Should ChatGPT bots be banned from the EEVblog Forum?

[wilfred]

How about the dyslexic Devil worshipper?
He sold his soul to Santa.

Another classic, never gets old.

[EEVblog]

This cancer is now apparently being pushed from many new accounts that show the Chinese flag. Let's just report those when it becomes obvious they are ChatGPT frontends.

If you have a list of accounts like this please provide it.

[jpanhalt]

Banned or not, bots should be identified as trollbots or similar so members don't mistakenly respond to them.

[Nominal Animal]

A much more useful expert system would be one that takes Rally English or translated text, and refines the wording to be simple, clear, and precise.  It could even ask the user clarifying questions (when the original text is ambiquous or misses a point).

In other words, a proper language model that takes the information stated in the input (and user-supplied answers to clarifying questions), and formulates a clear sequence of statements based on that.

[MK14]

I'm just thinking out loud at this point, but ...

If this bot-scripted-ChatGPT accounts issue, got really badly out of hand, and was causing major issues.  Especially with possible future ones (ChatGPT v5.0 v6.0 ...), which could be significantly harder to detect.

Forum(s), could offer something on the lines of a user-ID verification process.  In effect, to confirm they are human.  Their real life (personal) details, would still be secret from other users.  But they would have been verified.

I'm not suggesting we do that here, certainly not now, at least.  But just trying to suggest, that there are possibly viable solutions, if it gets really bad.

[tggzzz]

Forum(s), could offer something on the lines of a user-ID verification process.  In effect, to confirm they are human.  Their real life (personal) details, would still be secret from other users.  But they would have been verified.

That would tie in with the UK (and other governments) desire for "digital identities". It would presumably work as well as personal digital certificates[1], and use the same mechanisms as porn sites will have to use (?credit cards?).

Or not, as the case may be

[1] I had a late colleague who, at the turn of the millennium when the government first proposed such things, pointed out to them that "My name is Donald Duck and I have a certificate to prove it".

[Neutrion]

I'm just thinking out loud at this point, but ...

If this bot-scripted-ChatGPT accounts issue, got really badly out of hand, and was causing major issues.  Especially with possible future ones (ChatGPT v5.0 v6.0 ...), which could be significantly harder to detect.

Forum(s), could offer something on the lines of a user-ID verification process.  In effect, to confirm they are human.  Their real life (personal) details, would still be secret from other users.  But they would have been verified.

I'm not suggesting we do that here, certainly not now, at least.  But just trying to suggest, that there are possibly viable solutions, if it gets really bad.

And than the anonimity of the internet would be over, the Big Brother system would start to kick in.
Or no human interactions on the net, without ID. (The huge problem with the UK proposals as well.)
And if the certs get stolen? But even if in theory it would be an unhackable system which we all know exist until the practical implementation internationally gets through politics and regulation...

[pcprogrammer]

And than the anonimity of the internet would be over, the Big Brother system would start to kick in.

As if the internet is really anonymous. Your IP address is know, and if needed governments will force internet service providers to fork over the information about you. Social media sites get your information when you enroll. If needed they will be forced to fork over the information. Everything you do online is monitored as soon as you allow cookies and use search engines, etc.

Of course you can use private browser windows or vpn's, but if they want they will find you. Anonymity is an illusion.

[PlainName]

if needed governments will force internet service providers to

Nowt you can do about that, but I'm buggered if I'm going to allow some random internet perve the same access.

I guess the government know all your banking details and payments. You think that should be so easily available to Dave? Or whoever hacks into Dave's user database and releases it on the Dark Web? No, the one at least has to pretend to go through the motions of getting a warrant and all that. Let Rando Webo get their own warrant if they want my personal identifying info.

[Neutrion]

And than the anonimity of the internet would be over, the Big Brother system would start to kick in.

As if the internet is really anonymous. Your IP address is know, and if needed governments will force internet service providers to fork over the information about you. Social media sites get your information when you enroll. If needed they will be forced to fork over the information. Everything you do online is monitored as soon as you allow cookies and use search engines, etc.

Of course you can use private browser windows or vpn's, but if they want they will find you. Anonymity is an illusion.

Just because there is no EASY anonimity on the internet, and there are ways for governments etc. to find things out, it does not mean that it makes sense to give up the rest of the anonimity, and make it even easier to end up in a surveilance state.
I would rather push for even more easy anonimity, if it is about regulation.
Its like saying: "It is already quiet bad, why shouldn't we make it worse?"
There are historic lessons from not long ago about what happened when the state had the power to control everything.

But I would rather go the other way: Why the reciever side (us) should try to protect itself and give up basic freedoms just to let some companies make money?
Why not these companies sould held responsible for the consequences of their handling?
There is a shitload of regulations for every aspect of life to mitigate dangers, but these companies are alloved to mess around without consequences?

So I would rather unleash the total control on them, instead of everyday people.

And no it wil be not easy to just escape to some third world countries. But in worst case I would rather ban traffic from there rather than give up everything here.

I really wait for the day when politics start to take these issues at least as serious as some ridiculos gender and woke stuff. It is really time to have a chat with the local politicians and to make sure they got a bit more information about these issues than the marketing BS.
If people who are not able to decide wheter they are man or woman were able to do do it I don't see why some smart people couldn't make as much noise

[thm_w]

If a reputable user wants to include the occasional ChatGPT response, I see no issue with that.

I do. If you have nothing to say, why post? If people are asking in a forum, it's not to find answers they would have found using a search engine or chat bot available to anyone.
Why would you do that? That's absolutely beyond me. That's the very opposite of what a forum is for.

A large portion of beginner questions and some advanced can be answered with a simple google search.
Many people ask either due to: laziness, inability to search, not knowing what key words to use, just wanting confirmation, etc.

Answering with a link or the correct chatgpt response is not "saying nothing".

[Infraviolet]

Does the "one official bot" option mean a specific forum section where a bot interacts with otherwise human posts? One where peole can jointly "ask the bot" and laugh at how useless its results actually are when read carefully despite looking convincing at a glance.

[tggzzz]

I'm just thinking out loud at this point, but ...

If this bot-scripted-ChatGPT accounts issue, got really badly out of hand, and was causing major issues.  Especially with possible future ones (ChatGPT v5.0 v6.0 ...), which could be significantly harder to detect.

Forum(s), could offer something on the lines of a user-ID verification process.  In effect, to confirm they are human.  Their real life (personal) details, would still be secret from other users.  But they would have been verified.

I'm not suggesting we do that here, certainly not now, at least.  But just trying to suggest, that there are possibly viable solutions, if it gets really bad.

And than the anonimity of the internet would be over, the Big Brother system would start to kick in.
Or no human interactions on the net, without ID. (The huge problem with the UK proposals as well.)
And if the certs get stolen? But even if in theory it would be an unhackable system which we all know exist until the practical implementation internationally gets through politics and regulation...

If the certificates are compromised then you revoke the certificate.

If the certificate is biometric, then simply get another retina/iris/etc.

[Neutrion]

I'm just thinking out loud at this point, but ...

If this bot-scripted-ChatGPT accounts issue, got really badly out of hand, and was causing major issues.  Especially with possible future ones (ChatGPT v5.0 v6.0 ...), which could be significantly harder to detect.

Forum(s), could offer something on the lines of a user-ID verification process.  In effect, to confirm they are human.  Their real life (personal) details, would still be secret from other users.  But they would have been verified.

I'm not suggesting we do that here, certainly not now, at least.  But just trying to suggest, that there are possibly viable solutions, if it gets really bad.

And than the anonimity of the internet would be over, the Big Brother system would start to kick in.
Or no human interactions on the net, without ID. (The huge problem with the UK proposals as well.)
And if the certs get stolen? But even if in theory it would be an unhackable system which we all know exist until the practical implementation internationally gets through politics and regulation...

If the certificates are compromised then you revoke the certificate.

If the certificate is biometric, then simply get another retina/iris/etc.

Revoking cert.: And than going through the process of clearing what was made by the real me and what by others. And have to make sure that the revoking is not made  by a third party, who is not me.

But at least getting  new eyes, fingers, and face is an easy one.

[Bud]

Explain the concept or meaning of "stolen biometric certificate" (without talking nonsense).

[SiliconWizard]

If the certificate is biometric, then simply get another retina/iris/etc.

Yes, no problem.

[tggzzz]

Explain the concept or meaning of "stolen biometric certificate" (without talking nonsense).

Compromised is sufficient; stolen is not necessary.

Compromised is anything that renders a certificate useless. FFI, read com.risks.

[PlainName]

Just because there is no EASY anonimity on the internet...

First couple of reads I saw "animosity" 

[SiliconWizard]

"AI"-detection systems will very likely flourish in the next years. And then "AI" companies will find ways to circumvent detection systems. And then, ... and then.
That will be exactly as with the spam and spam filters, just possibly orders of magnitude more complex and more expensive to handle.
(Speaking of spam, I've noticed an alarming rebound in spam lately that manages to pass through even the most aggressive filters.)

And that will be that much time and money that we can't invest in something more productive.

And, we will eventually be flagged as far-right and discriminatory if we ever attempt to discuss that. 

[tggzzz]

"AI"-detection systems will very likely flourish in the next years. And then "AI" companies will find ways to circumvent detection systems. And then, ... and then.
That will be exactly as with the spam and spam filters, just possibly orders of magnitude more complex and more expensive to handle.
(Speaking of spam, I've noticed an alarming rebound in spam lately that manages to pass through even the most aggressive filters.)

And that will be that much time and money that we can't invest in something more productive.

And, we will eventually be flagged as far-right and discriminatory if we ever attempt to discuss that. 

Nah. Far left pinko virtue signallers.

Wadda you mean there's no difference between far left and far rigbt? If that is true then both Stalin and Goebbels were right!

[rstofer]

The ChatGPT responses I have seen have two characteristics not found in ordinary user responses:  There are no grammatical errors and there are no spelling errors.  It should be easy to create an NLP to look for these features.

[xrunner]

The ChatGPT responses I have seen have two characteristics not found in ordinary user responses:  There are no grammatical errors and there are no spelling errors.  It should be easy to create an NLP to look for these features.

Also, many ChatGPT responses have a closing summary that begins like these -

It is worth noting that ...

Ultimately, the decision between ...

Overall these points ...

Among other similar phrases.

[Halcyon]

If a reputable user wants to include the occasional ChatGPT response, I see no issue with that.

I do. If you have nothing to say, why post? If people are asking in a forum, it's not to find answers they would have found using a search engine or chat bot available to anyone.
Why would you do that? That's absolutely beyond me. That's the very opposite of what a forum is for.

Not only that, but if a "reputable user" does that, it just conveys the idea that the user they are replying to is too stupid to do that themselves. It's wrong on all levels.

We shouldn't be controlling what people say. If someone just wants to dribble words, I say let them. They aren't harming anyone. Some people just need to be heard and feel like they have a voice. Others get a feeling of satisfaction by contributing.

Yes you're right, we can all find answers if we look hard and long enough but a lot of times, it's not about that. I often ask questions here which I'm sure with enough research, I could find what I wanted, but why not reach out to a whole bunch of experts and have a discussion?

AI like ChatGPT is just another tool, and I encourage its use, just not by user accounts where that is the only source of content.

[tggzzz]

If a reputable user wants to include the occasional ChatGPT response, I see no issue with that.

I do. If you have nothing to say, why post? If people are asking in a forum, it's not to find answers they would have found using a search engine or chat bot available to anyone.
Why would you do that? That's absolutely beyond me. That's the very opposite of what a forum is for.

Not only that, but if a "reputable user" does that, it just conveys the idea that the user they are replying to is too stupid to do that themselves. It's wrong on all levels.

We shouldn't be controlling what people say. If someone just wants to dribble words, I say let them. They aren't harming anyone.

Not quite true.

Dribble sucks up some of my remaining life. While I've become good at quickly determining what to ignore, less experienced people may not have the skill/ability to do that.

When I spot particularly dangerous (in a strict or loose interpretation of that word) then I can feel obligated to correct the falsehoods.

Some people just need to be heard and feel like they have a voice. Others get a feeling of satisfaction by contributing.

Yes you're right, we can all find answers if we look hard and long enough but a lot of times, it's not about that. I often ask questions here which I'm sure with enough research, I could find what I wanted, but why not reach out to a whole bunch of experts and have a discussion?

AI like ChatGPT is just another tool, and I encourage its use, just not by user accounts where that is the only source of content.

I don't object to clearly marked chatbot output, for the same reasons I think it reasonable to tell people "google is your friend".

Unmarked chatbot output is an abomination that should be strongly deprecated.

[danbr]

Could be useful for giving reply's to common questions, or perhaps as an "enhanced" forum search