[lampiaio]

AI that is indistinguishable from reality is a certainty for the not-so-distant future.

That future will come, and it will come sooner than anyone's expecting.

Yet all I see is society trying to prevent the inevitable from installing itself (because it's "scary", "dangerous", "undermines the very pillars of society" etc.), instead of preparing itself for when the inevitable occurs.

People seem to have finally accepted we can't put the genie back in the bottle, so now we're at the stage where governments and institutions are all trying to look busy and pass the image of "hey, we're doing something about it, ok? You can feel safe".

Soon we will be forced to accept that all that wasted effort was but a futile attempt at catching a falling knife.

Maybe the next idiom in line will be "crying over spilled milk", because could someone point me to what is being done in terms of "hey, let's start by directly assuming a world in which anyone can produce unrestricted, genuine-looking content will soon come and there's no way around it -- what then?"

All I see is a meteor approaching and everyone trying to divert it, but no one actually preparing for when it does hit. Each day that passes I'm more certain that we will we look at each other like fools, asking ourselves "why didn't we focus on preparing for change, instead of trying to prevent change"?

[survirtual]

We've been preparing for a while? It's all that work people have been doing for years with asynchronous cryptography, ecc, and tech like what happens during heavy rain downpours and that coin with a bit in front of it.

These are all the proper preparation for AI. AI can't generate a private key given a public key. AI can't generate the appropriate text given a hash.

So we build a society upon these things AI can't do.

It has been a good run. We have done things like the tried and true ink stamping to verify documents. We have a labyrinth of bureaucracy for every little activity, mostly because it is the way that has always worked. It has surely been nice for the "administration" to sit around and sip lemonade in their archaic jobs. It has been nice to have incompetent people with no vision being appointed to high places for being born into the right families connected with the right people. That gravy train was surely a joy for those who were a part of it.

Sadly, it won't work anymore. We will need competent people now that actually care.

We need everything to be authenticated now with digital signatures.

It is not even that difficult a problem to solve. The existing systems are far more complex, far more prone to error, far more expensive, and far more difficult to navigate.

AI is giving us an opportunity to evolve. It is a time for celebration. Society will be faster, more efficient, more secure, and much more fun with generative content. AIs will produce official AI-signed content, and unsigned content. Humans will produce official human signed content, and unsigned content. Some AIs will use humans to sign content to subvert systems. But all of this pales in comparison to the fraud, waste, and total abuse of the current system.

[aussieguy1234]

Most nefarious AI content is going to be posted by humans misusing the AI tools, as opposed to some kind of AI gone rogue.

These humans would simply generate a public key from the private key, then post it under their human identity. The main threat from AI in the future IMO is not rouge AI, but bad human actors using it for their own nefarious agendas. This is how the first "evil" AI will probably come about.

[cush]

There are some interesting hardware solutions from camera makers that provide provably authentic metadata and watermarks to videos and images - mostly useful for journalists, but soon consumers will expecting this to be exposed on social media platforms and those they follow on social media. There really are genuinely valuable things happening in this space.

[samatman]

This will always be spoofable by projecting the AI content onto the sensor and playing it to the microphone. Which will give the spurious content a veneer of authenticity, this is within reach of a talented malicious amateur, and would be trivial for nationa-state actors to do at scale.

[lampiaio]

Thank you for pointing that out, I want to reply to everyone here but I don't think I have it in me to fight this battle. It seems my initial message of "have we questioned ourselves what we'll do should the countermeasures fail?" fell on deaf ears. I asked a very simple question: "what will we do / should we do when faced with a world in which no content can be trusted as true", and most replies just went on to list the countermeasures being worked on. I will follow my own advice and simply accept that is how the band plays.

[cush]

Of course. I don’t think anyone is going to be arguing that content captured by these cameras is real, it’s that the content is captured by the owner of that specific camera. There always needs to be some aspect of trust, and the value comes in connecting that with a trusted identity. Eg one couldn’t embed the CSPAN watermarks from a non-CSPAN camera.

[dorkwood]

> All I see is a meteor approaching and everyone trying to divert it, but no one actually preparing for when it does hit. Each day that passes I'm more certain that we will we look at each other like fools, asking ourselves "why didn't we focus on preparing for change, instead of trying to prevent change"?

I don't know, we've done a pretty good job at preventing nuclear war so far. We didn't just say "oh well, the genie is out of the bottle now. Everyone will have nuclear weapons soon and there's nothing we can do about it. All wars from now on are going to be nuclear. Might as well start preparing for nuclear winter." We signed treaties and made laws and used force to prevent us all from killing each other.

[refulgentis]

Forgive me on an initial reading, it is hard to have a nuanced discussion on this stuff without coming off like an uncaring caricature of one of two stereotypes, or look like you're attacking your interlocutor. When I'm writing these out, it's free association like I'm writing a diary entry, not as a critique of your well-reasoned and 100% accurate take.

Personal thoughts:

- we're already a year past the point where it was widely known you can generate whatever you want, and get it to a reasonable "real" threshold with less than a day worth of work.

- the impact is likely to be significantly muted, rather than an exponential increase upon, a 2020 baseline. professionals were capable of accomplishing this with a couple orders of magnitude more manual work for at least a decade.

- in general, we've suffered more societally from histrionics/over-reactions to being bombarded with the same messaging

- it thus should end up being _net good_, in that a skeptic has a 100% accurate argument for requiring more explanation than "wow look at this!"

- I expect that being able to justify / source / explain things will gain significant value relative to scaled up distributors giving out media someone else gave them without any review.

- something I've noticed the last couple years is people __hate__ looking stupid. __Hate__. They learn extremely quickly to refactor knowledge they think they have once confronted in public, even by the outgroup, as long as theyre a non-extremist.

After writing that out, I guess my tl;Dr as of this moment and mood, is there will be negligible negative effects, we already reached a nadir of unquestioned BS sometime between 2010 and 2024, and a baseline be _anyone_ can easily BS will lead to wide acceptance of skeptical reactions, even within ingroups.

God I hope I'm right.

[lampiaio]

I like the outlook you build through your observations, and I acknowledge the possible conclusion you arrive at as plausible. I do, however, put a heavier weight on your first point because I see what we have today in terms of image/video generation as very rudimentary compared to what we'll have in a couple years. A day's worth of work for a 100% convincing, AI-generated video immune to the most advanced forensics? We'll soon have it instantaneously.

Thank you for the preface you wrote, I completely understand your point of how easy it is to sound like a contrarian online, I'm sure my writing style doesn't help much on that front I'm afraid to admit.

[alex_duf]

What would you do then if you could prepare for a world where it's already here? Where the asteroid already hit, to use your own metaphor.

[skobes]

What sort of preparations do you recommend?

[lampiaio]

There's a saying in my local language that people usually say to someone who's going through a breakup or going through an unfair situation:

"Accept it, it hurts less".

I'm not saying it makes the actual situation any better; it obviously doesn't. But anyone can feel the rarefied AI panic in the air growing thicker by the minute, and panic will only make the situation worse both before and after absolute change takes place.

When we don't accept incoming change before it arrives, we surely are forced to accept it after it arrives, at a much higher price.

You asked about preparations: prepare yourself to see governments try (and fail) to regulate what processing power can be acquired by consumers. Prepare yourself for the serious proposal of "truth-checking agencies" with certified signatures that ensure "this content had its chain of custody verified as authentic from its CMOS capture up to its encoded video stream", in which a lot of time and effort will be wasted (there's already people replying about this, saying metadata and/or encryption will come to the rescue via private/public keys. Supposedly no will would ever film a screen!).

The above might seem an exaggeration, but ask yourself: the YouTube guidelines this post is about, the recent EU regulation... do you think those are enough? Of course they're not. They will keep trying to solve the problem from the wrong end until they are (we are) forced to accept there's nothing that can be done about it, and that it is us who need to adapt to live in such a world.

Enjoy the ride, I suppose.