Conversation

Notices

1. Embed this notice
   Mici_Melanzana (mici_melanzana@hachyderm.io)'s status on Saturday, 27-Jan-2024 23:37:38 JST Mici_Melanzana

   • Thomas 🔭🕹️

   @thomasfuchs I agree but a lot of this depends on what you consider artificial intelligence, to some people a text model is AI, you could argue an algorithm is already a form of AI, but I guess what most people consider “real” AI is a computer that can think like a human. That’s already ironic if you ask me.

   • Embed this notice
     A Slightly Orange Cat (mbrailer@mstdn.social)'s status on Saturday, 27-Jan-2024 23:39:56 JST A Slightly Orange Cat

     • Thomas 🔭🕹️

     @thomasfuchs I'm assuming that AI, like all computer applications, is just a highly complex mathematical process.

     Faster computers would allow for more complex functions, but it would still be just a process and not true creativity.

   • Embed this notice
     Och Menno (ochmennopodcast@literatur.social)'s status on Saturday, 27-Jan-2024 23:47:27 JST Och Menno

     • Thomas 🔭🕹️

     @thomasfuchs Railguns .. everyone forgets that darpa spent years on Railguns …

   • Embed this notice
     Och Menno (ochmennopodcast@literatur.social)'s status on Saturday, 27-Jan-2024 23:57:40 JST Och Menno

     • Thomas 🔭🕹️

     @thomasfuchs yup

   • Embed this notice
     Och Menno (ochmennopodcast@literatur.social)'s status on Sunday, 28-Jan-2024 00:03:33 JST Och Menno

     • Thomas 🔭🕹️

     @thomasfuchs zumwalt class was designed to carry Railguns ^^ https://en.wikipedia.org/wiki/Zumwalt-class_destroyer

   • Embed this notice
     Raven667 (raven667@hachyderm.io)'s status on Sunday, 28-Jan-2024 00:17:20 JST Raven667

     • Thomas 🔭🕹️
     • A Slightly Orange Cat

     @thomasfuchs @mbrailer I think at some point you could achieve the complexity of a human brain, which is itself not made of magic but of complex physics processes that could be emulated and modeled, but only at _enormous_ expense with silicon. All of the computer neural networks in the world put together can do, what, a tiny fraction of what _anyone_ can do with a kilo of grey matter and less than 100 watts, their approach will _never_ achieve the power efficiency that biology has done.

   • Embed this notice
     Michael Kranz (mkranz@hachyderm.io)'s status on Sunday, 28-Jan-2024 00:28:32 JST Michael Kranz

     • Thomas 🔭🕹️

     @thomasfuchs isn’t there also evidence that Moore’s Law may be so predictable simply because of the expectations *it* puts on the industry and consumers rather than because of any “natural” feature of transistors or innovation. Not sure how that artifice would be transferable to anything with real limitations (even wrt transistors, as we’re now seeing)

   • Embed this notice
     Keith (keydelk@fosstodon.org)'s status on Sunday, 28-Jan-2024 00:39:37 JST Keith

     in reply to

     • Thomas 🔭🕹️
     • A Slightly Orange Cat
     • Raven667

     @raven667 @thomasfuchs @mbrailer it’s not just the complexity that matters, but how the system is structured that gives rise to real intelligence. My biggest frustration with this current wave of AI hype is that it is so divorced from any studies of actual human cognition.
     We take “neural networks” (a statistical algorithm for finding trends in large and messy data sets), throw tons of data on it, and pretend it’s “on the verge of becoming AGI.”

   • Embed this notice
     Michael Kranz (mkranz@hachyderm.io)'s status on Sunday, 28-Jan-2024 03:16:34 JST Michael Kranz

     • Thomas 🔭🕹️

     @thomasfuchs oh, no doubt. I think, in the years since he coined it, though, many have taken it to be some kind of force of nature that results in technological progress happening, as you put it, linearly or geometrically, when we that wouldn’t have necessarily happened (for a time, at least) for transistors without the self-fulfilling prophecy. I don’t think any of that is what Moore intended, though

   • Embed this notice
     botvolution (botvolution@mastodon.sdf.org)'s status on Sunday, 28-Jan-2024 20:27:50 JST botvolution

     • Thomas 🔭🕹️

     @thomasfuchs
     Nobody tell them that x tends to zero