I would hate to have to exist without CoPilot...

[Zed Clampet]

I would hate to have to exist without CoPilot. Any problem I have with Windows, it immediately gives me the fix for it. These fixes are often not easily found online. For instance, using a remote service to access my desktop PC broke the onscreen keyboard. I told that to CoPilot, and it said "open Powershell as admin and paste this" and it gave me a line to copy and paste. About 10 seconds later the onscreen keyboard was working again.

Compare this to a Google search. First of all, the AI summary at the top fails to give you a working answer. Then you have to sift through dozens of different problems that people have had that don't apply to you. Every one of these problem solutions can be found in the Google AI roundup at the top, and none of them work. So the first page of Google is completely useless and a huge time waster, especially if you watch the "How to fix your onscreen keyboard" videos.

So far, every problem I've had with Windows, and I've had a bunch, CoPilot helps me fix immediately. If you aren't using CoPilot for this, then you are doing it wrong. Plain and simple. Usually it gives you a list of potential fixes. Giving as much info as possible in your question reduces the size of this list. The more vague you are, the bigger the list.

I also asked CoPilot, earlier, how many bosses were in V-Rising right now, and it gave me the right answer. I typed it into Google, and the AI summary gave me the wrong answer and all of the articles it linked were wrong and using dated material. Now, I went back to Google a couple of hours later, and asked again, and the AI part gave me the right answer the second time around, so it must have researched more after giving me the wrong answer the first time, so theoretically it should be getting better as it goes, but that didn't help me the first time I asked.

There are just so many questions that you can't get answers from Google that you can get answers from CoPilot.

 

[Zed Clampet]

I said something stupid above about Google Gemini and the V-Rising. When you type in your search query, it will be sent to one of probably thousands of Geminis. The one that got my V-Rising question the first time got the dates wrong, which is something AI still has some problems with. If it were 2025 at this time, its answer would have been correct. The AI that got my search request the second time got the date correct, so it gave me the correct answer. Whether they get the correct date or not likely depends on the seed they are working on.

@Kaamos_Llama @neogunhero I think that at some point each of you got a different answer from AI than I did, and we didn't know how that happened. It's because there are many instances of each AI and they are working on different seeds.

 

[Colif]

I generally don't get that many problems with Windows that I can't solve myself but then I did answer questions about windows for 10 years.

Unlike you I don't install a ton of programs any more, I put the same ones I always used on any PC and rarely add any thing new. Windows works fine if you leave it be

I generally get problems no one has had before.

I have ignored Copilot since they introduced it, I tried it then and it was a joke. Guess its nice to hear its actually useful for those who need it

 

[Pifanjr]

I've only used CoPilot once, when my Windows updates failed to install, and it was indeed pretty good at giving a bunch of suggestions that I hadn't found by looking for the answers myself, next to all the ones I had found myself.

Which is both an indication of how much LLMs have improved and how much search engines have gotten worse.

It should be noted however that none of its suggestions ended up working except the one to just to an in-place installation of Windows, which I had already found myself, so in the end it didn't actually help.

 

[Colif]

except the one to just to an in-place installation of Windows, which I had already found myself, so in the end it didn't actually help.

if you hadn't already found repair install, it would have helped.
repair installs generally need an installer for the same version update of windows (version: I mean the yearly update, not if its win 11 or 12), so they aren't always the perfect answer. Only works if you on latest already.
I did one on my last PC. Much better than a clean install, all you need after are a few patches.

It did its best, its not psychic so doesn't know what you have done before, unless you tell it. Having answered questions myself, you offer the fixes you know, unless the question shows they did that already. Its a matter of whittling down the possible answers.

 

[Kaamos_Llama]

*Angry luddite noises*

Random poorly considered shower thoughts:

-At what point is so much cognitive load shifted onto an AI assistant that we no longer have any true thoughts of our own?

-If everyone becomes reliant on AI for helping then to understand things, what does that mean in relation to the type of people who are telling the LLMs what information they draw from and how to present it?

- Isnt it a nice thing to have to communicate with other people to solve problems and learn things? That was the whole point of the internet to begin with.

-Does making things exponentially more convenient as time goes on really improve everyones quality of life in the long run?

Grumble.

OTOH.

I understand that when people first began to write things down some thought at the time it would ruin everyones memories and make everything worse. Maybe thats me right now.

 

[Colif]

At what point is so much cognitive load shifted onto an AI assistant that we no longer have any true thoughts of our own?

where do I start and the AI end? depends on intelligence and how much you trust the AI. Have to have some sense of self before you engage with a AI. Have to know its just a tool and not a person.

Should keep it away from kids but that might be too late. You have to know who you are before you ask it for help with personal things. You need to know what is a bad idea.

-If everyone becomes reliant on AI for helping then to understand things, what does that mean in relation to the type of people who are telling the LLMs what information they draw from and how to present it?

the 2nd part is already a problem, what sources do you use, and which are ignored... it should see them all. But then not everything is true. Who gets to decide what is real?

If everyone relies on it, no one will know what is real... it will be like TV before the internet... you have to believe what is shared with you is real. So is an AI agent telling you what to think any different to the TV doing the same?

Isnt it a nice thing to have to communicate with other people to solve problems and learn things? That was the whole point of the internet to begin with.

and the purpose of forums 😀

-Does making things exponentially more convenient as time goes on really improve everyones quality of life in the long run?

We aren't any smarter than people 1000 years ago, we just have better tools. Its possible that AI right now had too much investment and was over hyped. I am not saying its a bad thing but just like NFT, lots of companies invested lots of money buying hardware without spending enough on buildings to house them... and it takes years to make the data centres. So unless they can wait and not make any money in that time, some won't survive.

 

[Kaamos_Llama]

Should keep it away from kids but that might be too late. You have to know who you are before you ask it for help with personal things. You need to know what is a bad idea.

Yea thats a big problem. I saw a 7 or 8 year old kid walking along a while ago literally talking to ChatGPT on a phone as if it were a friend. So weird.

the 2nd part is already a problem, what sources do you use, and which are ignored... it should see them all. But then not everything is true. Who gets to decide what is real?

If everyone relies on it, no one will know what is real... it will be like TV before the internet... you have to believe what is shared with you is real. So is an AI agent telling you what to think any different to the TV doing the same?

Its ok, because the free speech people will definitely make sure its all completely balanced for everyone.

Spoiler

/s

and the purpose of forums 😀

Yar.

We aren't any smarter than people 1000 years ago, we just have better tools. Its possible that AI right now had too much investment and was over hyped. I am not saying its a bad thing but just like NFT, lots of companies invested lots of money buying hardware without spending enough on buildings to house them... and it takes years to make the data centres. So unless they can wait and not make any money in that time, some won't survive.

We are mostly better educated, but humans always gonna human I spose.

 

[BeardyHat]

AI is just another tool in the belt. I've found it mostly pretty useful for helping me troubleshoot issues quickly and efficiently, though it's not always foolproof, it tends to be a lot quicker than Googling and searching various forums and tutorials and it's especially useful given you can quickly bounce questions back at it for clarification or further information.

I've used it primarily to help me troubleshoot Linux issues, which it's been great for. As well as a helping me nail down a specific cause of a behavior I was seeing on my car; I suspected the alternator, but ChatGPT showed me how to use my multimeter to get a better diagnoses. It's also helped me with my RC car, which saved me a lot of time asking questions on Reddit, which usually is slow and not terribly helpful.

I don't use it much more beyond that and I hate how sycophantic it is, that drives me nuts. I don't need to be constantly reassured that I'm valid and intelligent, just answer my question without trying to suck my...well, you know.

 

[Pifanjr]

Isnt it a nice thing to have to communicate with other people to solve problems and learn things? That was the whole point of the internet to begin with.

Ew, no. If I have a problem I just want a manual or some documentation on how to fix the problem, I don't want to have to twiddle my thumbs waiting for a human to respond to my questions.

 

[Kaamos_Llama]

Ew, no. If I have a problem I just want a manual or some documentation on how to fix the problem, I don't want to have to twiddle my thumbs waiting for a human to respond to my questions.

Well, communication doesnt have to be real time, online it counts if you come across a 10 year old forum post that solves the issue.

 

[Rolfil]

I have to admit that I use AI a lot. I use Opera and have chatgtp, gemini, perplexity and claude all in the sidebar. But they can they be really stupid sometimes when they scrape things like 5 year old reddit posts.

Mostly I use it as a replacement for search because it is a lot quicker than clicking through half a dozen sites, then remembering to hit the search tool and set the date to the last month or year, searching those sites and then getting frustrated when I can't find what I'm looking for.

You do have to have your wits about you though which is why I sometimes will ask different ai's the same question.

 

[Zloth]

I recently used CoPilot to look up what 'sycophantic' meant. It told me it meant obsequious, which is a tougher word for me to figure out than sycophantic, which at least I think I understand.

Ew, no. If I have a problem I just want a manual or some documentation on how to fix the problem, I don't want to have to twiddle my thumbs waiting for a human to respond to my questions.

Yeah, I'm seeing that at work a lot. There's a big chasm between the people who want to know exactly how their programs work (and how the users work with the programs) and the people that just want something that works right now. The latter are feeling really empowered by CoPilot.

We went through all this before with offshore programming. You pay programmers in poorer nations pennies on the dollar, and they will write programs for you to your specifications. Except the programmer has no idea what the 'big picture' is, they just know they are getting this list of variables and need to send back this answer - and they do exactly that (probably). They will rarely ever ask questions about how all this is supposed to work if a server goes down for a day, or what are you supposed to do if the person has no middle name at all, or one of the other jillions of things managers always forget to plan for. Then, when it screws up, they can't be of much help fixing it. After all, they don't know such-and-such server went down for a day. The best they can do is give a list of things that might be happening and let somebody that understands how everything really works figure out which of those things is happening. And there's nobody on the planet with that knowledge.

Mangers loved off-shoring at first. You pay less than what you paid for one programmer and get a dozen good programmers working on your code! But then the problems hit, or changes need to be made, and they're screwed. They have to hire a dedicated programmer to figure out how everything works (and fails to work) to deal with those situations. So, now they are paying off-shore AND on-shore!

Now we seem to have a pretty similar situation - except it isn't managers working with other companies, it's individual programmers working with CoPilot. Like the off-shore folks, CoPilot doesn't have a clue what the 'big picture' is, it doesn't know when it needs to question the requirements, and it will only be able to give you a list of possibilities when something breaks.

 

[Colif]

Ew, no. If I have a problem I just want a manual or some documentation on how to fix the problem, I don't want to have to twiddle my thumbs waiting for a human to respond to my questions.

as a human who answered, its equally annoying when the person who asks never returns to answer your follow up questions.

AI solves a problem many people have, they don't know how to properly search to find the answers. AI is a shortcut

Often knowing what to search is the hard part