But provably safe infrastructure could stop this kind of attack at every stage: biochemical design AI would not synthesize designs unless they were provably safe for humans, data center GPUs would not execute AI programs unless they were certified safe, chip manufacturing plants would not sell GPUs without provable safety checks, DNA synthesis machines would not operate without a proof of safety, drone control systems would not allow drones to fly without proofs of safety, and armies of persuasive bots would not be able to manipulate media without proof of humanness.
Let’s say that you have a decider which can look at some complex real-world system and determine whether it is possible to prove that the complex real-world system has some desirable safety properties.
Let’s further say that your decider is not simply a rock with the word “NO” written on it.
Concretely, we can look at the example of “armies of persuasive bots would not be able to manipulate media without proof of humanness”. In order to do this, we need to have an adversarially robust classifier for “content we can digitally prove was generated by a specific real human” vs “content we can’t digitally prove was generated by a specific real human”.
But that also gets you, at a minimum, a solid leg up in all of the following business areas
Robust remote ID verification (id.me has a $1.5B valuation)
Provably bot-free online casino (billions in potential revenue)
So if you think this problem is solvable, not only can you make a potentially large positive impact on the future of humanity, you can also get very very rich while doing it.
You don’t even need to solve the whole problem. With a solid demonstration of a provable humanness detector, you should be able to get arbitrarily large amounts of venture funding to make your system into a reality.
The first step of creating a working prototype is left as an exercise for the reader.
Let’s say that you have a decider which can look at some complex real-world system and determine whether it is possible to prove that the complex real-world system has some desirable safety properties.
Let’s further say that your decider is not simply a rock with the word “NO” written on it.
Concretely, we can look at the example of “armies of persuasive bots would not be able to manipulate media without proof of humanness”. In order to do this, we need to have an adversarially robust classifier for “content we can digitally prove was generated by a specific real human” vs “content we can’t digitally prove was generated by a specific real human”.
But that also gets you, at a minimum, a solid leg up in all of the following business areas
Robust remote ID verification (id.me has a $1.5B valuation)
The ability to prove that advertisements are watched by a real human, eliminating ad fraud (one estimate I’ve seen says [$125B lost annually to ad fraud(https://www.anura.io/blog/ad-fraud-cost-advertisers-125-billion-in-2023#:~:text=Based on the data we,mostly focuses on U.S. losses.))
Provably bot-free online casino (billions in potential revenue)
So if you think this problem is solvable, not only can you make a potentially large positive impact on the future of humanity, you can also get very very rich while doing it.
You don’t even need to solve the whole problem. With a solid demonstration of a provable humanness detector, you should be able to get arbitrarily large amounts of venture funding to make your system into a reality.
The first step of creating a working prototype is left as an exercise for the reader.