People have speculated for decades what the biggest dangers of ever-evolving artificial intelligence will be to the world and to society.
I think the most pressing and realistic danger is that of "deep fakes": Images, audio and video footage that's so super-realistic that it's literally impossible to distinguish it from genuine material. At the moment of writing this it's still possible to detect deep fakes algorithmically and with "digital forensic" analysis, but this is becoming harder and harder by the month. The inevitable result is that we are very close to a point where it will be literally impossible to tell if something has been generated by an AI or whether it's genuine footage.
And that's not even the worst aspect of it: The absolute worst aspect is how trivial it is to generate these deep fakes. In the past convincing fake footage (especially video) required a huge amount of work and experience, and often expensive equipment and software. However, nowadays anybody can trivially create this kind of fake footage in mere minutes with almost zero effort, knowledge or experience.
There is also another very closely related side effect of that which, although might not be as "dangerous", it still has the potential of eroding society as we know it, and perhaps even destroying it: The sheer amount of lazily created AI-generated material will skyrocket.
The time will come, very soon, when 99% of content online, be it on YouTube or other video sharing websites, be it on image boards, be it on social media platforms, be it on discussion forums, be it on blogs, will be AI-generated slop.
And the vast, vast majority of it will be made not only to pander to the lowest common denominator, but moreover to try to gain as many views and interactions as possible via clickbait or shock value. "Fake news" will become the norm. Fake articles, fake news broadcasts, fake "interviews", completely made-up fake opinion pieces, fake events being depicted.
It would be one thing if the AI-generated slop was just for entertainment purposes, like cartoons or video skits, or even cat videos. The problem is that the vast majority will try to gain views and engagement through clickbait and shock value rather than mere innocent entertainment. And how do you do that? By generating fake stories, fake claims, fake interviews, fake statements made by people, fake events, all of which is passed as real rather than having been clearly marked as fiction.
We are very quickly becoming a world where you can't trust anything you see online. There's footage of some big accident that happened somewhere? Probably a deep fake. Some celebrity made a controversial statement on video? Probably a deep fake. People being interviewed on the street and giving funny answers to questions? Probably a deep fake. Someone playing an instrument in a very skillful manner? Probably a deep fake. A "scientific breakthrough" having been discovered and announced? Probably a deep fake. You can't trust anything.
All the genuine material and the real events will be buried under the gigantic amounts of AI-generated slop. Soon you will be outright unable to even find the real thing anymore.
Not only will this make people dumber, but it can also be used for nefarious purposes.
Comments
Post a Comment