Ki-61-II Kai & Engine Performance - Grok Edition

[Laurelix]

[Image of Ki-61-II Kai Otsu with the bubble canopy, Ko is the same plane but with the standard canopy]

Ki-61-II Kai Ko / Otsu
Empty Weight: 2855kg
Loaded Weight: 3825kg
Wing Area: 20m2
Engine: Ha-140

Grok Estimated Sustained Turn time at SL-1000m, no flaps
~21 Seconds at Military Power
~20 Seconds at WEP

Armament:
2x 12.7mm Ho-103
2x 20mm Ho-5

Maximum Safe Dive Limit: 850km/h IAS

 

[GTX]

Do we really need AI generated 'data' here? I would much prefer to see primary real sources.

 

[Laurelix]

Do we really need AI generated 'data' here? I would much prefer to see primary real sources.

its official sources but AI is used to calculate / estimate missing blanks

 

[GTX]

its official sources but AI is used to calculate / estimate missing blanks

Still see no benefit in allowing such AI generated so-called data.

 

[Laurelix]

Still see no benefit in allowing such AI generated so-called data.

most TAIC stuff for japan was calculated, the AI just does it for me and its based on already existing data which makes it more accurate than most TAIC reports

 

[GTX]

I think you will find many, if not most, members are uncomfortable with AI generated "data". There is a strong, valid IMHO, concern of such AI generated "data" polluting real historical references.

 

[Laurelix]

I think you will find many, if not most, members are uncomfortable with AI generated "data". There is a strong, valid IMHO, concern of such AI generated "data" polluting real historical references.

i agree the average person trying to do this with little knowledge could easily screw this all up
That's why i make sure the AI isnt pulling shit out of its ass but is simply filling in blanks on already known data

This is from Bunrindo who always quotes japanese official data
As you can see the engine power curve is there, as well as performance.. 610kmh at 6000m, 544kmh at 10,000m
6:30 to 5000m climb at military power.

Issue is the speeds at and climb rates at different altitudes arent listed and japanese dont include WEP figures.

 

[GTX]

is simply filling in blanks

This is still of concern because it makes it seem like this is real first hand data when it is not.

 

[Laurelix]

This is still of concern because it makes it seem like this is real first hand data when it is not.

the figures are not random but calculated based on what the horsepower is at the alttude that the sources dont provide speed for

 

[GTX]

the figures are not random but calculated based on what the horsepower is at the alttude that the sources dont provide speed for

It is still an AI generated result and not primary real sources. What validation is there that the so-called data here is real?

 

[Laurelix]

It is still an AI generated result and not primary real sources. What validation is there that the so-called data here is real?

Although the japanese primary source doesnt give the SL speeds for example... what grok calculated matches with the US TAIC report for Tony 2 max speed at sea level

 

[FulmenTheFinn]

I can understand using AI as an aid tool for performing certain calculations, but even then the results should be thoroughly checked and very plainly stated to originate with AI and thus be taken with a considerable amount of healthy scepticism. In my experience chatbots make mistakes with aircraft performance data calculations all the time, and spamming this sub-forum, or any forum purportedly dedicated to historical knowledge, full of entirely AI-generated tables and essentially presenting them as factual data is, IMO, crossing the line.

 

[Laurelix]

I can understand using AI as an aid tool for performing certain calculations, but even then the results should be thoroughly checked and very plainly stated to originate with AI and thus be taken with a considerable amount of healthy scepticism. In my experience chatbots make mistakes with aircraft performance data calculations all the time, and spamming this sub-forum, or any forum purportedly dedicated to historical knowledge, full of entirely AI-generated tables and essentially presenting them as factual data is, IMO, crossing the line.

For people like me who know how horsepower effects speed and how altitude effects speed and how critical altitude works, its easy to verify if the figures are plausible or nonsense.
Ive made sure the numbers are not nonsense.

But yes youre fully right to be sceptic, the average person trying to attempt this would have errors all over and i dont recommend anyone trying to attempt this as well.

 

[GTX]

The fundamental concern is that this AI generated "data" risks polluting the real historical sources. I suggest you watch the video here: How AI Slop is Killing World War II History to understand the concerns. I am not saying that you are deliberately faking things but please appreciate the concerns and why people are not comfortable with this.

 

[Laurelix]

The fundamental concern is that this AI generated "data" risks polluting the real historical sources. I suggest you watch the video here: How AI Slop is Killing World War II History to understand the concerns. I am not saying that you are deliberately faking things but please appreciate the concerns and why people are not comfortable with this.

Thats why i kinda limit it to planes with low data
No real need to do it for US or Soviet planes.
I did do the F6F hellcats cuz the data on them seems conflicting sometimes.

Its also good opportunity for people to find full official flight test data data and see how it compares.