/sci/ - Science & Math » Thread #11761766

44KiB, 866x438, 2-Figure1-1.png

View Same Google iqdb SauceNAO

Will automated proof solvers replace mathematicians?

Anonymous Thu 04 Jun 16:45:10 2020 No.11761766 View Reply Original Report

Quoted By: >>11762006 >>11762025 >>11762028

I thought math would be one of the last fields to fall to AI, but it seems that proof solvers will automate the field sooner than later.

Anonymous

Anonymous Thu 04 Jun 2020 17:57:56 No.11762006 Report

Quoted By: >>11762022

>>11761766
Proof solvers are pointless if they don't give human-readable proofs. It's trivial to make a proof solver that outputs an unending list of arbitrarily complex proofs, but without a human mind to comprehend these structures and guide the development of these subfields, it is pointless.

Anonymous

Anonymous Thu 04 Jun 2020 18:04:33 No.11762022 Report

Quoted By: >>11762026 >>11762035

>>11762006
i thought the point of statistical learning is that the computer does not need to understand human readbleness. you probably can't even define what the arbitrary threshold of complexity is, beyond which a proof is no longer human-readable. you just recognize it when you see it.

Anonymous

Anonymous Thu 04 Jun 2020 18:05:51 No.11762025 Report

Quoted By:

>>11761766
>>>/g/tfo retard

Anonymous

Anonymous Thu 04 Jun 2020 18:06:24 No.11762026 Report

Quoted By: >>11762057

>>11762022
Like I said, you can design a proof solver that generates valid proofs. That doesn't make those proofs useful.

Anonymous

Anonymous Thu 04 Jun 2020 18:07:18 No.11762028 Report

Quoted By: >>11762051 >>11762686

>>11761766
If P = NP then yes, otherwise no.

Anonymous

Anonymous Thu 04 Jun 2020 18:08:31 No.11762035 Report

Quoted By:

>>11762022
>i thought the point of statistical learning is that the computer does not need to understand human readbleness.
Yes, because statistical learning is used as an applied tool. Companies don't care WHY this is the set of ads you will optimally click on, they just care that you're clicking on them.

The point of math is understanding, not just the end result. It's not very helpful to shit out a gigantic list of theorems if nobody on earth has any clue why they're true.

Anonymous

Anonymous Thu 04 Jun 2020 18:14:00 No.11762051 Report

Quoted By:

>>11762028
>t. retarded cs major

Anonymous

Anonymous Thu 04 Jun 2020 18:15:50 No.11762057 Report

Quoted By: >>11762216

>>11762026
what do you even want out of proofs? is this something that can be clearly defined, or a vague notion that can only be expressed by means of word association?

Anonymous

Anonymous Thu 04 Jun 2020 18:23:53 No.11762080 Report

Quoted By: >>11762087

Technicians will be the last people to stop working, until the atlas robot can set up an extension ladder on a roof, climb up with it's tool bag, find the broken HVAC unit, troubleshoot it, and repair it, technicians will still be needed.

Anonymous

Anonymous Thu 04 Jun 2020 18:25:51 No.11762087 Report

Quoted By:

>>11762080
the boston dynamics cow can probably already do this

Anonymous

View Same Google iqdb SauceNAO Screenshot_20200604-150744.png, 271KiB, 1440x2560

Anonymous Thu 04 Jun 2020 19:11:11 No.11762216 Report

Quoted By:

>>11762057
>is this something that can be clearly defined
As in can it be formalized? Of course it can't. But clearly picrel is not advancing mathematics in any meaningful way. The purpose of mathematics is to discover and understand structures, not find truth values for arbitrary statements. Even if we could direct a proof solver to solve certain theorems we know in advance are important, this still would not reveal any deeper understanding to us and would leave us with no idea how to continue expanding the field. It would devolve math into a useless and effectively solved game of symbol pushing.

Anonymous

Anonymous Thu 04 Jun 2020 21:54:14 No.11762686 Report

Quoted By:

>>11762028
isn't that already a thing

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