/sci/ - Science & Math » Thread #11769135

84KiB, 1200x800, 5729582.jpg

View Same Google iqdb SauceNAO

Anonymous Sat 06 Jun 23:25:47 2020 No.11769135 View Reply Original Report

Quoted By: >>11769168 >>11769190 >>11769218 >>11769224 >>11769339 >>11771385

Are we going to hit a roadblock with AI any time soon?

Like, I'm sure researchers will squeeze out so much of neural networks and what not (like they did with ipv4), but all the papers I read are quite similar in some sense.

I don't see many theoretical stuff, it's mostly all implementations.

Anonymous

Anonymous Sat 06 Jun 2020 23:38:58 No.11769168 Report

Quoted By:

>>11769135
Firstly, coffee isn't good for you. Secondly,
>any time soon?
That happened a few years ago.

Anonymous

Anonymous Sat 06 Jun 2020 23:48:23 No.11769190 Report

Quoted By:

>>11769135
I just want the cognitive linguists to tell me what learning is so my brain can learn what learning is then learn it's own learning until I become the singularity and kill myself.

Anonymous

Anonymous Sun 07 Jun 2020 00:05:38 No.11769218 Report

Quoted By:

>>11769135
Not for a while I'd say. Computational resources are still improving and becoming cheaper, with quantum computing set to become the next thing (hopefully), so even if we stop developing new machine learning techniques just throwing more compute at the problem will likely see AI continue to become more powerful.

Anonymous

Anonymous Sun 07 Jun 2020 00:10:02 No.11769224 Report

Quoted By:

>>11769135
Wit literally anything made by humans ever there are roughly 4 types of people. People who cant use it or understand it, people who can use it but cant understand it, people who can use
and understand it, and people who can improve it. Most people by far will be the first three. It would be strange if we used a process most people could improve, so it is no surprise that most papers are just trying to use it in some sort of fashion. For a lot of people, this is their limit.

Anonymous

Anonymous Sun 07 Jun 2020 01:01:36 No.11769339 Report

Quoted By:

>>11769135
And that's why someone will end up finding a better, more efficient alternative to neural networks. Just like how they found ipv6 after they exhausted ipv4.

Anonymous

Anonymous Sun 07 Jun 2020 18:54:58 No.11771385 Report

Quoted By:

>>11769135
AI is a lot more than ANN research. For example, there is some very exciting stuff happening in the field of machine learning infrastructure (take a look at openmined). You also can't disregard symbolic AI developments.

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches