I needed to publish my opinion on why GPT / AI / ML / NN bans are usually not enforceable by the administration , or they need to be blanket utilized to all merchandise .
The thread was deleted , happily resulting from a bug of the mql5.com profiles i can return in time and gauge what occurred roughly.
A prelude :
What was within the thread ?
The OP (poster) had a few of his merchandise hidden as a result of somebody complained or he spammed them and there appears to be an anti spam bot that sends your product to be checked on the desk should you spam it . Fairly intelligent mechanism i prefer it , bravo mql5.com , step in the best course there.
The OP was utilizing gpt to reply questions , his solutions have been obscure , and he was insisting rather a lot on the flawed factors (from what i had time to learn) Moreover in his blogs i found customers complaining that a few of his merchandise didn’t function (so he was in violation of a number of guidelines together with the hype AND official exhausting coded guidelines) for among the merchandise .
Nonetheless the dialogue was on GPT . Now was he wrongly insisting he was proper ? In all probability from what i’ve learn so far .
May the moderators have eliminated -as i recommended in a single publish in there- the fuss the op did and maintain the legitimate af factors a few of us made in regards to the GPT hype rule ? Sure however maybe they have been drained and irritated and paired with the OPs timezone they might both lose sleep or delete the thread .
Anyway . I am going to blur the names and avatars of the mods posts clearly .
On to the opinion :
In the beginning of the thread the op claimed the ban was for gpt within the titles (and outline and logos) so the moderator replied with a reference to a publish by the Metaquotes moderator account from the Russian discussion board that mentioned this :
So there’s hype talked about there , which means -probably- that the product was flagged for one thing and the test additionally caught the factitious intelligence + gpt phrases within the descriptions .
right here is the unique hyperlink https://www.mql5.com/ru/discussion board/451234/page2#comment_48339526
So the administration appears to not need AI/ML/GPT/NN for use as “hype” to enhance the perceived efficiency of the product .
(the place is the factitious intelligence hype talked about ? afterward that thread the desk informs the op he used “synthetic intelligence” in his description)
right here is the hyperlink : https://www.mql5.com/ru/discussion board/451234/page3#comment_48343897
However these can’t be arbitrarily enforced .
Right here is why :
Let’s begin with Neural networks.
For example Mark sweats his hours off and creates a neural internet . Locations the neural internet within the ea to commerce. Contains neural internet within the title and descriptions based mostly on what he did . Then i waltz out there , i add neural internet , ai , deep studying within the title or description however my ea does none of that. We add our eas . The one method for mq to know certainly one of them is an actual neural internet ea is to take a look at the supply code . Now if Mark says maintain on , my ea requires loading this onnx mannequin then mq nonetheless can not know if that mannequin is behind the wheel of the buying and selling choices with out wanting on the supply code . Mq cannot take a look at the supply code so there isn’t a foundation for banning Mark’s or my EA . They must disguise not one of the 2 EAs or each of the EAs.
Except they appeared on the supply code.
Let’s go to chatGPT / GPT / LLM
Mark could be very sensible and decides to deploy an LLM mannequin he forked from meta (fb). He buys a server and a website identify and has the mannequin within the server or on a separate laptop with particular replace intervals the place it sends new information to the server . So , what’s going to Mark’s ea do on this case ? it’s going to ship alongside an emblem and the server will reply with the latest replace it acquired from the LLM mannequin from Mark’s separate PC . Similar if the mannequin is within the server , the EA will ship an emblem and the server will reply with its newest replace for that image , or , if the ea can wait , it’s going to crunch it and compute it proper then and there (not environment friendly nevertheless) If Mark need’s to make use of chatGPT he might want to facilitate communication with openAIs API ,with the token they’ve given him (which he has purchased and actively pays per use) . Now for the reason that fee is per question and per question measurement , and “avatar” measurement if im not mistaken ,Mark can not danger the queries going on to his server (from the EA). So , he’ll both have a separate PC which does the backwards and forwards with the openAI API and sends or shops the responses at replace intervals on the server . Okay ? One other method might be to make use of an onnx mannequin (the smallest is 900 megabytes , greatest 14gigabytes) and have his shoppers load it . Which means we would additionally see necessities for large quantities of RAM for the EA to function on MT5 and nearly inconceivable on MT4 . Now i come alongside , i purchase a server i purchase a website identify and i’ve my ea talk with it . Each Mark and that i publish our eas , each Mark and that i add AI , GPT ,Synthetic Intelligence and all that lingo in title and the overviews . There isn’t any method for mq to inform Mark’s or my EA is pretend . They would want to see the supply code , the backend of the server , if there was a 3rd PC managing the replace intervals they’d must see that too . If Marks EA operated with onnx fashions they might not know if the mannequin was wired within the buying and selling choices -again- with out wanting on the supply code.
So there isn’t a foundation for hiding Mark’s or my EA in that case both .
Once more its non-enforceable or it should be blanket enforced.
Once more , they have to disguise not one of the 2 EAs or each of the EAs.
There’s one blind spot which i can not test due to the laptop being weak right here .
If Mark compiles a mannequin inside the EA (which means opening up a supply file of minimal 920MB or 14GB in measurement which my laptop cannot do right here) and if that may be compiled , i can not test what the dimensions of the ex4/ex5 might be .
However nonetheless even when that was evident by the bloated ex4/ex5 (if it didn’t go over the add file measurement restrict!) there wouldn’t be a option to know the mannequin is wired within the buying and selling choices with out wanting on the supply code.
So for all instances the rule can’t be utilized arbitrarily even whether it is as a result of the spam bot caught a product (for different causes) and when it was checked it was discovered to be in violation of the hype rule .
So , it should be utilized to ALL merchandise that use this lingo , or , NO merchandise in any respect . That’s what my protest is about .
Sure they is probably not checking all merchandise however the rule can’t be utilized arbitrarily if a product occurs to be checked .
They personal their database they’ll comb via it with scripts and mechanically disguise violations . In the event that they actually needed to go after hype.
“However they are saying they do not test merchandise anyway…” yeah , however resulting from this rule being inconceivable to test ,as a result of from the second one of many search phrases seems on a product its 50/50 if its true of not , it should be blanket utilized universally + immediately OR not utilized in any respect.
That’s the reason the present strategy isn’t viable and unfair .
If mql5.com needed to be honest -again as a result of this can be a non-rule , cannot be checked , its obscure anyway , they must :
Not apply this rule to anybody . Apply it to everybody mechanically however present these affected with the means to change their titles . Or change the market solely the place distributors should add supply code as a substitute of ex information.
This publish isn’t about “how unethical the OP of the thread was for together with gpt in his overviews” , it goes with out saying anybody who does this
(together with a number of unhidden merchandise on the entrance web page proper now) is unethical . I am not arguing as per the moral spectrum of that particular case.
I will not publish this within the discussion board as a result of it is going to be deleted .
So based mostly on the above , if mq desires to implement it these merchandise are in violation of the hype rule :
Google search this “web site:mql5.com/en/market synthetic intelligence” minus the quotes Google search this “web site:mql5.com/en/market gpt” minus the quotes
Some have been flagged already however guide checks are meaningless -and a mockery to all the vendors-
Edit : Evidently that if a product is “checked” by the desk manually for whether or not or not it communicates with chat GPT , which means the EA should contact a url to obtain buying and selling directions, it’s in violation of the market rule 6 as that is a further limitation . In order that they move the test for gpt and ingore rule6 in that case….
When you agree share in your profile :