Using Ai to connect events.

[Snautzer]

I tried to use AI to connect a general location,as given near Celle from a MACR, to a specific airfield


In this case it was Joseph F Kerch 795975 Pilot POW P-51 42-106698 ( 355th Fighter Group 357th Fighter Squadron) The MACR mention attacking airfield near Celle 15 APR 44


I ask GROK this ( using DaZengs work as main source)et all 15 APR 44 remarks from https://www.ww2.dk/Airfields%20-%20Germany%20[1937%20Borders].pdf, sort closets to Celle Germany


result:
Only 8 such dated remarks appear in the entire document (no other date formats or additional matches were found).

1. Fassberg (52°55′10″N – 10°11′30″E) — 33.5 km
   15 Apr 44: low-level attack by VIII Fighter Command P-51 Mustangs – claimed 2 x He 177s destroyed.
2. Salzwedel (52°50′10″N – 11°13′00″E) — 80.0 km
   15 Apr 44: low-level attack – 1 x Bf 109 G-6 from IV./JG 3 destroyed.
3. Hagenow (53°25′50″N – 11°13′30″E) — 117.8 km
   15 Apr 44: low-level attack by VIII Fighter Command P-51 Mustangs – claimed 1 x Bf 110, 2 x Ju 88s and 5 x He 177s destroyed, plus 2 x He 177s and 1 x unidentified aircraft damaged.
4. Oldenburg (53°10′40″N – 08°10′30″E) — 141.8 km
   15 Apr 44: low-level attack by VIII Fighter Command P-47 Thunderbolts – claimed 2 x Ju 88s destroyed and 2 more damaged.
5. Hesepe (52°27′00″N – 07°57′05″E) — 145.3 km
   15 Apr 44: low-level attack – 1 x Ju 188 A-2 from I./KG 2 shot up.
6. Fritzlar (51°07′00″N – 09°17′00″E) — 176.5 km
   15 Apr 44: low-level attack by VIII Fighter Command P-47 Thunderbolts – claimed 1 x Ju 52 and 1 x unidentified aircraft destroyed.
7. Flensburg (54°47′33″N – 09°25′43″E) — 244.7 km
   15 Apr 44: low-level attack by 4 VIII Fighter Command P-51 Mustangs – 9 x He 111s from I./KG 26 shot up and destroyed on the ground.
8. Aachen-Merzbrück (50°49′20″N – 06°11′15″E) — 334.8 km
   15 Apr 44: low-level attack – 1 x Do 217M from III./KG 2 destroyed.

As the MACR states the engine was cutting out and J-851 gives Ellendorf as crash place, which is near.


So Fassberg it was. I know for sure because in J-851 it states "Der Aufschlagort des Flugzeuges wird durch Fliegerhorstkommandantur Fassberg gemeldet"


So i feel confident that this way of using sources can be productive.




Google AI will not open an online pdf, and sice the copy function of windows (clipboard) is limited i used GROK

Hope it can be useful to somebody.

[Laurent Rizzotti]

As a data quality manager in computer science, I will give you a general advice to always check the result of "open IA" (meaning the IAs available to general public for free).

Current IA will very rarely reply "I don't know". It will search the internet (that is not always telling the truth) and build a reply from whatever it has found. Another problem with the IA is that it may make guesses in his work, and after some time will consider these guesses as proofs, as they are the basis of his reasoning.

Another issue I have with the IA is that there are now scores of archive documents available, but the IA will not scan them, only what is available on html pages, and sometimes on PDF when you point it in your prompt.

In my own queries, both professional and for my history hobby, IA is wrong about half the time.

As in your precise case, J-851 is not attached to MACR 4070, so the USAAF made no link between the two after the war. I have not J-851 available, but I presume the name of the pilot or the serial appears in it.

What is in the MACR is for me two foggy to be really usable: both pilots attacked an airfield and two minutes later the missing aircraft was seen still flying at 2500 feet with flames going out of his aircraft. As the witness did not see the pilot bail out or the aircraft crash, the pilot may have baled out some time later, or he may even have already baled out when the other pilot last saw his aircraft flying.

I can't find a place called Ellendorf in Germany, but there is an Ellerndorf a dozen of km NE Fassberg airfield, this is consistent with 2 minutes flying after the attack.

As two HE 177s were destroyed at Fassberg according to your site, it could be interested to find is an US unit made claims for them, especially if 355th FG or 357th FS did.

[Snautzer]

How can you even comment if you have no knowledge of the sources ?
Surely you know as a data quality manager in computer science you should first take notice of sources then research results no?


I am very aware af how AI currently works for the general public.


Quote: As in your precise case, J-851 is not attached to MACR 4070, so the USAAF made no link between the two after the war. I have not J-851 available, but I presume the name of the pilot or the serial appears in it."


Just little things you get. J-851 is a German file and i have not and presume is as bad as it comes in giving critique..


So if one combine the MACR the KU or KUJ or J files and Loss Register Studiegroep Luchtoorlog 1939-1945 France crashes 39-45 IWM American Archive Find a grave sometimes Dazeng,( i have a more) one can get sometimes interesting results.
I went through hundreds of files KU KUJ and J files alone. As base source material.
Now if one would prompt AI using these very valid sources still not good?
If so i know some companies who think differently


And as for "IA is wrong about half the time" well that is just how you prompt the machine now is it? And how you tell it to use what sources.
And who is wrong then, the machine or the one asking half baked questions on a specialist subject.


and as for"two HE 177s were destroyed at Fassberg according to your site" ...sources Laurent, in this case Dazeng work, not my site. Mistake right there. Not AI. yours.


As this is an aviation site i will not debate this matter here anymore.