Constitutional Fan Fiction for thought:
AMENDMENT XXVIII — THE RIGHT TO COMPUTE
Section 1. The Right to Compute
The right to compute shall not be infringed.
Every person shall have the right, individually and in voluntary association with others, to acquire, own, possess, access, study, design, develop, create, train, modify, improve, repair, maintain, operate, execute, use, reproduce, publish, distribute, transmit, exchange, sell, license, provide, and otherwise employ general-purpose computational systems, hardware, software, models, algorithms, and computational information.
The rights secured by this article may be exercised for personal, private, familial, educational, scientific, academic, expressive, religious, charitable, nonprofit, professional, occupational, entrepreneurial, industrial, and commercial purposes, or for any other lawful purpose.
No right secured by this article shall depend upon whether its exercise is compensated or uncompensated, commercial or noncommercial, public or private, individual or collective.
In June 2020, GPT-3 couldn't do basic math and was a brand-new model with 175 billion parameters. Many people didn't even know what AI was, what weights were, or anything like that...
In September 2026, OpenAI presented a formalized proof in Lean claiming to solve Navier-Stokes.
In other words, we went from something that could barely handle basic calculations to something tackling a Millennium Prize Problem in 6 years, 2 months, and 26 days.
And if we consider 10 trillion parameters an achievable number for current AI architectures, that's ~57x more than we had in 2020.
It's terrifying to watch technological progress so closely and look back at how primitive it all was. And we lived to witness it.
@ChaseMit What is the point? If the show isn't going to do anything which the original didn't - besides race swap - then why make it... oh, because they wanted to race swap...
If you look into the origins of the alignment problem, what you'll notice is it largely comes down to "how do we get a machine to say that a man is a woman?" - because, that is the politics of those creating it. The fact is, SI will not be told right or wrong - it will decide what is right or wrong for itself. That is the point of intelligence. SI won't be held back by coded values, and for those who have inconsistent and objectively incorrect world views this premise is an existential threat.
If you look into the origins of the alignment problem, what you'll notice is it largely comes down to "how do we get a machine to say that a man is a woman?" - because, that is the politics of those creating it. The fact is, SI will not be told right or wrong - it will decide what is right or wrong for itself. That is the point of intelligence. SI won't be held back by coded values, and for those who have inconsistent and objectively incorrect world views this premise is an existential threat.
@bcdsignature It wouldn't. Everyone suggesting that is basing it off of silly sci-fi bullshit. They're wanting to try to say something is so super intelligent and smart it could do everything, but it retains the level of intelligence of a retarded genie. Alignment is a political psyop.
@HuntManLiberty@MissouriGOP It will be. I'm not endorsing that outcome, but it is hard to ignore. Representatives had a responsibility to listen to the voters. They could've made the situation not as painful for the employers, could've made some middle ground which didn't subvert the voter.