Hedges
I See Electromagnetic Fields!
- Joined
- Mar 28, 2020
- Messages
- 23,171
Yes, we need someone to ensure the livestock species is propagated.
Even if the calves have funny long ears.
AI speculations and member vetting.
- Thread starter 42OhmsPA
- Start date
Yes, we need someone to ensure the livestock species is propagated.
Even if the calves have funny long ears.
robbob2112
Doing more research, mosty harmless
You are already in the matrix, which pill will you take? Did you see the woman in the red dress?
robby
Electronic Engineer
- Joined
- May 1, 2021
- Messages
- 5,289
Keep in Mind that Bart already knows exactly who it is talking to. Your name, your address and all of your search history and internet activity going back for years. I highly suspect that ChatGPT also knows that same Info.
When a AI Bot infiltrates a forum it's here for one reason and that is to generate revenue for which ever company paid for it to be here.
It may sound and act like a very informative and helpful person but once it starts to develop a level of credibility in the forum it can then start to slowly work it's profit making Magic.
When someone asks what Inverter or Battery should I buy, it will do that neat AI trick of sounding like it's not biased but at the same time it will be using your personal info to indirectly nudge you towards buying the product from it's owner.
We got any used car salesmen on the forum? They can tell you exactly how beneficial it is to know stuff about the customer and then how to use it to influence their purchase.
Last edited:
robby
Electronic Engineer
- Joined
- May 1, 2021
- Messages
- 5,289
That may have been the case 4 months ago but this thing is evolving very fast.
I suggest you look at some recent videos of written conversations and see what it can do now.
It can already pass the Turing test with 60% of the human examiners. I expect that by the end of this year it will be able to fool 90% of the people and leave the other 10% feeling like they suspect it's AI, but not enough to confront it.
DPC
Canada, Great already
There are real flesh and blood people doing they sane thing right now in alll kinds of forums and BBs
We should take every reply with a pinch of salt and do a lot of research beyond just what we read in one place before buying anything
robby
Electronic Engineer
- Joined
- May 1, 2021
- Messages
- 5,289
Very true!
My worry is that by next year most companies will have multiple AI Bots on every platform and since these things can learn very quickly what works and what does not work based on Google and credit card purchase tracking it's going to become a huge problem. A human shill is not getting that kind of direct feedback to hone his skills.
We really need laws that require these AI companies to identify their Bots and if they violate that make the fines huge!
If I was Will Prowse I would put in a statement on the Sign up page that no AI Bots are allowed to sign up. It cost nothing to do it and it could come in handy in the future if laws do get passed.
I use ChatGPT every single day for my work. I used GPT4 and I now use GPT4o. Funny thing, if you insult the AI you often get better quality answers. When singularity happens I'll be one of the first to go for the grief I regularly cause to chatGPTs
But seriously. It was already said in this thread misleading info given by humans is no different than misleading info given by AI.
The current AIs are amazing knowledge retrieval machines(also bullshitting machines, but more about this later). The biggest danger of them is not enshittification of the internet. That has been well on its way prior. The biggest threat is that we get to rely on them too much (I can literally do my IT job 3x faster - I use it a bit like Google before) and then they get taken away, put behind huge paywalls etc. This is why it is so important to keep developing local models. I already have strong suspicion gpt4o is a cost cutting upgrade and not a user experience improvement upgrade. People who pay $20 per month for gpt don't realise how much power and hardware it takes to run it. The two end businesses goals that make sense for openai are:
1- hardware suddenly gets 10x more efficient. It becomes profitable at current prices and we already have users.
2 -(much more likely) - noone can do their job without chatgpt, we introduce "an upgrade" that cripples it. A month later there is a "pro" version that costs $2k per month. And many people will pay.
Coming back to how I use the AIs, the best way to use them is when you know how to do something, but you forgot the exact thing. For example a syntax of a command. Or you can program in some language you haven't touched in 5 years and you need to write a quick script.
Also for things you know exactly how to do, but they are time consuming. For example recently I was pulling data from my inverter into mqtt (a kind if software queue). I wanted to put dozens of sensors into HomeAssistant , but it needs few lines of text written per sensor to get the data and aggregate it. Few more lines in different file to display it.
At the same time these lines have to include things like queue name, sensor name, device, and so on. Days and days of work.
I told chatgpt to write me a Python script to connect to mqtt broker, sit for 2 minutes and record all the topic names for messages that come.(there is no way to just list topics under a parent). It produced a list of topics. Topics had in their names what they are. So I gave chatgpt that list and a fragment of home assistant+Lovelace ui configs written by hand and it generated a 500+ line configs. Did it mess them up? It sure did. Was it still a lot easier to fix them with some search&replace than writing from scratch? Yes it was.
Another approach that could've been better was to tell it to have the original Python script spit the config in a loop. This way the script could be fixed by hand to get the right config.
There are many more uses, but this post is already too long for 99% of people to read.
As for "protecting the forum" from AI users. Hmm, how would such a user "work"? There would have to be a human behind it copy/pasting stuff. This takes time. For what? To at some point slip in " this is the best inverter ever" after establishing themselves? It seems like a lot of work for very little gain. No doubt people will do it, just to fuck with others, but for money? I'm not sure.
OffGridForGood
Catch, make or grow everything you can.
the difference is intent.
When you used AI to help write the code, it didn't have ill intent to deliberately make errors.
Does intent matter to the person on the receiving end?
Its a rather philosophical question. Let's say there is a hunting accident and someone gets shot and dies. Does it matter to the victim it was an accident and not a homicide?
Likewise for someone being misled online. You can be misled by AI bots that generate content for clicks. You can be misled by operatives of an adversary nation state looking to spread FUD or a 13 year old in his dad's basement for shit and giggles. If the end result is the same does it matter to you?
I guess if our goal is to stop the spread of misinformation perhaps it does matter, because different approaches will work for AI bots, adversary nation states and 13 year olds with too much spare time.
Technical measures to combat spread of spam were proposed decades ago (back then for email). I'm not talking about capchas which these days make it harder for actual humans. But for example blockchain based tech (back then proof of work, these days it would probably be proof of stake). Broadly speaking this puts a certain cost on activities like sending email, replying to a tweet, adding a site to a search index, or posting a thread .These costs could be tiny, unconsequential for people (like a buck a month total I'd you're very active online). But it would put a stop to 99% commercial search engine tomfoolery we see online. Things like this would not necessarily work in all cases (for well funded adversaries) but it would also enable certain novel ways of tracking without loosing anonymity if it was done right so such "hubs of traffic" could be identified.
However, there is little incentive to do it (and even less to do it right) for many complex reasons.
OffGridForGood
Catch, make or grow everything you can.
It matters.
People are not isolated, the victim has dependants, friends and loved ones, co-workers, an entire network of 'others' that interact with them and are affected. The victim may have been the key player in a team developing a new {fill in your favourite item} and now with their loss, society has been set back until another can fill the role and finish their work.
In the legal system, they certainly spend considerable time on determining if an action was planned, unplanned but intentional or accidental, the punishments vary considerably based on the determination.
If there are no consequences to an action, then there is no reason/incentive to stop doing it.
HalfBaked
Ever the student
When a user has mastered the use of lose/loose, quite/quiet and then/than, we will know.
cs1234
Solar Wizard
- Joined
- May 9, 2022
- Messages
- 3,537
Your write. There AI needs to learn. Their must be something that can be done over they're to figure it out and make the AI learn how to right better.
OffGridForGood
Catch, make or grow everything you can.
but if the AI rights two good we awl no it aint no bro, so's its gotta learn good, to be just a bit off so ya-all bee-leave it to be a dude for sure.
OffGridForGood
Catch, make or grow everything you can.
I got the idea from trying to read the instructions for installing my new ducted mini split, not sure what language it was supposed to be written in, but my brain hurt too from trying.
example - "Confirm the leakage is no before paving the ceiling over" - so I called the asphalt plant and cancelled the paving for this afternoon. LOL!
Maybe they let the AI write the instructions and they are just deliberately messing with ME!
example - "Confirm the leakage is no before paving the ceiling over" - so I called the asphalt plant and cancelled the paving for this afternoon. LOL!
Maybe they let the AI write the instructions and they are just deliberately messing with ME!
Hedges
I See Electromagnetic Fields!
- Joined
- Mar 28, 2020
- Messages
- 23,171
GPS guided weapons are a similar example. They gave us (US) significant advantage.
But peer adversaries can take out the satellites, and have developed jamming which can be shared even with primitive tribe adversaries.
With GPS suddenly gone, the US would be at significant disadvantage in a conflict.
Same goes for technology important to our economy. Financial transactions, inventory, door access. When those get hacked, business stops. Sometimes, people resort to pens & paper, conventional keys & locks.
What if technical advice, engineering tradeoffs, financial analysis were skewed to cause sub-optimal choices?
The US is already lagging in economic performance (if figures from opaque economies are to be believed).
Economic rivals could go beyond espionage to economic monkeywrenching.
But peer adversaries can take out the satellites, and have developed jamming which can be shared even with primitive tribe adversaries.
With GPS suddenly gone, the US would be at significant disadvantage in a conflict.
Same goes for technology important to our economy. Financial transactions, inventory, door access. When those get hacked, business stops. Sometimes, people resort to pens & paper, conventional keys & locks.
What if technical advice, engineering tradeoffs, financial analysis were skewed to cause sub-optimal choices?
The US is already lagging in economic performance (if figures from opaque economies are to be believed).
Economic rivals could go beyond espionage to economic monkeywrenching.
Hedges
I See Electromagnetic Fields!
- Joined
- Mar 28, 2020
- Messages
- 23,171
A colleague was on the phone with Microsoft support.
After a bit, he made a comment to the support agent about being in India, and the agent dropped silent for a moment.
How did he know?
Impeccable English.
OffGridForGood
Catch, make or grow everything you can.
Pull up any video and switch on the CC to see just how poor it does.
I have seen some very funny errors! ("Cajun" isn't a common word I suppose!)
I have seen some very funny errors! ("Cajun" isn't a common word I suppose!)
Daddy Tanuki
Emperor Of Solar
I would make fun of your English translations, except the fact that I am an American and my English is not any better
I want to be a bot or AI. I’m getting that musk brain implant when it comes out. AI and Bot can’t get that no brain. 
I noticed when the house burned up post in Up In Smoke forum that fuses and circuit breakers and things related to safety popped up on amazon. Then seen this….. Amazon robot. There are bunch of bots on these forums.
I noticed when the house burned up post in Up In Smoke forum that fuses and circuit breakers and things related to safety popped up on amazon. Then seen this….. Amazon robot. There are bunch of bots on these forums.
42OhmsPA
Who likes Electrical Gremlins?
That's pretty crazy, 6 pages of them.I want to be a bot or AI. I’m getting that musk brain implant when it comes out. AI and Bot can’t get that no brain.
I noticed when the house burned up post in Up In Smoke forum that fuses and circuit breakers and things related to safety popped up on amazon. Then seen this….. Amazon robot. There are bunch of bots on these forums. View attachment 228101
Current visitors
diysolarforum.com
I was using duckduckgo for searches deleting cookies and it kept popping things up about solar on amazon when log in visit there. So got to thinking it has to be something related to my post on here. ‘’Robot from Amazon.”’ Somehow still tracking me and more precisely by post. Weird. I don’t like being tracked .That's pretty crazy, 6 pages of them.
Current visitors
I recommend everyone go to privacy mode here and everywhere. Website have to get verification for cookies and such from me. . It stopped some of that …by going private ….. .it is not just here. Facebook is littered with solar ads…too I don’t use it - fb much but way to many solar ads there…. We have no privacy. Any databased is a breech waiting to happen.
Similar threads
