Are forums dying?

[SparkyFX]

they at least did a search first, which will have included AI results

They often have not. Neither the forum search nor websearch. The actual constraints are found out in the conversation. And that was the case before AI, so i don´t see AI as the big culprit here.

[wilfred]

I am sure that's what keeps this place going, but it is only because the moderation allows significantly diverged chat. Often, some unexceptional starting topic will digress into something completely unrelated but nevertheless more interesting than the initial topic. Of course, not always  But those whining about off-topic stuff sometimes, I think, miss the cart for the horse.

The conventional English expression would be "Cannot see the wood for the trees." meaning to lose sight of the bigger picture.  Then there is "putting the cart before the horse"  meaning to do things in the wrong order.

But with regard to off-topic stuff there is interesting off-topic stuff which engages the community and then there is off-topic stuff which does not. Instead it becomes a private bun-fight conducted in full view of everyone and disregards the social norms of polite conduct. And then there are the times when what is said is just plain silly and one feels a strong sense of fremdscham.

Furthermore I am definitely in the court of those who believe some people just want to engage like minded fellows in conversation with their questions. It is utterly irrelevant whether AI or an internet search could have provided an answer. And really anyone who disagrees can keep silent and read the next topic.

[paulca]

I think social interaction is declining in general. More and more couples sitting at a table in front of some cafe, holding their smartphones, heads down, doomscrolling, and not talking to each other.

Yes and if you follow the research you will find it has accelerated since 2020.  Its not just doom scroll addiction either.  Social media was the only platform people had at the start of Lock Down.  It had potential for a while to bring people together for good.  But...  vaccines, masks and political adgenda seen to that.  Polarisation.  Move people to either side of the argument and turn it up.  Leads to people staying on platform to argue, leads to more add impressions.

This has spread to the real physical world.  People don't say, "Hi, how you doing?" to strangers.  People don't stop to talk to strangers anymore.  People expect others to be confrontational or polarised against them because that is what social media taught them.

In the younger generations the simple ability to "behave" in public has been deminishing.

[PlainName]

People don't say, "Hi, how you doing?" to strangers.

Hopefully not! A chap at work says this to me every day, and my automatic response of 'OK' seems both pointless to me and unfriendly to him. I am not going to go into my medical and mental health issues with even my partner every bloody morning, and if the answer is always going to be 'Fine' or equivalent, what's the point of asking? What is wrong with just saying 'Hello' and getting a 'Hi' back?

[RoGeorge]

Here, in Ro, at the countryside (but only inside small villages, where everybody knows everybody else) the custom is to greet everybody you cross by on the country roads of the village, whether you know them or not.  And if you are new/not known, villagers who don't know you will often stop you out of curiosity, and ask you "who are you", as in, who are your relatives, from which family of that village you belong.

The funny thing is, the Ro question is "Al cui ești?", which would translate word-by-word "To whom do you belong to?" 

Though, it is not about you being considered the property of somebody else, the question means "from which family are you?".  Expected response is to name who your parents/grandparents/or some other relatives known to the villagers are.

[PlainName]

The funny thing is, the Ro question is "Al cui ești?", which would translate word-by-word "To whom do you belong to?"

Ooh, that could wind me up something rotten 

But great example of how digression can lead to something interesting and otherwise below one's radar 

[jnk0le]

a form of karma points system, but most people hated it.

[golden_labels]

This has spread to the real physical world.  People don't say, "Hi, how you doing?" to strangers.  People don't stop to talk to strangers anymore.  People expect others to be confrontational or polarised against them because that is what social media taught them.

In the younger generations the simple ability to "behave" in public has been deminishing.

While reading this brain’s response was “but in here (…),” followed with both disagreement and agreement! So here are all the three aspects.

… but in here:
Here’s an ASCII art approximation of a typical city bus of people in Warsaw:

Code: [Select]

  __________________
 /:(:(:(:|:(:|:(| :(\
 `-O------------+-O-'Outside rural populations, in particular older ones, I can’t recall seeing expressions other than :| and :( through my entire life. If it’s not people walking dogs, in cities there is close to zero interaction.⁽¹⁾ No “hello” in common corridors, unless some kid says “good day” that to an old person. No talking in stores or waiting queues, nothing. Just robots (robitniks? ) moving back and forth, from and into their dwelling holes and workplaces. Saying “hello” to a stranger may even feel like trespassing social norms, and a random person talking to you on a tram without a particular goal may be suspicious, in some cases received as aggression. I know it differs by region in Poland, but only a bit compared to the full scale of interactions possible.

So in here I don’t see the “they don’t say ‘hi’ anymore” trend. It’s simply the norm.


… agreement:
I can’t agree more with the 1995–2010 group missing basic skills in social interaction. It’s not completely new and it’s not universal, but compared to older generations, including my own around the same age, the shift is hard to miss. Yes, any non-positive reaction is considered an act of aggression, everything is black-and-white, details and depth don’t exist. Forget about social norms. Not because the group rejects them. That I would not whine about. It rather feels like the mere concept is absent and any rebel-like or naïve-cynical responses are merely a defense mechanism masking it.

I’ll not judge the general behavior. These things change from generation to generation. Even Romans both whined about it, and complained about Romans whining about it. Time flows, we have to accept the culture changes. And that we tend to paint ourselves as much better than we were at their age.

However, on a bit objective side, there are some behaviors and reactions we may have valued, but which are indeed missing. I’ll not say if it’s good or bad. But the change itself may be seen as factual and that has some cause. I’ll not blame social media for that. I’ll blame parents and the society. That is: I’ll blame our generation. These reactions aren’t missing because Facebook or TikTok magically sucked them out youth’s brains. Similar to like playing GTA will never make a person get into a car and driver over pedestrians. Or how playing FPS games doesn’t make one grab a gun and do a school shooting. These reactions and behaviors were never put there in the first place. If young people don’t know how to say “sorry” while bumping into sombody, or that they should help a person in distress, or how to react to common interactions, it’s because we failed to give them the correct examples.


… disagree:
Let me share a few things.

• 10% of Polish youth actively avoids internet,⁽²⁾ and we see anti-smartphone trends in western Europe and North America.
• If I hear any serious, quality conversation about current politics in public transport, it’s 15–25 yos.
• 18–30 yo voters remain strongly anti-establishment, as they have been in past generations.
• I see an increase in activism among youth in here. Ranging from protests, volunteering, to charity. And they fluently use current tools to achieve that.⁽³⁾ That has been in decline since 2010.
• Polish goth community has been shrinking rapidly until just before Covid, and the metal concertgoers were getting old. But around 2018–2019 the trend changed. I do see more and more young people seeking actual human interaction and something beyond scrolling goop content.
• Catholic media report increase in younger people interested the religion. I’m aware these are rather panic attempts to show the church is not dead yet, but in the end these young people are in fact there and seeking spirituality.

⁽¹⁾ I know that to many Polish people that may sound outrageously inaccurate. Calibrate your scale first using other locations and cultures. I didn’t realize something is off until I learned the full spectrum.
⁽²⁾ Krajewski, Kurz, Lewicki, Frąckowiak “Aktywni i (nie)obecni” (NCK, 2026)
⁽³⁾ And regarding effectiveness, let’s just say that this year a YouTube influencer from Poland collected over 292M PLN ($77M, 67M €) in a single live stream, followed by three continuous live fundraisers in a row amassing over 15M PLN ($3.9M, 3.4M €) to sponsor treatment for 3 children. He beaten the biggest charity organisation in Poland, WOŚP.

[Haenk]

Here, in Ro, at the countryside (but only inside small villages, where everybody knows everybody else) the custom is to greet everybody you cross by on the country roads of the village, whether you know them or not.

Same in rural Germany - but, (unless they know each other), that's it. I would find it very odd, if someone would ask me further personal questions, or any other questions besides asking for the way.
Lately I noticed in my medium sized town people started to greet me when just walking around. (No, I'm neither famous nor looking rich...)
Maybe not all hope is lost.

[Berni]

Same here, greeting people you don't know is common in rural areas. But not in a city.

Tho it typically wouldn't form into a conversation. Those only happen when people know each other. That's how you often get scenes of someone standing at a fence while talking to the person who lives there. They passed by and seen the person in the yard, planned to say a few words and continue on where they are going, so they don't actually step in on the yard. But then the conversation ends up taking 1 hour.

But conversations sometimes strike up when people have nothing to do. Like waiting at the hair salon, a stranger might join into a conversation and just keep talking. Again happens in rural areas, and they might ask that "who are you" question (in a similar form as RoGeorge explained). This is not something that would happen in a hair salon in the city.

But i am not all that talkative personally so it is not something i do. Then again being a early millennial likely also has something to do with it.

[Zero999]

I think social interaction is declining in general. More and more couples sitting at a table in front of some cafe, holding their smartphones, heads down, doomscrolling, and not talking to each other.

Yes and if you follow the research you will find it has accelerated since 2020.  Its not just doom scroll addiction either.  Social media was the only platform people had at the start of Lock Down.  It had potential for a while to bring people together for good.  But...  vaccines, masks and political adgenda seen to that.  Polarisation.  Move people to either side of the argument and turn it up.  Leads to people staying on platform to argue, leads to more add impressions.

This has spread to the real physical world.  People don't say, "Hi, how you doing?" to strangers.  People don't stop to talk to strangers anymore.  People expect others to be confrontational or polarised against them because that is what social media taught them.

In the younger generations the simple ability to "behave" in public has been deminishing.

I don't post much on social media, except for YouTube comments,  but I found from family members it did help bring people together during lock down. Yes, there was a lot of discussion surrounding the pandemic response, which I saw as a good thing, because it held the authorities and mainstream media to account. There were the quacks who pushed a lot of nonsense and some were drawn into it, but they were a minority. Unfortunately there was a lot of censorship, including medical experts who asked valid questions, and governments still continue to try to censor things they dislike.

We still see the benefits of social media today, with people circulating footage of the authorities (police, parking wardens, immigration officials etc.) behaving badly, which wouldn't have seen the light of day otherwise. Yet, whenever I read articles or listen to the news, I hear more about the negatives of social media: cyber bullying, extremism, misinformation etc. when more often than not, I see it as a net benefit to society. I know that my smartphone can be used as a tool against abuse from the authorities. If I'm stopped by the police, I can simply start recording them, which I'm well within my rights to.

Forums such as this one are better than social media, because they are moderated by a group of experts and the discussion is separated into clear topics and threads. Social media have used fact checking organisations, but the problem with them is bias and that they're just journalists, rather than experts and later community notes, which don't have the same bias, but are still the general public, not experts. Moderators here are invaluable for focusing on cutting the crap and keeping the discussion on topic.

Can EEVBlog do anything differently to thrive in an AI world and distinguish itself from other technical forums? 

a) Automatically respond to new topics with an AI answer by a designated AI user (marked as such and all) for everyone to read
Pro:

• would streamline both approaches into a thread, allows people to consent or dissent with AI right away
• guarantees an answer
• fixes the problem that many people might ask the same/a similar question, each individually hassling with the AI to squeeze a solution out of it. Would make the forum an even better resource  because you can see other peoples prompts

Con:

• which model?
• Would need to be free
• Might need protections against all kinds of abuse
• the moderators would need to be OK with that (more effort)

I don't know why, but I actually like that idea. Perhaps there should be a delay so the AI only posts if no one else has within 24 hours, to give everyone the chance first. It could also auto-report suspected spam, although keep it to one report per spam account, to avoid overloading the moderators with reports.

The moderators and perhaps a handful of well-respected members, should have the power to edit the AI's posts, delete any hallucinations, and ban it, if it goes rogue.

[Berni]

One of the problems with AI is that they only produce a good answer when given a good prompt.

A lot of the new users here write a very vague post about trying to do something but not providing any specifics needed to properly answer their question. So the next 5 posts is other forum users fishing all of the needed details out of the user.

If you just throw that post at an AI you will get a long monologue about the general topic without any concrete answers in it. That's just a bunch of low value crud filling up the forum.

Perhaps it could work with a specifically forum trained AI model that does moderation work (so to help moderators) but while it is reading all the posts also decides if a thread is good fit for an AI response and sends that off to the big expensive smart AI model to actually answer. Then again if a user wanted an AI answer they could have just typed their post into ChatGPT and got the answer instantly.

Maybe it would be more interesting if the model was trained on the forums own posts, but i doubt anyone is willing to sink time and money into trying to train one just for shits and giggles. Would be amusing to see how long it takes for that AI to get into a heated argument with a user. 

[EEVblog]

Can EEVBlog do anything differently to thrive in an AI world and distinguish itself from other technical forums? 

a) Automatically respond to new topics with an AI answer by a designated AI user (marked as such and all) for everyone to read
Pro:

• would streamline both approaches into a thread, allows people to consent or dissent with AI right away
• guarantees an answer
• fixes the problem that many people might ask the same/a similar question, each individually hassling with the AI to squeeze a solution out of it. Would make the forum an even better resource  because you can see other peoples prompts

In that case why would you need the AI to respond with a post on the forum? Why not just your favorite AI directly? Every AI on the planet is already trained on all this forums content.

I think at one point we talked about having an dedicated AI bot section that just responded to posts in that section, so it was completely isolated. Running locally on the EEVblog servers.
Or like a chat window interface so it wasn't responding with archived posts.
IIRC not many people cared for the idea.

[Zero999]

Can EEVBlog do anything differently to thrive in an AI world and distinguish itself from other technical forums? 

a) Automatically respond to new topics with an AI answer by a designated AI user (marked as such and all) for everyone to read
Pro:

• would streamline both approaches into a thread, allows people to consent or dissent with AI right away
• guarantees an answer
• fixes the problem that many people might ask the same/a similar question, each individually hassling with the AI to squeeze a solution out of it. Would make the forum an even better resource  because you can see other peoples prompts

In that case why would you need the AI to respond with a post on the forum? Why not just your favorite AI directly? Every AI on the planet is already trained on all this forums content.

Because that's not what happens in reality. Too many new members ask simple questions which are easily answered by a search engine, which annoys many regular posters. Offloading this work to an AI would save many peoples' time and patience.

[fzabkar]

On the subject of Discord, where is the benefit in communicating with someone in real time if the most appropriate information is provided by someone in another time zone, as is often (usually?) the case.

[paulca]

The AI auto-responder has "value" I think, but.  Anything "free" or "cheap" will be "rubbish".

It would work if the bot behind it was not a "public facing off the shelf", but a EEVBlog maintained API account and a specific focus on answering the 99% with post explicitly from this forum.

That is not going to be cheap.  Not at the volume of posts it could be answering.

And... If there was budget to spend I would spend it on "prevention" rather than "tonics" and fix the search function!  Even put a bot there!  Its the one forum you don't get people whining "Use the search function before asking" and we all know why.

[paulca]

On ideas for differentiation. 

Collaboration initiatives.  Shared projects.  Shared online workspaces.  Dave has the kudos and credibility to lead these and has led them IIRC. 

Is there legs/routes to turn that up?  Maybe not just Dave driving, but ... embracing the more modern world through the collaboration tooling and actually getting ahead of the game there.

EDIT: You are going to tell me I missed the dozens of shared projects littered throughout the forum, like entire redesigns of existing test gear etc.  So maybe Im thinking about less backroom basement affairs and more "generally fun".

[RoGeorge]

The problem with AI is that the answers are fabricated, they are fantasy, not facts.  The answers are NOT the result of reasoning.

Most people are too gullible to notice that.  They want to believe the AI has agency, and thus they anthropomorphize the AI as if AI were an all-knowledgeable teacher.  Well, it is not that.  The gullible take fantasy/probabilistic words-matching as reasoning, which sadly, it is not reasoning, not at all.

And, to make their bambuzlement even bigger, some AI answers are correct. 

It is very hard to break this illusion in the gullible's mind, particularly because some of the AI answers are coming from a known-good lookup-table of answers.  You have to make them drop the wrong AI answer first (which is rather hard, because the first impression sticks the most, it is hard to unlearn), then you have to teach them how it really is.  Very unproductive for me, and confusing for them.

That is why I don't usually respond to users who quote AI then try to make sense of that AI fantasy.  It's a double effort, and most often they will fail to understand how it really is, simply because their mind was put on the wrong track by AI. 

[SiliconWizard]

Can EEVBlog do anything differently to thrive in an AI world and distinguish itself from other technical forums? 

a) Automatically respond to new topics with an AI answer by a designated AI user (marked as such and all) for everyone to read
Pro:

• would streamline both approaches into a thread, allows people to consent or dissent with AI right away
• guarantees an answer
• fixes the problem that many people might ask the same/a similar question, each individually hassling with the AI to squeeze a solution out of it. Would make the forum an even better resource  because you can see other peoples prompts

In that case why would you need the AI to respond with a post on the forum? Why not just your favorite AI directly? Every AI on the planet is already trained on all this forums content.

I think at one point we talked about having an dedicated AI bot section that just responded to posts in that section, so it was completely isolated. Running locally on the EEVblog servers.
Or like a chat window interface so it wasn't responding with archived posts.
IIRC not many people cared for the idea.

I don't see the point apart from checking a box "we now have AI on EEVblog".

What could be useful is better search on the forum. LLMs are IMO not a fit for that. Yes they have been trained with forums content, but then have digested all this content  as a black box and this is very far from an exact search yielding exact posts. A LLM specifically and only trained on this forum content could maybe work better than using a general LLM for looking up information that has already been discussed, but I'm not sure the work involved would be worth it, and that would require a lot of computing resources.

Probably just a bit smarter search without any LLM would already be good enough, but I think it's a rabbit hole. It's harder than it looks, and again you probably want the search feature to be as lean as possible, so that it doesn't grab all CPU time on the servers.

[Randy222]

One of the problems with AI is that they only produce a good answer when given a good prompt.

AI only has available to it an existing knowledge base, at least the AI that is used for LLM.

Some AI models can derive things (nasty sequences of viruses, as example), but only using the existing ruleset given to it. "This is a jpg", "this is a jpg of a dog", "there are the rules for making bio bugs", etc etc.

When will AI develop on its own some new physics? Likely never. In this realm AI things are only assistants.

When two humanoid bots can on their own clone themselves, then that's when we take notice of that ut-o. I suspect many hundreds of years for this ability. But there does appear to be the initial seed that would allow it, humans provide all the tools up front, like CNC machines, metals, PCB making tools, etc etc, and then the bots use ML/AI models to complete the tasks on their own. Some day yes, anytime soon no.

[SparkyFX]

In that case why would you need the AI to respond with a post on the forum? Why not just your favorite AI directly? Every AI on the planet is already trained on all this forums content.

More or less to get a conversation starting, at the same time providing oversight from the community, kind of getting the best of both worlds instead of just antagonizing AI in comparison to forums. It's just an idea however, i can imagine what effort that would mean.

As a forum reader i would say it would give more input to think about (or think about what might be missing).
It could also mean that people fear of being in competition with AI, so i would just limit it to one answer, not having a replacement interface for AI. Giving it a delay would be a nice touch.

In any case, AI or human, details to a question need to be given or you can't get a good response.

I personally have mixed results from  utilizing AI in trying to work with it. I used Gemini to summarize large amounts of data sheet values (key was to ask for a list in csv-format) but the small context window makes it forget things very fast and e.g. lists suddenly shrink by a few values in the conversation.

The problem with AI is that the answers are fabricated, they are fantasy, not facts.  The answers are NOT the result of reasoning

Yes, some vectors pointed to the most likely combination of words that were close to each other during training. It is not that much different from how humans answer, though.

[abeyer]

Because that's not what happens in reality. Too many new members ask simple questions which are easily answered by a search engine, which annoys many regular posters. Offloading this work to an AI would save many peoples' time and patience.

Would it? Or would those same regular posters who read new member questions and then complain about them just read the autoreplies and complain about those instead? I could absolutely see them devolving into "nitpick the AI response" threads.

[PlainName]

When two humanoid bots can on their own clone themselves, then that's when we take notice of that ut-o. I suspect many hundreds of years for this ability. But there does appear to be the initial seed that would allow it, humans provide all the tools up front, like CNC machines, metals, PCB making tools, etc etc, and then the bots use ML/AI models to complete the tasks on their own. Some day yes, anytime soon no.

That's a fallacy of some kind. Even we, sentient, conscious and allegedly intelligent beings can't make clones even with all the tools developed on the thousands of years. All we can do is fuck and hope some greater being takes pity on us (or, alternatively, hope they don't make it work this time). Being able to procreate isn't a precursor to being conscious (or vice versa).

[RoGeorge]

The problem with AI is that the answers are fabricated, they are fantasy, not facts.  The answers are NOT the result of reasoning

Yes, some vectors pointed to the most likely combination of words that were close to each other during training. It is not that much different from how humans answer, though.

It is not that different from human thinking, indeed, except that AI is lacking the reasoning about that proximity of the words, which is the main difference between being a fantasy vs being an inferred/deduced truth.

[PlainName]

except that AI is lacking the reasoning about that proximity of the words

Got a cat. In the morning I feed him and he sits there watching me. I make my tea, toast, etc., and then wash up the tools, grab my tea and head out to my office. The cat knows exactly when I am going to open the door, despite my going near to it several time (the hand towel is on the back of it). I am fairly certain he has no idea what I am doing, but he somehow knows that this trip to the door is the one that opens it. I can vary my routine - make toast then tea, skip the toast, etc., but it is very rare indeed that he gets it wrong.

I doubt if he is reasoning since he doesn't know what I am doing. It is purely statistical association as far as I know. Is my cat not conscious?