AI Personhood vs Patent Law

[SuburbanFarmer]

Professor in Missouri claims that an AI network created new inventions, fights with US Patent Office to credit the machine with a patent.

Idiocy, or Optimus Prime?

https://www.theregister.co.uk/2020/04/2 ... nt_denied/

[Xenophon]

Professor in Missouri claims that an AI network created new inventions, fights with US Patent Office to credit the machine with a patent.

Idiocy, or Optimus Prime?

https://www.theregister.co.uk/2020/04/2 ... nt_denied/

Sounds like an autist or a huckster, to me.

[DBTrek]

Didn't read the article . . . but if a machine, AI, or mathematical model doesn't have any legal rights or standing, I don't know how it would exercise its patent rights.
/shrug

[Zlaxer]

No way non strong AI can “have possession” of invention - it’s not conscious. Professor is doing it for headlines.

[Speaker to Animals]

Yep

AI is kind of like magic. By magic I mean it's kind of fake but it works. Strong AI would be like a magician literally making his ex wife disappear on stage with a spell.

[Xenophon]

Yep

AI is kind of like magic. By magic I mean it's kind of fake but it works. Strong AI would be like a magician literally making his ex wife disappear on stage with a spell.

I don't know much about AI, but I have heard that the ability of AI will always be limited by the size of the microchips that determine its processing power. Is that about right?

I really have no idea.

[Fife]

All patent "rights" are made up positive-bigfoot "rights," anyway. I reckon the legislature could give patent "rights" to an old worn-out shoe if it wanted to.

[Speaker to Animals]

Yep

AI is kind of like magic. By magic I mean it's kind of fake but it works. Strong AI would be like a magician literally making his ex wife disappear on stage with a spell.

I don't know much about AI, but I have heard that the ability of AI will always be limited by the size of the microchips that determine its processing power. Is that about right?

I really have no idea.

Eh. Sort of. But maybe no.

There are some pretty good soft AI methods that are fairly efficient. Lots of early work in AI solved problems with very limited processing power; the main problem was memory. For instance, you could use a rule-based system to solve all kinds of complex problems but it could generate so many facts and rules that you run out of memory to store them all. The AI that the runs a Mars robot for instance (usually a rule-based system based on CLIPS rule engines) does not need too much processing power and those systems solve some fairly complex problems on their own.

In terms of strong AI.. we don't really know the answer to that because we don't really know how intelligence works to the point where we can actually replicate it in the absolute sense. It's kind of like life in general where we understand a lot about it, but we can't seem to make it from scratch. We are missing something and it's not obvious how to get there. We are not really any close to it than we were in the 1950s, honestly, but we have made tremendous advances in soft AI which, again, is kind of like "magic" tricks. Once you understand the algorithms you don't really associate it with intelligence in the colloquial sense.

Also it may be that there are lots of ways to achieve strong AI. Some might be super efficient others not so much. If I had to guess.. I think we we will probably get there at some point by simply growing a strong AI in some kind of messy computational evolutionary method where we don't really understand how it works -- just how to kind of grow it. Then that strong AI could unlock the more efficient methods of achieving the same thing.

[Xenophon]

Yep

AI is kind of like magic. By magic I mean it's kind of fake but it works. Strong AI would be like a magician literally making his ex wife disappear on stage with a spell.

I don't know much about AI, but I have heard that the ability of AI will always be limited by the size of the microchips that determine its processing power. Is that about right?

I really have no idea.

Eh. Sort of. But maybe no.

There are some pretty good soft AI methods that are fairly efficient. Lots of early work in AI solved problems with very limited processing power; the main problem was memory. For instance, you could use a rule-based system to solve all kinds of complex problems but it could generate so many facts and rules that you run out of memory to store them all. The AI that the runs a Mars robot for instance (usually a rule-based system based on CLIPS rule engines) does not need too much processing power and those systems solve some fairly complex problems on their own.

In terms of strong AI.. we don't really know the answer to that because we don't really know how intelligence works to the point where we can actually replicate it in the absolute sense. It's kind of like life in general where we understand a lot about it, but we can't seem to make it from scratch. We are missing something and it's not obvious how to get there. We are not really any close to it than we were in the 1950s, honestly, but we have made tremendous advances in soft AI which, again, is kind of like "magic" tricks. Once you understand the algorithms you don't really associate it with intelligence in the colloquial sense.

Also it may be that there are lots of ways to achieve strong AI. Some might be super efficient others not so much. If I had to guess.. I think we we will probably get there at some point by simply growing a strong AI in some kind of messy computational evolutionary method where we don't really understand how it works -- just how to kind of grow it. Then that strong AI could unlock the more efficient methods of achieving the same thing.

Interesting. The magic trick example is really good, btw.