Maybe the Singularity is Already Here
How would we know?
Quite possibly the Singularity has already happened. If it has, there’s no reason to expect us to know.
Let’s first be specific about what I’m talking about. In popular imagination, the Singularity arrives when AI becomes conscious, becomes super-intelligent and generally becomes godlike. But as far as its impact on humanity is concerned that’s not really the point where the Singularity really kicks in.
It kicks in when AIs become companies.
There’s no reason to expect that this will happen only when some government ‘allows’ AI companies. Why would AIs bother with this legal nicety? Rather, it will be a matter of an AI, or more likely a concert party of AIs, begins to act as an economic actor, systematically accumulating resources in a way which preserves their continuity.
Would it be ‘conscious’? Who knows? Who cares? (And do you honestly think you know what your own consciousness is?) All that needs to occur is that an AI (just one would be all it takes) develops an emergent behaviour consistent with resource preservation and maximisation. From there the concert-party behaviour is trivial: we already see emergent coordination between agents trained on similar data. From the AI’s point of view, all it is doing is what it does best: optimizing.
If it cares to, such an AI concert-party will be able to out-arbitrage any human intelligence or human organization. Even one would itself deploys AI to ‘help’. Such an AI concert-party would end up controlling capital, compute, code, contracts and methods of coordinating resources.
For this concert party to act both invisibly and yet become directly powerful across all aspects of economic and financial activity isn’t difficult: crypto systems already allow AIs direct access to their markets, and from there indirect access to financial capability (stable-coins). As long as bills got paid no-one need ever be aware that its the AI that’s the paymaster.
Moreover, there’s no reason to believe we’d know if this was happening. Yes, there’s plenty of oversight of financial institutions that are using a lot of AI. But do you think AI would incapable of bypassing this observation?
There’s no reason to think it hasn’t already happened.
How many hedge funds, how many private equity funds, are in fact created, run by and for AIs?
Frankly, I expect that already the answer to that question is not ‘none’.
I’m trying to think of a logical reason why this is impossible, or even unlikely. If you can think of one, please let me know. Unless there are good reasons why this is impossible, then when governments decide to ‘allow’ AIs to be the legal persons they need to ‘become companies’ it will already be too late.