense is if the scraping interferes with the normal functioning of the board (which, I presume, is something RCK means above).

Other than that I'd struggle to see any justification of keeping the awesome knowledge contained on eab exclusive to a selected group of people.And like it or not, in a few years LLMs will be the prime way people search for information.

[pixie]

Originally Posted by jman

I sporadically use LLMs so I may be wrong w.r.t. ChatGPT specifically citing sources from EAB. I've tried other LLMs in context of programming-related domains and received varying quality responses.

Set aside all that, which is not greatly relevant, my point is about not allowing LLMs to scrape user-generated content on EAB.

I thought your point was "exploiting user-generated content without giving proper credit or citing sources in a reliable way". While not being directly comparable, the day I couldn't use google to search eab has hampered a lot the way I search eab for info.

[commodorejohn]

"In a few years" they'll be gone entirely after the bubble bursts

[dreadnought]

Originally Posted by commodorejohn

"In a few years" they'll be gone entirely after the bubble bursts

I'd like to say "ok, let's chat again in a few years" but I'm pretty sure we can do it sooner than that

In fact I'd say it should be pretty obvious it won't be another dotcom or whatever right now, but I get it that some folks like to live on the controvershial edge.