Can AI use synthetic data generation to compute continuously to generate alternate timelines based on if different decisions had been made in the past?
[ - ] MaryXmas 1 point 9 monthsAug 4, 2024 21:58:38 ago (+1/-0)
Yes, but that is what life is called. Your experiences shape who you are and change the future. AI works the same way. The way you control for this is to identify seeds or starting points. Again, the theory isn't the issue, getting a model to do what you want it to can be fairly challenging, even if you know what you want. That is why I was asking for a use case. Like a big game of plinko. I would also argue that continuous/discreetly doesn't matter as a descrete decision could be made in a dynamic context based on the past.
A good way to describe AI would be plinko. You start at one point at the top and then each subsequent decision it makes is based off of the historical pattern that has come before. It will fall onto the next peg and that will contribute to the journey. It makes this journey by leveraging the most likely percentage for the path.
The seed is where you would start. This is typically random but reflects where on the top of the plinko board you drop the coin. Theoretically, if you drop the coin at the same place twice, it should have an identical journey because the decisions are made of math percentages. So if you can rotate through the seeds, you could see how each modification of scenario would impact the decision making and basically view the learned impact on the scenario which is what you are asking for.
This is mostly in theory as it would be incredibly difficult to view each step of the journey. Even reviewing the results would take a massive amount of time and processing power. This is stuff you would likely need a custom model to research but there are some models that will show the seed. I don't have an example of which ones.
The temperature is analog. A digital system could only measure it discretely. Could a digital AI take discrete data points and turn them into analog information? A sort of calculous of computing? Does my prompt make sense?
Yes,it makes sense. It wouldn't ever build the entire line, you would get it to dynamically infer the missing midpoints of the line depending on the context you provide it. You would keep zooming in closer and closer.
[ + ] xmasskull
[ - ] xmasskull 1 point 9 monthsAug 4, 2024 16:49:29 ago (+1/-0)
[ + ] Deleted
[ - ] deleted 1 point 9 monthsAug 4, 2024 19:24:59 ago (+1/-0)
[ + ] MaryXmas
[ - ] MaryXmas 1 point 9 monthsAug 4, 2024 20:05:28 ago (+1/-0)
[ + ] we_kill_creativity
[ - ] we_kill_creativity [op] 1 point 9 monthsAug 4, 2024 21:42:01 ago (+1/-0)
[ + ] MaryXmas
[ - ] MaryXmas 1 point 9 monthsAug 4, 2024 21:58:38 ago (+1/-0)
Again, the theory isn't the issue, getting a model to do what you want it to can be fairly challenging, even if you know what you want. That is why I was asking for a use case. Like a big game of plinko.
I would also argue that continuous/discreetly doesn't matter as a descrete decision could be made in a dynamic context based on the past.
[ + ] we_kill_creativity
[ - ] we_kill_creativity [op] 0 points 9 monthsAug 4, 2024 22:20:50 ago (+0/-0)
[ + ] MaryXmas
[ - ] MaryXmas 0 points 8 monthsAug 5, 2024 04:50:27 ago (+0/-0)
The seed is where you would start. This is typically random but reflects where on the top of the plinko board you drop the coin. Theoretically, if you drop the coin at the same place twice, it should have an identical journey because the decisions are made of math percentages. So if you can rotate through the seeds, you could see how each modification of scenario would impact the decision making and basically view the learned impact on the scenario which is what you are asking for.
This is mostly in theory as it would be incredibly difficult to view each step of the journey. Even reviewing the results would take a massive amount of time and processing power. This is stuff you would likely need a custom model to research but there are some models that will show the seed. I don't have an example of which ones.
[ + ] we_kill_creativity
[ - ] we_kill_creativity [op] 0 points 8 monthsAug 5, 2024 22:30:53 ago (+0/-0)
Can AI take discrete data points and turn it into continuous data? Can it take digital data and turn it into a smooth line?
[ + ] MaryXmas
[ - ] MaryXmas 0 points 8 monthsAug 6, 2024 14:51:21 ago (+0/-0)
[ + ] we_kill_creativity
[ - ] we_kill_creativity [op] 0 points 8 monthsAug 6, 2024 22:39:22 ago (+0/-0)
[ + ] MaryXmas
[ - ] MaryXmas 0 points 8 monthsAug 7, 2024 00:12:30 ago (+0/-0)
[ + ] DitchPig
[ - ] DitchPig 0 points 9 monthsAug 4, 2024 17:50:03 ago (+0/-0)
Yes.