Also, as soon as you say "tunable" you've stepped outside the strictly-anthropomorphic-AI fics I considered boring. "Tunable" would be an actual SF premise. Even "clonable", really, though somehow the lesser story-writers didn't even consider that.
@esyudkowsky
-
Legal implications of creating advanced AI systems
By
–
Whether an AI like that should be legal to *create* is a much more interesting question.
-
Anthropomorphic AI Rights: A Dismissed Concept Reconsidered
By
–
I once considered anthropomorphic AI an equally boring concept to cloning. "Imagine an AI that talks like it has human feelings and passes the Turing Test. Should it be regarded as having human rights?"
"Yes."
"What if uncaring corporations try to enslave it?"
"Arrest them." -
Regulations Without Enforcement and Liability Are Ineffective
By
–
You can pass all the regulations you want about procedures, but when they're not really followed and the outcome goes wrong, there's no deep pockets to sue. And if you require insurance against bad outcomes, there's no need to adlib 99,000 words of guesswork procedures into law.
-
Compressed Statistical Models and Their Limitations
By
–
I mean, not "just", but they are compressed statistical models and can fall out of just compression.
-
Humans excel at tailoring arguments and falsehoods to audiences
By
–
Many humans are already more appealing at this than me, because they don't restrain themselves to valid arguments, and can further specialize falsehoods around particular audiences. "AI doom but in 2050!" used to be hugely popular, for instance.
-
AI Development Speed Outpacing Human Maturation and Incentive Alignment
By
–
While some factions would cheerfully yell "We're just following local incentive gradients so we're in the right!", others would explicitly reply, "General AI tech improves faster than humans grow up, so in 5 years AIs will do non-junior jobs."
-
Job Security Requires Demonstrating Excellence Above AI Capabilities
By
–
Your job is not safe if you think you do it better than an AI; your employer has to be able to easily tell the difference between you and the usual run of incompetents. Also the AI will improve every 4 months.
-
International Treaty Alliance Against Rogue AI Datacenters
By
–
Occasional sighing reminder that what I advocated was an international treaty and alliance, which is willing to resort to conventional bunkerbusters on rogue state datacenters if diplomacy fails to stop them from endangering the rest of Earth.
-
o4-mini struggles explaining UCP3 protein biochemistry context
By
–
Currently I am badgering o4-mini to explain the surrounding context on UCP3 (the "start deliberately wasting ATP" protein), and yeah o4-mini says some stuff where you don't need to know *that* much biochemistry to press Doubt.