Avoiding dangerous super AI by regulating meta level agency | ChatGPT and Tesla FSD examples

แชร์
ฝัง
  • เผยแพร่เมื่อ 30 ต.ค. 2024

ความคิดเห็น • 6

  • @jverart2106
    @jverart2106 ปีที่แล้ว +2

    I really enjoyed your analysis and perspective. As for the open-ended agency and computing power, I'm looking at Nvidia and other companies making huge leaps in hardware and processing power. They talk about incredibly massive growth, let alone the advancements on quantum research. Perhaps that computing power is soon available. The problem with the research of innovation on AI is that, differently from pathogens or nuclear energy, we don't see the consequences explicitly on humans as a physical danger and the mere curiosity of manufacturing a "virtual super being" is probably overwhelming for every company and country.

    • @Go-Meta
      @Go-Meta  ปีที่แล้ว +2

      Thanks! And, yeah, I agree, it's curiosity and prestige of doing it first that is the most likely reason people will keep on trying to build such an AI and, as you say, the consequences are not as obviously bad as, say, nuclear weapons. And I totally share the curiosity (it's why I studied AI decades ago!) but we see the same curiosity/risks/regulation setup in areas like human cloning. The longer term problem, as you say, is that if computer power continues to get smaller, faster, cheaper, then cost will no longer be a barrier to experimenting. So, yeah, I guess one day we will build entities with meta level agency. For now we should slow down (this aspect) and learn more.
      Thanks for the comment 👍

  • @caseyarmstrong7113
    @caseyarmstrong7113 7 หลายเดือนก่อน +1

    Great content. Another question I find interesting is is it even possible to create an intelligence that would want to achieve its own goals. Maybe without being a organic creature is very difficult to create that? I don't know. But it's something I always wondered.

    • @edhero4515
      @edhero4515 6 หลายเดือนก่อน

      An open, self-organising process, carbon-based or not, is on a continuous journey from the moment it begins, towards an ever more comprehensive vision of its goals. So the answer to your question would be: it is inevitable.

  • @m_sedziwoj
    @m_sedziwoj ปีที่แล้ว +2

    Yeah, maybe we should not design that system, but who don't want to be first who design it?
    This is why we need regulations, and with fast progress with neuromorphic computing, it may in short time become not so expensive.

    • @Go-Meta
      @Go-Meta  ปีที่แล้ว

      Yeah, we may not have long before building AI systems with meta level agency becomes relatively cheap to do. We therefore need to use this time wisely.