1 hour ago, RobJC said:

That is a very simplistic way of looking at the power of AI. It can also speed up learning dramatically, giving a wholistic view of the topic much quicker. I was wiring up some Victron components on my RV yesterday and wasn’t sure I had it working right. AI helped me understand (based on photos) that is was right, and why. Now I have a much better understanding of how these components work, and AI was providing valuable insight along the way. It told me about things i wasn’t asking about, and it did it extremely effectively. It was like having an expert next to me, being a patient teacher at the same time. And it remembers everything. The next time i ask about anything related to my RV it considers everything i have installed, other issues i have mentioned, etc.

Again, awesome.

Too bad that the likelyhood of it telling you lies ist high.

I gave it a task a while ago to find some stuff out in a field where I actually already knew the answers. And it took me half an hour of correcting it over and over again until it finally stopped making stuff up.

That's the problem I know people who consult it for serious matters and don't realize that it makes things up. Anyone who trusts this technology right now doesn't know what they're doing; it needs to mature a lot more I've seen it fail even at simple math calculations.

One thing AI can do is find and compille massive amounts of data. You could spend an entire weekend searching and not come close to what AI could do in a moment.

The question is - where is it getting that information from? That's the big issue. I find a lot of my MSFS 2024 releated seaches end up referencing threads right here at Avsim. That's generally a good thing, people here know their stuff way more than most. The problem is, maybe what I'm looking for isn't exactly the same thing being discussed. There's also the case of someone saying "I did X and Y happened". That's a 100% fact, but that doesn't mean that's the same outcome everyone will get. We've all seen weird things that behave differently with MSFS.

From there it will scour reddit and even youtube videos. Here's where you're already venturing into uncertain territory. Reddit can be hit or miss (and has a lot of bots itself). Youtube? Anyone can upload a youtube video.

If you've ever googled a random Windows problem, you know that much of what gets returned is flat out wrong or "snake oil". There's your "download this EXE to fix it" scams.

AI is meant to convince you that it knows what it is talking about, not necessarily give you the most accurate or best information. It does seem to have some logic to not just take everything, but it is far from perfect.

That's not to say this doesn't have value, it absolutely does. I was asking about an actual MSFS 2024 pause function, not just active pause. I entered it into an AI search and the results of a thread here were posted. That's good, it was exactly what I was looking for in this case. It gave me answers like just hit esc or there's an FSUIPC option to use (which I did).

That begs the question though, what value is AI providing if I already know to search here first? I don't need a supercomputer taking up all our hardware components to tell me what I already know.

2 hours ago, kerosene31 said:

That begs the question though, what value is AI providing if I already know to search here first? I don't need a supercomputer taking up all our hardware components to tell me what I already know.

You are at the core of the bubble that is AI. It's all about stocks and what it is really good for is using up insane amounts of energy resources.

I really have to wonder what its inventors were thinking, if they were thinking at all.

We are sitting in the midst of a man made environmental catastrophe and the goal is to reduce our carbon footprint and reduce anything that uses up more energy and they come up with a technology that increases exactly that consumption exponentially?

How stupid are these people?

9 hours ago, Farlis said:

Too bad that the likelyhood of it telling you lies ist high.

I gave it a task a while ago to find some stuff out in a field where I actually already knew the answers. And it took me half an hour of correcting it over and over again until it finally stopped making stuff up.

AI gets things wrong but it also gets a lot right. And it will only get better. All those corrections are making an impact, and within a few years those mistakes will be less and less. It is a game-changer for me. If it isn’t for you, I guess don’t use it.

I provided a spreadsheet with historical stock prices and volume, and asked it to use a very specific technical analysis method to give me price targets. It was so darn good that i run my other trades by it, not as the final say, but as a valuable data point.

On 6/28/2026 at 2:03 PM, Rocky_53 said:

I think the issue with AI, is that far too many people simply think it's always correct, I have had many suggesitons, tips, information that have been simply wrong. Often, a long way off being correct.

And lately, more and more people have then been posting the incorrect or partially correct "answers" here or Reddit or Discord or name your forum of choice. If the poster wanted an AI response instead of a response from a human that knew what they were talking about, they could just go straight to ChatGPT or Claude or Gemini or name your AI of choice.

LLM's do not know the answer. They do not know how to reason. They take the massive amount of data they have access to and churn out a response. The source of that response could be a sarcastic troll or a combination of information that may be related to original question but isn't actually the answer to the original question at all. I've asked AI about things readily available online... I asked the length of a 5090FE and it gave me random lengths from (presumably) other 5090 cards. I've asked AI the location of MSFS 2024's application folder and it gave me the location of 2020's folder instead. And it gave those wrong answers very authoratively!

AI isn't taking jobs because it knows how to do anything... it's taking jobs because the former human employees wrote very detailed processes and the AI was trained on that data. It's why AI is good at coding: there are millions of GitHub code repositories out there that helped train AI. But AI doesn't actually know how to code... it just knows how to spit out code that was already used, hopefully in a working application.

AI doesn't have access to the MSFS source code. It can't give any super special tips and tricks for MSFS that aren't already available online. Sometimes those tips and tricks are valid, and sometimes they are snake oil... just like AI itself.

AI is going to flop. And flop big time. Just as electric cars are not going to replace gas driven vehicles. Isn't going to happen.

1 hour ago, udidwht said:

AI is going to flop. And flop big time. Just as electric cars are not going to replace gas driven vehicles. Isn't going to happen.

😂😂😂😂😂

3 hours ago, RobJC said:

AI gets things wrong but it also gets a lot right. And it will only get better. All those corrections are making an impact, and within a few years those mistakes will be less and less. It is a game-changer for me. If it isn’t for you, I guess don’t use it.

I provided a spreadsheet with historical stock prices and volume, and asked it to use a very specific technical analysis method to give me price targets. It was so darn good that i run my other trades by it, not as the final say, but as a valuable data point.

Exactly this. As I mentioned, I use Claude and Claude Cowork on a daily, and it is, at the very least, a massive time saver.

The "negative" comments about AI on this thread mostly show a complete lack of knowledge of the current state of AI, or are based on what AI used to do a couple of years ago.

Obviously, if you use AI the way you use Alexa, the results are going to be predictable.

53 minutes ago, Ricardo41 said:

Exactly this. As I mentioned, I use Claude and Claude Cowork on a daily, and it is, at the very least, a massive time saver.

The "negative" comments about AI on this thread mostly show a complete lack of knowledge of the current state of AI, or are based on what AI used to do a couple of years ago.

Obviously, if you use AI the way you use Alexa, the results are going to be predictable.

Absolutely correct. It is like the people who said the internet wouldn’t catch on. I bet if we spend 5 minutes with them seeing how they use AI we would see the problem.

8 hours ago, Funky D said:

And lately, more and more people have then been posting the incorrect or partially correct "answers" here or Reddit or Discord or name your forum of choice. If the poster wanted an AI response instead of a response from a human that knew what they were talking about, they could just go straight to ChatGPT or Claude or Gemini or name your AI of choice.

LLM's do not know the answer. They do not know how to reason. They take the massive amount of data they have access to and churn out a response. The source of that response could be a sarcastic troll or a combination of information that may be related to original question but isn't actually the answer to the original question at all. I've asked AI about things readily available online... I asked the length of a 5090FE and it gave me random lengths from (presumably) other 5090 cards. I've asked AI the location of MSFS 2024's application folder and it gave me the location of 2020's folder instead. And it gave those wrong answers very authoratively!

AI isn't taking jobs because it knows how to do anything... it's taking jobs because the former human employees wrote very detailed processes and the AI was trained on that data. It's why AI is good at coding: there are millions of GitHub code repositories out there that helped train AI. But AI doesn't actually know how to code... it just knows how to spit out code that was already used, hopefully in a working application.

AI doesn't have access to the MSFS source code. It can't give any super special tips and tricks for MSFS that aren't already available online. Sometimes those tips and tricks are valid, and sometimes they are snake oil... just like AI itself.

Absolutely

AI is being used by hackers to breach cybersecurity, and by cybersecurity analysts to prevent breaches. One example is both Google and Mozilla have recently used AI to find many high-severity security flaws in their browser code.

Businesses are using AI to create apps, bots, and agents, and to automate processes; these all reduce human involvement and save them money.

Love it or loathe it, AI is here to stay.

On 6/29/2026 at 9:03 AM, RobJC said:

It can also speed up learning dramatically

Studies have shown otherwise.

And studies have shown using AI actually slows down productivity, one reason is because people have started second guessing themselves a lot more often and stop ask the AI if they're right.

Edited Tuesday at 01:29 PM3 days by Tuskin38

Someone needs to come up with a term for AI Luddite. According to AI, there isn't one yet.

44 minutes ago, Tuskin38 said:

Studies have shown otherwise.

And studies have shown using AI actually slows down productivity, one reason is because people have started second guessing themselves a lot more often and stop ask the AI if they're right.

Don’t agree in this.

I am working for a banc on the mortgage department.

A year ago we had 80 employees. Then they introduced AI. It made mortgage quotes 100 times faster and now there are only 10 employees left for checking the quotes.

The Insurance department : the same.

In the future AI will take over many tasks.