Can Descartes can help navigate Deepfakes? The new case for radical skepticism
With Open AI's newest text-to-video Sora, image manipulation, voice cloning and AI content mills there is reason to distrust everything we see and hear online.
A couple of weeks ago Open AI released its newest generative AI model Sora. The program takes text prompts and turns them into realistic video - the output is nothing short of incredible. Already it has seen ripple effects across the film industry, with director Tyler Perry withdrawing plans for an $800 million investment into building a new sound stage. Why would filmmakers buildings, stages, sets and people when they can create viable background footage just by typing a sentence. This kind of new AI technology could replace the need for background or stock footage altogether, with filmmakers potentially being able to conjure up any scene they wanted to with the type of a few prompts.
Aside from the business applications and concerns for jobs, this technology rings new alarm bells for the make up of truth on the internet as a whole. Already, the reliability of the information landscape online is under threat, with AI created blogs, news articles, fake photos of people’s faces, and synthesised voice clones. However, the realism and simplicity of use of Sora is going to be a new frontier once it is released to the public.
Imagine the potential of creating a realistic video of anything at any time, simply by typing a sentence. Open AI are putting some safeguards in place, their videos are currently watermarked, and they are 'red teaming' the program for safety. However, once a technology is in the wild, safeguards can only last so long. It wasn't long before ChatGPT itself was cloned into FraudGPT or WormGPT.
The Case for Radical Scepticism
AI is causing this very distrust in belief in the world around us, our access to knowledge of the way the world is, is being called into question. One way around this is to follow the idea of radical scepticism. Radical scepticism is the argument that it is not possible to know anything ever. This kind of argument is possibly best known as being one which the 17th Century French Philosopher Rene Descartes put forward.
Descartes wanted to discover what he could, without a doubt, know to be true. He finally settled upon the often quoted "I think therefore I am" (cogito ergo sum). This informs a branch of philosophy known as epistemology, or the study of how we know that something is true.
Descartes starts in the most sceptical place he can. He gets us to think that everything we see around us is created by this kind of evil demon, bent on deceiving us. The sky, the air, the earth are all created just to trick us into believing they are real. So, the solution is to begin to doubt everything. On Descartes thinking even if we can't know what is true, by doubting everything, we at least won't be affected by things that are fake. It is better to not believe anything, rather than be tricked once.
If we start with this radical scepticism when we look online, then we won't be duped by deep fakes or misinformation. We should note this isn't doubting just for the sake of it, but it is to reach a kind of certainty about what is real.1
So Descartes calls us to question our senses, as with Sora, or image generators like Midjourney - we are not able to trust what we see with our eyes. AI voice cloning technology causes us to doubt what we hear. We can position AI is the "evil deceiver" causing us to err even when we are reasoning. (Note I don't mean to call AI evil in this case, but only using the terminology of Descartes' thought experiment)
To combat this kind of falsity, we have to withdraw the mind from the senses. Online, this means not blindly trusting what we are seeing or hearing. Anything we view online - be it reading, seeing, watching - can no longer constitute knowledge. There is no guarantee that some part of it wasn't synthetically made.
Are we just virtual Brains in a Vat?
As mentioned, AI is causing distrust in belief in the world around us. If I watch a video on Instagram of puppies playing in the snow, can I be sure those puppies exist?
One problem is that, if the technology is good enough there seems nothing different in me watching a fake video which I believe to be true and watching a real video which I also believe to be true. Does my experience change if the video of the puppies turns out to be faked? What if I never find out it was fake, and go on the rest of my life believing the video to be real? Is my experience any different than if the video were real?
What we have above is similar to a thought experiment known as the "brains in the vat". Suppose there was a brain in a vat which was hooked up to electrodes and fed experiences of the real world. The brain would believe what they were experiencing to be true, and have no way of knowing or proving otherwise. This is also, more popularly, the subplot to the Matrix (spoilers).
One point of difference which philosophers point out is that, a real video gives me more of an insight into the way the world really is. However from my experience of just watching the videos I have no way of knowing that I am learning about the world as it really is or not.
Does distrust lead to conspiracy theories?
There is a downside to radical scepticism or distrust online, the proliferation of fake content will cause people to believe that real information is fake.
Last week the Associated Press issued a take down of a photo of Kate Middleton, the Princess of Wales and her children. The photo, which was purported to have been taken by prince William, was found by internet detectives to have been manipulated in some minor ways. Changing the composition of an photo like this are against APs strict policies on photographic standards. These standards are in place, because the AP has a great deal to lose in terms of reputation and credibility if they are seen to be endorsing false images.
However the manipulated image (which is probably just a bad photoshop job) only served to fuel conspiracy theories around Kate Middleton. The princess had not appeared in public for a number of weeks after recovering from surgery, and the edits to the picture only fuelled speculation that there was something else really going on.
Conspiracy theories such as this are the price we pay for radical sceptisism. When people are convinced that they are being lied to or mistrust information from places of authority, they will create new narratives to make the pieces fit. People begin to doubt genuine video from legitimate sources. This is also the cost for not guaranteeing authenticity.
Is true knowledge dead?
Descartes' thought experiment causes us to shift perspective. Instead of beginning from the position of proving something to be a fake, we should instead demand justification that it is true. How we go about guaranteeing authenticity is another matter, will water marks or digital signatures be enough to earn the trust of the public. Or has AI heralded the death of knowledge, the death of epistemology for good.
On Descartes account, if we strip away everything from the world, distrust everything, then all we are left with are our thoughts. After all, if we didn't have thoughts, how could we be able to think enough to doubt everything. There wouldn't be anything to think those thoughts! So if I can think, then I must exist in some way. I think, therefore I am.