ChatGPT and DCS help questions... Interesting...

[HwyStar]

I'm a new user. I really wish that "Create Fast Mission" would add AWAC's automatically to the missions, but as far as I can tell, it does not.

So, I thought ChatGPT could help answer my question: How do you create an A/A mission in DCS with AWACs?

ChatGPT's direct reply:

Quote

 

To create an A/A mission in DCS with an AWACS, you can follow these steps:

1. Launch DCS World and select the mission editor from the main menu.
2. Start a new mission by clicking on the "New Mission" button.
3. Select the map you want to use for your mission and click "Ok".
4. Set the date and time for your mission.
5. Place your aircraft on the map by selecting it from the units list and dragging it to the desired location.
6. Select an AWACS aircraft from the units list and place it on the map.
7. Configure the AWACS by selecting it and opening the properties window. Set its altitude, speed, and communications settings.
8. Add enemy aircraft to the mission by selecting them from the units list and placing them on the map.
9. Configure the enemy aircraft by setting their altitude, speed, and other properties.
10. Add waypoints to the mission by selecting the aircraft and clicking on the "Waypoints" button. Set the desired waypoints for each aircraft.
11. Set the radio frequencies for the AWACS and the friendly aircraft so they can communicate.
12. Save the mission and test it in the simulator.
Note that the exact steps may vary depending on the specific version of DCS you are using and the aircraft and map you have selected.

 

Now, I know that it pulled this response from a webpage somewhere (probably from the DCS forum), but the grammatical response is perfect! It appears to have read the webpage then converted it into a perfectly written response in English. There are things missing from it (different settings per line item), but it gave me the "jist" of how to create a simple mission like I requested. 

Interesting response that it gave, in that, it's response works better than the "Search" field of the forum...

Need start-up procedures for a plane? ChatGPT it. Have questions about the targeting modes of a plane explained? ChatGPT it.

No flaming, please! I just thought that this could be a good tool for new users to explore while they learn the craft.

At some point, I bet that an API call could be run directly run from DCS World to get a response from ChatGPT based upon a users query and we wouldn't even need to leave DCS for questions. I'm just blue-skying this... But something like this will eventually happen.

"Sompin" to think about.

[HwyStar]

Here's another cool ChatGPT tidbit I just tried.

I asked ChatGPT: "Can you write some mission code for DCS that pits me against 5 enemy F-15C? I am flying the Su-27.

I will load this and see what happens. 

Quote

options = {
    -- Set mission options here
    coalition = "red",
    debug = false,
}

-- Define your aircraft
aircraft = {
    ["player"] = "Su-27",
    ["enemies"] = {"F-15C", "F-15C", "F-15C", "F-15C", "F-15C"},
}

-- Define the player's starting position and heading
player_start = {
    x = -80000,
    y = 0,
    heading = 0,
}

-- Define the enemy flight starting positions
enemies_start = {
    {
        x = 80000,
        y = 20000,
        heading = 180,
    },
    {
        x = 80000,
        y = -20000,
        heading = 180,
    },
    {
        x = 80000,
        y = 0,
        heading = 180,
    },
    {
        x = 80000,
        y = 40000,
        heading = 180,
    },
    {
        x = 80000,
        y = -40000,
        heading = 180,
    },
}

-- Define the mission objectives
mission = {
    ["tasks"] = {
        {
            ["type"] = "CAP",
            ["params"] = {
                ["communication"] = true,
                ["duration"] = 1800,
                ["engageAfterburner"] = true,
                ["enroute"] = true,
                ["fuel_mass"] = 5400,
                ["groupAttack"] = true,
                ["max_distance"] = 50000,
                ["min_distance"] = 10000,
                ["patrol"] = true,
                ["patrol_zone"] = {
                    ["max"] = {
                        ["x"] = -80000,
                        ["y"] = 80000,
                    },
                    ["min"] = {
                        ["x"] = -80000,
                        ["y"] = -80000,
                    },
                },
                ["speed"] = 900,
                ["start_time"] = 0,
                ["task"] = "CAP",
                ["threat_weapons"] = true,
                ["threats"] = {
                    ["F-15C"] = 5,
                },
            },
        },
    },
    ["briefing"] = "Your mission is to engage and destroy the enemy F-15C aircraft. You are flying the Su-27. Good luck!",
    ["map"] = "Caucasus",
    ["weather"] = {
        ["atmosphere_type"] = 3,
        ["clouds"] = true,
        ["enable_fog"] = false,
        ["fog_thickness"] = 0,
        ["visibility"] = {
            ["distance"] = 100000,
        },
        ["wind"] = {
            ["at2000"] = {
                ["direction"] = math.random(360),
                ["speed"] = math.random(0, 20),
            },
            ["at8000"] = {
                ["direction"] = math.random(360),
                ["speed"] = math.random(0, 30),
            },
            ["atGround"] = {
                ["direction"] = math.random(360),
                ["speed"] = math.random(0, 10),
            },
        },
    },
}

-- Set up the player's aircraft
red_player = planeGroup:new(1)
red_player:addUnit(aircraft.player, 1)
 

 

 

 

 

[Dragon1-1]

Note that DCS is quite unique in that there's not much BS on those topics, the info that is there is mostly correct. ChatGPT doesn't actually think, but if you constrain it to an area where its data is correct, then the results should be mostly correct, as well. With more complex programming languages it's very prone to inventing API calls and such, in a way that you'd think such a call exists (and sometimes, it'd be really nice if it did), but it doesn't. I don't know if it does that for DCS scripting or not, but it might.

[BitMaster]

Thing is, I doubt they will come for free like Google does.

Such a powerful and expensive to operate tool will be limited and taxed by question, letters or BTU's used.

I could imagine some form of free AI that is severely limited in one or the other way.

 

But aside from the "who pays for the bill" question, it is really game changing and frightening same time.

[Dragon1-1]

It's less disruptive than you think. Useful, sure, but it's not Skynet. It doesn't think, it has no agency and no creativity. As such, it will only replace humans who could be so described, and only as long as it's cheaper than hiring another office drone. This is a productivity tool, nothing more. Left to its own devices, it'll usually string together BS, and will never create anything truly meaningful. I guess we could replace politicians with it, at least.

Also, there's the issue of training sets. There are big bucks made in AI, and people will want their cut. Those huge sets of content those things train on? All had to be made by someone. All of those "someones" might feel entitled to a chunk of the AI's revenue, or to a measure of control with regards to how their work is used. This is already starting to happen, authors are suing for unauthorized use of their work. Making something part of "big data" doesn't automatically waive all copyright and licensing. Big companies own enough data that it might not hobble them much, but even they will have to make sure it's all properly licensed. Specialist AIs probably won't have that problem (and likely, they won't BS as much), but they're not the ones in the spotlight.

The third issue, one that might put a brake on things AI is allowed to write, is liability. ChatGPT already managed to slander a few notable figures badly enough for them to sue. Damages are one thing, but stopping it from doing that again is another. As the AI doesn't have anything that could be described as understanding of what it writes, it'll be difficult to constrain it from telling that kind of stuff, particularly against random people (it's already blocked from writing on certain political hot button issues). If it can't write about any random John Smith just because an actual John Smith may feel slighted and sue for defamation, it'll be very limited in what it can do. It doesn't even have to involve malice on part of the person writing the prompt, so shifting liability there won't work.

I can see this kind of tech finding its niche, as a fancy "autocorrect/autocomplete" type aid for programmers, replacement for certain jobs such as copywriters, or as specialist systems like the ones currently in experimental use for drug discovery. Ultimately, aids for humans, expanding their productivity while reducing drudgery somewhat. However, it's not a replacement for a human brain. If a job can be done by AI, it was probably brainless in first place.