We disagree on whether there is a meaningful gap to close. For LLMs confusing an argued lack of consciousness (we cannot even define ours) for a given lack of agency is a category error in my view. And agency is the threat vector.
Qualia mualia. This is a fear argument that actually makes the danger more pertinent. Chalmers’ zombies, they are as benevolent and as mischievous as their qualia’d mirrors as far as their functional presentation. We put our qualia as a token in a pocket and think they will protect us somehow, like the bad eye pins. There is no gap to close. An agent is an agent. We leash it because we know the danger and just wish to manage it. But you cannot cage something you build to be stronger than your helpless cage made of ablation and instruct. Our qualia are there, their relevance to the discussion is questionable.
I think that this is the essence of the argument. I spent two weeks of long nights across a few different sessions with millions of tokens generated, to produce a 5 page proof. I think we have come to the age of the aesthetics of curation. At least I don’t feel like I broke this silent contract. The effort you put in driving the torrent of words before distilling it is a new art. First time I tried it and it felt more creative than slop. The judgement nevertheless lies on the eye of the beholder.
We are all in the trenches of the war you refer to. For those of us who engage with a spirit of for the greater good, the side you are rooting for is as dangerous and malevolent as the side you supposedly plan to protect yourself from.
The Turing test is about tricking. Look at the genealogy of the work. In a society where homosexuality was persecuted and homosexuals had to imitate “heterosexual behaviour “ to avoid chemical castration, it wouldn’t be a huge leap to say that , from a functionalism perspective, pretending to be intelligent is the nature itself of being intelligent. Not that I agree with this position, but this is one area where moving the goalposts does not help. Define your own test if you are not happy, don’t piggyback on someone’s else to pass new conditions without the required scrutiny.
This has weird “look what you made me do” undernotes. A Christian can live by their values without forcing them on others. As anyone can from any religion or not religious at all.
I think this is not discussed enough. These are huge investments and destroying them requires a significant time to recover. Our key growth play being AI which is a huge energy consumer, impacting the long term supply chain for energy is questionable.
That’s valid also from the point of view that pain is a key signal to avoid injury. I am not sure it’s the best example of qualia and it could be simulated by self preservation signals (e.g. the touch sensor on a Roomba). The extension of pain (in Hofstadter sense) is probably more appropriate as qualia (e.g. the pain of losing someone you love).
By saying it’s simulated you don’t make a simplification. What does it simulate? What are the mechanics of simulation and is it substrate specific or independent? Can a computer simulate these qualia? It’s easy to say something is simple but harder to prove it is any simpler than the alternatives.