In many cases, it already isn't, because things can be taken out of context and people may erase portions of video that create a very different scenario than what is shown to the public.
The scary part is thinking about when deep fakes will be undetectable. Especially with all these neural networks that are used for facial recognition. Eventually AI will be able to make perfect videos of people doing things they never did.
I think/hope DNA will remain be the de facto form of evidence because of this.
DNA is as unreliable as video evidence. All DNA can do is be exclusive or inclusive to a degree of certainty, i.e. only 1% of people would match this DNA profile. DNA cannot outright determine the person. To put this in perspective, many DNA matches are to prior known criminals because they're in the database. But if 1% of people would match, that's 3 million Americans who would match. How many of that 3 million could reasonably have been in the area and don't have an alibi? That number is small, but it's not 1 most of the time.
There's a great Netflix show about these things called Exhibit A. It goes over how "experts" and bad science bad understanding of good science have made things like video and DNA evidence into things that are manipulated into pinning people who ultimately weren't guilty.
In many cases, it already isn't, because things can be taken out of context and people may erase portions of video that create a very different scenario than what is shown to the public.
The scary part is thinking about when deep fakes will be undetectable. Especially with all these neural networks that are used for facial recognition. Eventually AI will be able to make perfect videos of people doing things they never did.
I think/hope DNA will remain be the de facto form of evidence because of this.
DNA is as unreliable as video evidence. All DNA can do is be exclusive or inclusive to a degree of certainty, i.e. only 1% of people would match this DNA profile. DNA cannot outright determine the person. To put this in perspective, many DNA matches are to prior known criminals because they're in the database. But if 1% of people would match, that's 3 million Americans who would match. How many of that 3 million could reasonably have been in the area and don't have an alibi? That number is small, but it's not 1 most of the time.
There's a great Netflix show about these things called Exhibit A. It goes over how "experts" and bad science bad understanding of good science have made things like video and DNA evidence into things that are manipulated into pinning people who ultimately weren't guilty.