Is it time to stop using human Adjudicator's and go to A.I.

[Onionhead2]

With all the debating over score movement,maybe its time to remove the human element. 

[Cappybara]

Who programs and trains the AI? 

Humans.

[Onionhead2]

1 minute ago, Cappybara said:

Who programs and trains the AI? 

Humans.

Blue Devil Alum. of course

[Ashontheinternet]

5 minutes ago, Onionhead2 said:

With all the debating over score movement,maybe its time to remove the human element. 

isn't the fun part of competition the unpredictability and movement throughout the season? i sure hope this idea never becomes a twinkle in the eyes of the drum corps overlords, else we be felled on black days. 🙃

[Ashontheinternet]

1 minute ago, Onionhead2 said:

Blue Devil Alum. of course

code would look something like this:

> IF: BLUE DEVILS PERFORM

>> GIVE: FIRST PLACE SCORE

[Chief Guns]

Sarah Conner is coming for you. 

[Onionhead2]

R2B2 with green shirts

[Ashontheinternet]

1 minute ago, Sensioto said:

No, but it would be cool to explore the capabilities of AI in composing drill or selecting music based upon a given theme to supplement ideas proposed by design teams.

which corps would be the most likely to have ChatGPT create a show theme and pick musical selections? my money's on bloo. they seem quite on the forefront of experimenting with tech in their programs what with Keytars and FX and chapels - er, uh, narration.

[MikeN]

Ask an AI to describe GE, you'd probably get the same word salad the humans use... 😞

Mike

[jwillis35]

Interesting question. I prefer the human element as subjective as it is. I think it's interesting to get those subjective views on show design, performance, GE. It's definitely not perfect but I don't know that it has to be. 

The interesting thing about AI is that it could potentially be used for the performance captions but it would take a lot of programming. You need vision and audio sensors at a super high level. Like if the AI could be setup to use vision (cams that are posted center low, center high, right and left) where it can be programmed to view feet, legs, upper body and somehow judge overall technique, timing, and then run a series of tests to come up with a score...that could be very cool. But the AI would have to already have a high-end benchmark of what it compares these stats to. It has to know what to look for. 

From an audio perspective the AI would be setup to take in the audio (brass, percussion, front ensemble). Can it separate these things because I do not think you want to judge all three together - as powerful as AI is. Can one program of AI lock into the brass and decipher attacks, releases, articulations, control of tone and volume, and especially musicality? It can certainly measure volume, playng time, even things like control and demand -- all this based on the benchmarks that are programmed into the AI Brass (for instance). 

Overall I think it would be interesting to see how an AI score stacked up against the human judges. It would take a lot of work to make sure it knows what to look for or listen to. It would also take a lot of trial. 

One area it could be used would be to judge and react to the audience. It can track applause points, standing ovations, the volume of the applause (a show that produces some big moments that the crowd reacts to), and other stats to try and determine how the audience reacts to each show. This could be a neat little set of stats that get added onto the recaps but are not part of the overall score (at least for now). But perhaps could be added to the GE caption in some small way. Who knows?