Check out Google Genie Project.
Ha! - I wondered when someone would mention this project! 
Ever since MSFS was released, the game has always pushed the hardware envelope. A world that is streamed as an interactive video rather than one that is computed on the users’ hardware as a 3-dimensional mesh might reduce the minimum hardware requirements considerably.
In this “streamed video” paradigm, users would benefit as - just to provide one example - they no longer would require the latest & most expensive hardware in order to enable the highest fidelity of the simulation.
For Microsft / Asobo, this paradigm might, at last, separate program operation from user hardware, whatever that might be, be it PC, Xbox, PS5, SteamDeck, or whatever else is released to market in future years.
This server-side version of MSFS might allow far greater latitude in what could be included in the simulation. Features that might be ruled out as too hardware intensive could now be considered, as the vast majority - if not all - of the programs processing would now be done server-side rather than client-side.
When watching some of the examples given in Project Genie, I saw huge potential in how AI could transform flight simulation:
However, I’m not sure that we will be seeing this anytime soon, if at all. There are a number of very complex technical challenges to solve before we see a realistic AI World that truly represents our planet Earth. For example:
-
Real-World Fidelity: The worlds in Project Genie are imaginary. Portraying an accurate three dimensional view of the entire planet in AI is orders of magnitude more complex.
-
Physics Accuracy: Project Genie currently struggles to replicate real-world physics, which might be fine in an imaginary world, but not OK for a real-world flight simulator.
-
Complex Interactions: Interactions between entities in the current Genie worlds are oversimplified. Simulating intricate and highly detailed scenarios, such as ground operations at major airports would be considerably harder for an AI system to replicate, especially if expected to do so in real-time or near to real time. The same applies to systems such as the weather and aircraft cockpit systems.
-
Persistence. This is now available in a limited fashion in Genie3, but would need much improvement for a flight simulator.
-
Text-Driven Rendering: The current reliance on explicit text prompts provided by the user may not be practical for creating a flight simulator. The created environment is currently only as good as the data provided by the user. A significantly different input system that pulled data from many sources as well as direct interaction with developers would probably be required. How 3rd-parties, be they payware companies or the community, could participate in this type of server-side system is TDB.
and what might doom this concept from the outset:
- Scalability. Large-scale simulations - such as creating an entire planet world-wide - puts an enormous strain on the AI’s computational resources, and could lead to inconsistencies, anomalies and errors. Ensuring visual fidelity and reliability while maintaining performance would be critical.
The infrastructure required to do this at scale - think of 1 million+ users flying simultaneously in their own AI worlds - may be simply too cost prohibitive to make it a commercially viable proposition.
Project Genie, whilst impressive and full of potential, is currently one of many projects being worked on in Google labs. How it - or a Microsoft equivalent - transforms into a commercially viable product remains to be seen.
More AI slop is the last thing we need. Folks are going to really miss human programmers and artists when corporate replaces them to save a buck, and all we’re left with is an endless stream of hallucinated junk. I guess we’re already heading down that road, more and more quantity of content with less quality of content.
Genie looks like a tinker toy gimmick that users play with a bit for a bit until they’re bored. I don’t see this becoming a serious tool for game development.
I don’t know if you are following what is happening right now, but your idea might be a bit behind the facts of today. From a first hand perspective I can tell you that Vibe Coding with Copilot is mind bogglingly accurate. I am creating a complete, and complex project with project planning, milestones, deliverables, management summaries, cost planning, code architecture, you name it.
Hallucination is very rapidly becoming something from the past.
As with most things, non-experts won’t even be able to tell when AI hallucinates because it also does a good job sounding like it knows what it’s talking about. Eventually, the mistakes just become normalized, we just digest whatever it thinks it needs to tell us, and expertise goes into the bin. There will be a reckoning.
Do you have code inspectors? Just like you would with a building or any engineering project? I know we’re a bit off topic, but who knows when we’ll see the first AI generated airplane in the sim.
I find the opposite. Because I think because I may call myself an expert, I am able to give the right prompts and follow ups. And co-develop en application small step by small step. I think that when people talk in very generic terms about the supposedly lack of quality of AI, that they never have seen how this co-creation actually works these days.
Being derivative by way of an expert’s prompts is still being derivative. I can manipulate a yoke to fly an airplane, but theoretically my dog can as well. Only one of us is going to do it right.
What happens when a non-expert prompts it, or it prompts itself, or by means of some other algorithm? GIGO is still going to be a thing.
And I’m not saying we don’t need to be ready for an AI-driven world, but I don’t have high hopes for the outcome, in several ways.
Yes. Sanity checks are routinely executed by real humans at the moment. But with the standard rules of code reviewing in the context, AI can do the same quickly, and thoroughly, and find any orphaned functions, double declarations, you name it. It can even find performance killers. A whole new world has opened for me, and my creativity goes through the roof, because I am not limited anymore by my own slow and tedious C#-coding.
Many modern frameworks come with generators to stand up a basic application structure complete with data models and data access modules. The generated code is not meant to be production ready, but it does the heavy lifting for routine tasks that have to be done for every project.
I wish I could show you what I am working on at the moment. My colleagues were flabbergasted when I presented the first POC this morning.
But I agree with you on the dangers. We need to be very aware of them. Human in the loop we call it. Keep watching Harari 
.
Sorry for this side track. Couldn’t resist. Back on topic now?
Could be an interesting question for next QA. What percentage of MSFS live code is vibe coded or AI assisted. In my company it sits at 40% at the moment.
In General MS and Asobo did a brilliant Job bringing flight Simulation tob the next level.
Unfortinately I’m not flying a lot and switched completely to airliners as the distortion in ultra widescreens makes it Impossible für me to get a glimpse of speeds and distances which ist essentiell when flying helicopters or GA planes.
In VR that might be great, but in home cockpits with multiple beamers/screens it IS a huge problem, killing immersion completely - not only for me: Adjustable Projection Type and Field Of View would be highly beneficial for Widescreens and Curved Monitors
What makes me wounder ist the fact, that this issue has never been addressed in any stream or in the wishlist.
So keeping MSFS a Platform mainly for single screen 16:9 aspect ratio gamers, I’m sure, more and more advanced setup users will be forced to switch to XPlane.
Still hoping for at least an official comment regarding the huge distortion issue.
Joe
I have several control profiles for different aircraft.
But none of them are within FS24, since the controls setup is so horrendous only a programmer can like it. No tester was ever asked if this was even remotely practicable or they were asked and they all lied.
I have been a FSUIPC user since Pete released the first version decades ago. I have tried Axis-n-Ohs but could not really warm up to it. And right now SpadNxt fully covers all my needs.
Only exception is the Skyraider. Can not get the throttle to work unless it is assigned in FS24.
The main reason there was a FS2024 is release a PS5 version and reduce the size of the base client. I don’t think there will be an FS28. FS24 was already a complete mess for third-party developers with many devs still don’t use new SDK (FBW, iFly)
I get your point but let me explain how I came to my conclusion. The part of “… old codebase is no longer….” is meant to be ironical.
I think the new version will be driven by economics instead of technical necessity.
My assumption is based on that selling a new 200 EUR/USD Game/Simulator with a subsequent selling of Marketplace addons will create more revenue than only selling further marketplace addons for an existing game.
Finally I hope you are right and I am wrong.
Well, the old code base was not suitable for making study level airliners and other aircraft that have complex systems. Sure, there are a lot of developers who have not embraced the new SDK because it requires programming skills, not just hacking up some XML files.
The encryption that came with 2024 was to prevent devs from copying others work and selling it as their own. This is a good thing, but it has slowed the rate of new content even more.
If you look under Settings, Legal, and Credits, you will see that Boeing and Airbus do not grant licenses to MS for their trademarks. This makes them partners in the sim, and we saw that last year with the announcement that Boeing is using the sim for training. I assume Airbus is doing the same. This is good for us as I don’t think there are enough of us to support a sim this large and complex. So, having those large customers keeps the lights on for us.
Complexity has drawbacks though. The amount of effort it takes to bring a 2020 aircraft up to 2024 standards may be so much that the return on investment doesn’t make it worthwhile. The community doesn’t want to buy the same aircraft twice, even though it might be completely new under the hood. However, we have all paid for this content over and over in FSX, P3D, XP, AFS, etc.
I guess the big question is, “What will it take to get me to spend more money?” I have only a handful of planes I fly regularly out of the hundreds that are available. Sounds are everything for me. The first think I do when a plane comes out is check YT for sounds. If I don’t like the sounds, it’s not happening. Some developers make great planes with bad sounds, and I’ve given up hoping they will improve.
Next is flight model. If a plane doesn’t fly straight and level, meaning I don’t have to dance on the pitch trim the entire flight, I don’t fly it.
The thing is, there are a number of developers I just won’t buy anymore. I’m sure they don’t wish to drive their customers away, but they end up doing that because their busted stuff stays busted.
One plane I fly very often is the JU-52. It sounds and looks and handles great. I fly that in 2024 more than any other aircraft. If you are looking for market research, your new products are not reaching this consumer. I try new content, but I’m buying less and less. I don’t trust that new content will be good because there’s a serious lack of integrity in this industry, and there’s a lot of junk in the sim.
I would even go that far to say it is leading by example.
How many of the 125 Aviator Edition planes are native 2024?
If not even MS/Asobo putting the effort in to bring all the planes they sold a second time up to native 2024. Hell they even did charge full price, not even an upgrade fee for them.
Why should a much smaller 3rd Party Developer do this for free or put the effort in to change them.
Most of them – not all – did follow the example from the Mothership. I do not blame them. If they request an upgrade fee for a native 2024 plane, I am personally totally fine.
Just my 2ct
Simple, ZERO. But Microsoft already say have zero upgrade fee for all Microsoft Content.
But many aircraft original dev is not longer active, but Inibuild and SH Software Aircraft have more chance get update then other add on
Msfs24 is Vista of msflightsims, that is sad true but true… 
Not even close.