Raw LLM Responses

I find these how to spot AI videos can end up having a negative impact. People want to know the rules for how to spot AI so when they watch something like this they think they have a set of rules to follow, like video length and obvious things like text not making sense or someone walking behind something then disappearing but AI is advancing at such a rate that the limitations keep shifting. The AI companies are probably working specifically on fixing these exact problems so they are likely to be the first things fixed (Gemini Veo already has an “extend” option to generate more video). People use to say, they could always spot AI images and then they would list the most obvious issues like peoples hands not being right, now AI can render hands quite well and depending on which AI you’re using, the text might also be quite good. One of the worst things is people falling for confirmation bias where they can spot some AI videos and images so they think the ones they spotted must be all of them. They don’t consider the fact that if they saw AI that was good enough to fool them, then it obviously didn’t register as AI so they are left only factoring in the imagery that they confirmed was AI. If you really want to test yourself, find an online AI recognition test where a bunch of images or clips are shown and you have to guess if they are AI or not. Then you will see how many you couldn’t pick and how many false positives you identified because you just wanted to say everything was AI to convince yourself that you couldn’t be fooled. Saying that real videos are fake, is just as bad as not spotting the real fakes.