This is more of a probability question. Having said that, based on each leaked script, you generate the possibilities of each of the characters dying (for example yes 0.2%, no 0.7%, unknown 0.1%) - you also give a weight for each leaked script based on on its perceived authenticity (from 0.1 to 1). You multiply the weight of each script by each of the possibilities and then you add the result for each possibility and each character across all scripts. So, for example, say you have 2 scripts. The first script has perceived authenticity of 20% (0.2) and the second script has perceived authenticity of 40% (0.4). In the first script, Character A has 50% (0.5) chance of dying, 20% (0.2) of not dying, and 30% (0.3) chance of unknown outcome. In the second script, character A has 20% (0.2) chance of dying, 50% (0.5) chance of not dying, and 30% (0.3) chance of unknown outcome. So the following applies for Character A:
- Dying: (0.2 x 0.5) + (0.4 x 0.2) = 0.18 (18% chance of dying)
- Not dying: (0.2 x 0.2) + (0.4 x 0.5) = 0.24 (24% chance of not dying)
- Unknown outcome: (0.2 x 0.3) + (0.4 x 0.3) = 0.20 (20% unknown outcome)
The probability of not dying is 24%, which is the highest.
Note: I am not saying that probability has nothing to do with AI (in fact, it is very important in AI).