Q&A: Prophecies of Epistemic Destruction – Artificial Intelligence
Prophecies of Epistemic Destruction – Artificial Intelligence
Question
Hello,
I wanted to ask what the Rabbi thinks about the implications of artificial intelligence for our rational understanding.
Today it is already fairly clear that the combination of deepfakes and artificial intelligence together will be able to trick us into experiencing anything. Add to that transhumanism (the integration of technologies into the human being), such as Elon Musk who founded Neuralink.
So what does the Rabbi think about this, once there is a positive doubt regarding the brain-in-a-vat problem? What will be the way to hold on to our beliefs another 20 years from now?
Answer
I don’t understand the question. You haven’t defined a situation, and you’re asking a general question. Every situation requires treatment in the way appropriate to it. Even today there is quite a bit of fake content, and we have to navigate within it. That is what we will also have to do in the future. I also don’t see why this is connected to the brain-in-a-vat question.
Discussion on Answer
Strange question. If the assumption is that it will be possible to produce a deepfake that cannot be distinguished from real information, then it cannot be distinguished from real information. So what is the question? The question is whether that indeed will be the situation (that is, that there will be no way to tell the difference). If it is, then it is. I don’t see what the question is. Do you want me to predict by what methods they will act to get around this? Prophecy is not one of my many virtues.
A prophet who prophesied in the name of God that next year a universal deceptive demon will arise, and a year passes and no demon is seen—has the prophecy been disproven, and will the prophet be punished as a false prophet, or can he claim: I truly prophesied, and the demon is deceiving you.
In other words: can a prophecy that cannot be falsified be verified, and is a prophecy that cannot be verified really a prophecy?
If the latter wording deals with a general prophecy addressed to the public, then it seems simpler, because it can be through subjective knowledge. A prophecy that cannot be falsified can be verified—for example, a prophecy that at least one person in the audience will suddenly feel a tingling in his spleen at a certain rhythm—by whispering the tingling, the prophecy will be confirmed. (And if everyone testifies that they felt nothing, it seems reasonable to me that we do not accept testimony that is conditional on everyone having spoken the truth.) And a prophecy that cannot be verified can be falsified—for example, a prophecy that no one in the audience will feel tingling—by whispering the tingling, the prophecy will be disproven. Is that really so?
And regarding the universal demon: plainly, this prophet is no different from anyone else who says, “True, it appears to you that I desecrated the Sabbath, but a demon is deceiving you.” But there is some tingling sensation here that when it comes to a prophet, the conclusion is being assumed from the outset. If in your opinion this really is somewhat different from anyone else who claims “a demon is deceiving you,” Perhaps you could clarify.. (I wrote here because it seems that the questioner asked about a universal deceptive demon, and when in your answer you referred to prophecy, it seemed that you deliberately chose a kind of prophecy that assumes the deception is limited.)
Interesting,
do you think that the scenario described by many futurists and researchers regarding deepfakes will in fact materialize?
The situation is this: in about 20 years, many AI researchers claim that reality will change.
There are claims that 10% of researchers think the results will be disastrous. (And some present it as 1%.) The concern is that at that time computers will exploit human beings, in the best case, and not the other way around. (For example, the paperclip scenario, where someone programs the computer to produce as many paperclips as possible, and in service of that it destroys everything that moves to make it happen, and if necessary exploits human beings too.)
In any case, there is another claim I was focusing on,
that at that time it will be possible to create deepfakes that cannot be distinguished from reality, just as today it is already possible to imitate people’s voices in a way that cannot be distinguished. In combination with transhumanism that communicates directly with consciousness.
If so, for a person who lives in that period and knows that the computer can deceive us in a way that cannot be distinguished,
will a person who sees anything be able to trust his own eyes? Because he has an excellent positive reason to fear that he should not.