4 ms·

> This is sometimes called the AI singularity or a superintelligence explosion. Nitpicking but an intelligence explosion due to an AI able to enhance itself is

> This is sometimes called the AI singularity or a superintelligence explosion. Nitpicking but an intelligence explosion due to an AI able to enhance itself is called the "technological singularity": https://en.wikipedia.org/wiki/Technological_singularity https://en.wikipedia.org/wiki/Technological_singularity

GalenErso 3y ago
Who f-ing cares if we call it a "superintelligence explosion", the singularity, or Roko's Basilisk? Everyone is trying so hard to coin the new concept that will enshrine their name through the ages that we're losing sight of the forest for the trees.
- TacticalCoder 3y ago
  It is definitely not a new concept. Vernor Vinge came up with the term in the eighties or early nineties. The Wikipedia page I linked to has more than a hundred references.
  - tim333 3y ago
    Actually John von Neumann in the 1950s.
m3kw9 3y ago
It won’t explode most likely because of hardware limitations, unless you think this is purely a software issue but last 10 years has proven otherwise
- ctoth 3y ago
  It's hard not to just want the damn AI to take over already when you read confident comments like OP, but the truth is: > The new MIT study adds to a growing body of evidence that the size of algorithms matters less than their architectural complexity. For example, earlier this month, a team of Google researchers published a study claiming that a model much smaller than GPT-3 — fine-tuned language net (FLAN) — bests GPT-3 by a large margin on a number of challenging benchmarks. And in a 2020 survey, OpenAI found that since 2012, the amount of compute needed to train an AI model to the same performance on classifying images in a popular benchmark, ImageNet, has been decreasing by a factor of two every 16 months. [0]: Improved algorithms may be more important for AI performance than faster hardware: https://venturebeat.com/ai/improved-algorithms-may-be-more-important-for-ai-performance-than-faster-hardware/ https://venturebeat.com/ai/improved-algorithms-may-be-more-i...
  - m3kw9 3y ago
    These are still very 1 dimensional type of intelligence comparatively to super intelligence we fantasize about, I’m confident because I see the probability ASI on current hardware isn’t there. Just look how much we understand our own brains and you think current hardware can best that by just algorithm isn’t a good bet.

GalenErso 3y ago

Who f-ing cares if we call it a "superintelligence explosion", the singularity, or Roko's Basilisk? Everyone is trying so hard to coin the new concept that will enshrine their name through the ages that we're losing sight of the forest for the trees.

TacticalCoder 3y ago
It is definitely not a new concept. Vernor Vinge came up with the term in the eighties or early nineties. The Wikipedia page I linked to has more than a hundred references.
- tim333 3y ago
  Actually John von Neumann in the 1950s.

m3kw9 3y ago

It won’t explode most likely because of hardware limitations, unless you think this is purely a software issue but last 10 years has proven otherwise

ctoth 3y ago
It's hard not to just want the damn AI to take over already when you read confident comments like OP, but the truth is: > The new MIT study adds to a growing body of evidence that the size of algorithms matters less than their architectural complexity. For example, earlier this month, a team of Google researchers published a study claiming that a model much smaller than GPT-3 — fine-tuned language net (FLAN) — bests GPT-3 by a large margin on a number of challenging benchmarks. And in a 2020 survey, OpenAI found that since 2012, the amount of compute needed to train an AI model to the same performance on classifying images in a popular benchmark, ImageNet, has been decreasing by a factor of two every 16 months. [0]: Improved algorithms may be more important for AI performance than faster hardware: https://venturebeat.com/ai/improved-algorithms-may-be-more-important-for-ai-performance-than-faster-hardware/ https://venturebeat.com/ai/improved-algorithms-may-be-more-i...
- m3kw9 3y ago
  These are still very 1 dimensional type of intelligence comparatively to super intelligence we fantasize about, I’m confident because I see the probability ASI on current hardware isn’t there. Just look how much we understand our own brains and you think current hardware can best that by just algorithm isn’t a good bet.