I'm not suggesting AI systems should replace us guys on Avsim when seeking help and advice,  but as an exercise and out of interest, I thought I'd give it a go. What I discovered simply blew me away.

This is my question followed by the answer...

Q)  For some reason I have just started getting stutters on final approach to some of the big airports in MSFS 2020. Can you suggest a reason for this?

A) There could be several reasons why you are experiencing stutters on final approach in Microsoft Flight Simulator 2020. Here are some possible explanations:

1. High graphics settings: If you have set your graphics settings to high or ultra-high, this could be causing your computer to struggle to keep up with the demands of the simulator. Try reducing your graphics settings to see if this improves performance.

2. Outdated drivers: Outdated or incorrect graphics drivers can also cause stutters and other performance issues. Make sure you have the latest drivers installed for your graphics card and other hardware components.

3. Background processes: If you have other applications or processes running in the background while playing MSFS 2020, they could be using up valuable system resources and causing stutters. Try closing any unnecessary programs before starting the simulator.

4. Overclocking: If you have overclocked your CPU or GPU, this could be causing instability and stutters. Try returning your system to its default clock speeds and see if this helps.

5. Insufficient RAM: If you do not have enough RAM installed in your system, this could cause stutters and other performance issues. Consider upgrading your RAM if possible.

6. Internet connection: MSFS 2020 requires a stable internet connection to stream scenery and weather data. If your internet connection is slow or unstable, this could cause stutters and other issues.

Try these suggestions to see if they improve your performance on final approach in MSFS 2020. If the issue persists, you may need to troubleshoot further or seek additional assistance from Microsoft or the flight sim community.

Here's another example:- 

Q) Which flight simulator provides the most immersive experience?

A)  There are several flight simulators available that provide a very immersive experience, but the one that provides the most immersive experience depends on the user's preferences and requirements. Here are a few popular options:

1. Microsoft Flight Simulator: The latest version of Microsoft Flight Simulator (MSFS) is considered by many to be the most immersive flight simulator available. It offers stunningly realistic graphics and a vast, detailed world to explore. It also features live weather, real-time air traffic, and real-world flight planning tools, making it feel like a truly authentic flight experience.

2. X-Plane 11: X-Plane 11 is another popular flight simulator that offers a high level of immersion. It features realistic flight models and physics, and its global scenery database provides detailed and accurate representations of airports, cities, and terrain. It also has a large community of third-party developers who create add-ons to enhance the experience.

3. Prepar3D: Prepar3D is a professional-grade flight simulator developed by Lockheed Martin. It is used by airlines, military organizations, and flight training schools. Its advanced features, such as realistic weather, air traffic control, and mission-based training scenarios, make it a highly immersive option.

Ultimately, the most immersive flight simulator depends on the user's preferences and what they want to experience. It's recommended to try out different simulators and see which one suits your needs the best.

 

Edited February 27, 20233 yr by Rockliffe

@Rockliffe, I'm not sure this qualifies for a MSFS Tips and Tricks post. It's fairly generic advice regarding simulator problems. Comparison of the three major sims is hardly new info.

I've moved it to the main MSFS forum but it may be it's more suitable in Hangar Chat. Was it your intention it is most suitable in Tips & Tricks for one of the three sims? I'm not convinced. Let's see how things go.

I tried it asking it the answer to a couple of long standing things I didn't have a clue about. 

As expected, the results looked convincing but were totally useless and wrong. 

1 hour ago, Ray Proudfoot said:

@Rockliffe, I'm not sure this qualifies for a MSFS Tips and Tricks post. It's fairly generic advice regarding simulator problems. Comparison of the three major sims is hardly new info.

I've moved it to the main MSFS forum but it may be it's more suitable in Hangar Chat. Was it your intention it is most suitable in Tips & Tricks for one of the three sims? I'm not convinced. Let's see how things go.

Yeah, sorry Ray. I have posted it in the wrong forum as you quite rightly pointed it out. Call it old brain syndrome! 

36 minutes ago, superspud said:

I tried it asking it the answer to a couple of long standing things I didn't have a clue about. 

As expected, the results looked convincing but were totally useless and wrong. 

Sure, of course it's not foolproof, but it's amazing at just what it comes up with.

"not foolproof" = virtually useless.

Here on Avsim, if someone gives you dumb advice, 10 other people will correct them and you won't follow the dumb advice.

ChatGPT is working without a net, and without a brain. So it's going to give you dumb advice and there won't be anyone to say it's dumb.

 

I've messed with it some. It's a very advanced Ask Eliza. It might even be able to pass the Turing test if the test were administered by a particularly credulous proctor, but only in the same way that a magician can saw a lady in half without becoming a murderer - it's all smoke and mirrors.

 

It's not smoke and mirrors. It's fantastic in my line of work. It's helped me significantly with distributed microservice design and finding faults with my architecture. For software engineering and architecture it's absolutely brilliant. My whole team are using it. It makes mistakes at times, but it'll make fewer and fewer over time.

 

Smoke and mirrors, it is not!

1 minute ago, Bigbluss said:

It's not smoke and mirrors. It's fantastic in my line of work. It's helped me significantly with distributed microservice design and finding faults with my architecture. For software engineering and architecture it's absolutely brilliant. My whole team are using it. It makes mistakes at times, but it'll make fewer and fewer over time.

 

Smoke and mirrors, it is not!

Now you're for it! You've just dissed the whole of AVSIM's 'Experts at everything' 😵

30 minutes ago, Bigbluss said:

Smoke and mirrors, it is not!

It absolutely is. There's no real intelligence in artificial intelligence. You feed the AI gobs of information, it processes it and then uses it to establish patterns with which it tries to guess the proper output.

If you took a new AI, and fed it only those kids books that say "A is for Apple," etc, you'd end up with a system that thought AI meant Apple Insect. Do the same thing with a human and they'll intuitively know that A might be a component of the word apple, but not to think "Apple" every time you see "A."

AIs have become very sophisticated, and when I said they're smoke and mirrors I didn't mean they aren't useful. But the idea that we can feed, for instance, a distributed microservice design requirement to an AI and have it, guaranteed, spit out an excellent design without knowledgeable supervision is absurd. As you said yourself, it makes mistakes, and if you weren't there to apply your expertise to vetting those results, they'd be useless because we'd never know which designs were good and which were garbage. AI is a tool, but as with most tools there are applications for which it is ideal and applications for which it can be used, but the results won't be optimal. I can butcher a chicken with a table saw, but the resulting dinner will be substandard.

Where using ChatGPT to fix computer issues stumbles is that it's useful if the output is given to someone who already understands computers because that person can vet the answer to determine whether or not it's likely to be a good answer, but otherwise, you don't know if what it tells you to do will screw something up or not, because *it* doesn't know.

That list of solutions above looks exactly like those stupid websites that give 5 or 6 SEO-optimized suggestions to fix various Windows problems. The answers almost never work, but they rank high in search engine results. 

Edited February 27, 20233 yr by eslader

23 minutes ago, eslader said:

It absolutely is. There's no real intelligence in artificial intelligence. You feed the AI gobs of information, it processes it and then uses it to establish patterns with which it tries to guess the proper output.

If you took a new AI, and fed it only those kids books that say "A is for Apple," etc, you'd end up with a system that thought AI meant Apple Insect. Do the same thing with a human and they'll intuitively know that A might be a component of the word apple, but not to think "Apple" every time you see "A."

AIs have become very sophisticated, and when I said they're smoke and mirrors I didn't mean they aren't useful. But the idea that we can feed, for instance, a distributed microservice design requirement to an AI and have it, guaranteed, spit out an excellent design without knowledgeable supervision is absurd. As you said yourself, it makes mistakes, and if you weren't there to apply your expertise to vetting those results, they'd be useless because we'd never know which designs were good and which were garbage. AI is a tool, but as with most tools there are applications for which it is ideal and applications for which it can be used, but the results won't be optimal. I can butcher a chicken with a table saw, but the resulting dinner will be substandard.

Where using ChatGPT to fix computer issues stumbles is that it's useful if the output is given to someone who already understands computers because that person can vet the answer to determine whether or not it's likely to be a good answer, but otherwise, you don't know if what it tells you to do will screw something up or not, because *it* doesn't know.

That list of solutions above looks exactly like those stupid websites that give 5 or 6 SEO-optimized suggestions to fix various Windows problems. The answers almost never work, but they rank high in search engine results. 

I agree with you in places here - and I'm glad I do because otherwise I'd be out of a job. I can't agree that it is smoke and mirrors, because nobody is or ever has suggested that this is a general AI as far as I know.

We've got a long, long way to go before anything approaching general AI, and even if we do it'll be up for debate as to whether or not it constitutes sentience, but it's no more smoke and mirrors than Google is as a search engine. The way I see ChatGPT, is as an incredibly useful stateful search engine that is capable of consolidating the results into a useful and personalised response. It absolutely requires someone with expertise to work with it and double check its work, but generally speaking, its really, really useful. 

If you ask it follow up questions it tends to get more detailed. I might ask about a high level concept 'how do I handle messaging between services in distributed architecture' to which it might respond with Kafka, RabbitMQ or Redis. I can then start to drill down into the differences between them, ultimately ending up in exactly how to implement the service I choose at the level of code - without having to look at the docs. I'm very excited about the future of this tech. Its for sure not genuinely intelligent, but as far as ML algorithms gone pertaining natural language, this one is really impressive imo.

Edited February 27, 20233 yr by Bigbluss

26 minutes ago, eslader said:

It absolutely is. There's no real intelligence in artificial intelligence. You feed the AI gobs of information, it processes it and then uses it to establish patterns with which it tries to guess the proper output.

 

This is more or less what the human brain does, too, except for it has additional information available through physical senses and the central nervous system with its reward system which influence the output. What is "real" intelligence to you? And before you start to mention "free will"  - neurology says that the free will does not exist (and has quite some evidence for that), so we are technically machines that have a determined output based on input and our established patterns. And as for "consciousness" - there is and can be no proof that either we have consciousness or ChatGPT has no consciousness, because there is no objective concept of what consciousness is and how it can be proven or disproven.

EDIT: Also ChatGPT and alike are and will become very useful and the standard in everyone's household within a couple of years. Your definition of "useful" is too narrow, because you want it to do all things better than a human being. However you have to also consider how fast it can do these. While I can write a sophisticated Java code in a couple of hours, ChatGPT can do the same (and less buggy) within 1 minute. Imagine a computer that can - when permitted - program itself or other computers!

Edited February 27, 20233 yr by Fiorentoni

8 minutes ago, Fiorentoni said:

This is more or less what the human brain does, too, except for it has additional information available through physical senses and the central nervous system with its reward system which influence, which influence the output. What is "real" intelligence to you? And before you start to mention "free will"  - neurology says that the free will does not exist (and has quite some evidence for that), so we are technically machines that have a determined output based on input and our established patterns. And as for "consciousness" - there is and can be no proof that either we have consciousness or ChatGPT has no consciousness, because there is no objective concept of what consciousness is and how it can be proven or disproven.

Ah!  You beat me to it.

Given our most recent neurological theories about the nature of consciousness being an emergent phenomena with how our physical brain works, I think as this whole AI thing progresses we are going to be confronted with the “what’s the difference” question between what these Chat AI’s are doing vs what is happening to us.  At its base level our brain looks for patterns we have stored in ‘memory’ fro past experience and puts forth an appropriate response.  Perhaps the difference is only that we combine that with real time feedback from sensory inputs feeding back in a loop that gives us that sense of being self aware…. Quite a discussion for a flight sim forum huh…. 🤪

 

Edited February 27, 20233 yr by VFXSimmer

8 minutes ago, Fiorentoni said:

This is more or less what the human brain does, too, except for it has additional information available through physical senses and the central nervous system with its reward system which influence the output. What is "real" intelligence to you? And before you start to mention "free will"  - neurology says that the free will does not exist (and has quite some evidence for that), so we are technically machines that have a determined output based on input and our established patterns. And as for "consciousness" - there is and can be no proof that either we have consciousness or ChatGPT has no consciousness, because there is no objective concept of what consciousness is and how it can be proven or disproven.

EDIT: Also ChatGPT and alike are and will become very useful and the standard in everyone's household within a couple of years. Your definition of "useful" is too narrow, because you want it to do all things better than a human being. However you have to also consider how fast it can do these. While I can write a sophisticated Java code in a couple of hours, ChatGPT can do the same (and less buggy) within 1 minute. Imagine a computer that can - when permitted - program itself or other computers!

I heard a fantastic quote about this from a neuroscientist on a podcast I listen to: "We are what the processing of information feels like" or something to that effect. We're along for the ride but we're sure not as hell no in the driving seat. Really interesting stuff. When you dig deep enough, there's just absolutely no space for free will. An odd feeling - not that we could possibly feel any other way 🙂

Also, didn't expect so many people interested in the nature of consciousness and free will on a sim forum 😄

39 minutes ago, eslader said:

...You feed the AI gobs of information, it processes it and then uses it to establish patterns with which it tries to guess the proper output.

That's exactly what you do 🧐