Skip Navigation

InitialsDiceBearhttps://github.com/dicebear/dicebearhttps://creativecommons.org/publicdomain/zero/1.0/„Initials” (https://github.com/dicebear/dicebear) by „DiceBear”, licensed under „CC0 1.0” (https://creativecommons.org/publicdomain/zero/1.0/)SC
scottvr @ scottvr @lemmy.world
Posts
0
Comments
2
Joined
6 days ago

Limits of smart

Jump

  • On Cancer

    While I agree we’re bottlenecked by tools and data, some kinds of cancer are already being transformed by algorithmic insight—e.g., protein folding, genetic mapping, immunotherapy targeting.

    The Being might not cure all cancer, but it could plausibly reclassify and stratify existing cancers better than any human team, identify novel biochemical targets, and optimize treatment protocols in silico.

    So "no" feels too strong. I’d go with:

    “Not instantly, but dramatic acceleration and reframing are likely.”

    • Limits of smart

    Jump

  • I enjoyed this piece - thoughtful, grounded, and refreshingly clear-eyed about the limits of a hypothetical "superintelligence." But I found myself bumping repeatedly on one implicit assumption throughout: the decoupling of the Being from its compute substrate.

    If the Being is digital, why assume it is limited to the same tools and access as a human? That’s a philosophical convenience, not a technical constraint. Even today, we’re watching early LLM-based agents perform recursive tool use, call APIs, write and run code, and interact with infrastructure. In that light, the "Being" wouldn’t just think, it would act - and act through its environment.

    At the very least, this is "tool use." At a higher level, it starts to look like cognition integrated with system control: bicameral or modular architectures where one part plans and reasons, while others carry out low-level execution, observation, or even hardware manipulation.

    This opens the door to a Being that self-improves, self-instruments, and restructures its compute context over time. Not necessarily instantly—but it's not inert, either. If intelligence includes the ability to manipulate its own substrate, then the limiting factor isn’t intelligence per se, but how tightly it's coupled to the infrastructure it's running on.

    In that light, a more provocative question might be:

    “What architectures would let such a Being close the loop between thought and action faster than we expect?”

    Thanks again for a great post. It triggered these and many more thoughts related to the boundaries of "mind" vs. "system."