3 ms·

Rather than going into these strange discussions about the scientific method, nondeterminism and philosophical debates I'm still scratching my head when languag

Rather than going into these strange discussions about the scientific method, nondeterminism and philosophical debates I'm still scratching my head when language became synonymous with AGI. Forget all this highbrow stuff, my still intact test is, can something walk into any random house on the street, plumb a toilet and brew a cup of coffee. Moravec was right in the 80s when he pointed out that the easy things are hard and the hard things are easy. Simulating human cognitive abilities primarily exists to satisfy human testers, it doesn't say much about general intelligence.

grandmczeb 3y ago
> can something walk into any random house on the street, plumb a toilet and brew a cup of coffee. Stephen Hawking would have failed this test.
- danielbln 3y ago
  I would fail that test, you'd end up with a pretty good coffee and a still broken toilet, yet I consider myself generally intelligent on most days.
- Barrin92 3y ago
  I'm not sure if that's a quip or a serious point but that is because Stephen Hawking in old age lost that capacity, not because he principally lacked it. ChatGpt et al did not walk around last week and are now suffering from robot ALS because there servers have too much dust on them, they just lack that modality altogether (like any other apart from generating text)
pixl97 3y ago
Language has become synonymous with intelligence, not only in AI, but just regular I. It is a means of transferring information from a source to a sink. If the sink is intelligent it will be able reconstitute the ideas that were transferred too it. We consider animals less intelligent because almost none have the ability to transfer intelligence via some other medium. You're also hitting on a problem with the word intelligence and the fact it's about all encompassing as the word universe. There is cellular intelligence encoded in RNA and DNA. There is multicellular intelligence between organs and different components. There multiple layers of the mind, one part for organs and constantly running system, one for locomotion, and another for abstract thinking. Trying to lump this as one thing makes it very difficult to understand the capabilities of a system. Also, you're not paying much attention to LLM enabled robots these days. For example some of Google's larger robot models can travel into a room they have not seen before and look in cabinets for an object they have not seen before and successfully find it. Not a good time to be in denial about evolving robotic capabilities.

grandmczeb 3y ago

> can something walk into any random house on the street, plumb a toilet and brew a cup of coffee. Stephen Hawking would have failed this test.

danielbln 3y ago
I would fail that test, you'd end up with a pretty good coffee and a still broken toilet, yet I consider myself generally intelligent on most days.
Barrin92 3y ago
I'm not sure if that's a quip or a serious point but that is because Stephen Hawking in old age lost that capacity, not because he principally lacked it. ChatGpt et al did not walk around last week and are now suffering from robot ALS because there servers have too much dust on them, they just lack that modality altogether (like any other apart from generating text)

pixl97 3y ago

Language has become synonymous with intelligence, not only in AI, but just regular I. It is a means of transferring information from a source to a sink. If the sink is intelligent it will be able reconstitute the ideas that were transferred too it. We consider animals less intelligent because almost none have the ability to transfer intelligence via some other medium. You're also hitting on a problem with the word intelligence and the fact it's about all encompassing as the word universe. There is cellular intelligence encoded in RNA and DNA. There is multicellular intelligence between organs and different components. There multiple layers of the mind, one part for organs and constantly running system, one for locomotion, and another for abstract thinking. Trying to lump this as one thing makes it very difficult to understand the capabilities of a system. Also, you're not paying much attention to LLM enabled robots these days. For example some of Google's larger robot models can travel into a room they have not seen before and look in cabinets for an object they have not seen before and successfully find it. Not a good time to be in denial about evolving robotic capabilities.