/sci/ - Science & Math » Thread #11923399

73KiB, 1280x720, Bd4OIA9CEAAeXmC.jpg

View Same Google iqdb SauceNAO

Anonymous Tue 21 Jul 18:18:26 2020 No.11923399 View Reply Original Report

Quoted By: >>11923403 >>11923405 >>11923418 >>11923425 >>11923525 >>11923677 >>11923940 >>11924810 >>11925283 >>11925384 >>11925580 >>11925602 >>11925804 >>11926400

>2020
Why isn't artificial general intelligence available yet? What's missing?

Anonymous

Anonymous Tue 21 Jul 2020 18:20:14 No.11923403 Report

Quoted By:

>>11923399
>What's missing?
The genocide of all animefags

Anonymous

Anonymous Tue 21 Jul 2020 18:20:55 No.11923405 Report

Quoted By:

>>11923399

We don't know how the brain functions

Anonymous

Anonymous Tue 21 Jul 2020 18:25:32 No.11923418 Report

Quoted By: >>11923963 >>11925572

>>11923399
Because we are no closer to true AI than we were decades ago. All the AI you see mentioned in the media is pure marketing hype. It's all simply Machine Learning, a bunch of algorithms to process and analyse large amounts of data. It's not intelligence or thought, it's quicksort()

Anonymous

View Same Google iqdb SauceNAO 1590855459219.jpg, 134KiB, 1242x1236

Anonymous Tue 21 Jul 2020 18:27:42 No.11923425 Report

Quoted By: >>11923921

>>11923399
>What's missing?
Real anime waifus to encourage the AGI researchers to work hard and do their best.

Anonymous

Anonymous Tue 21 Jul 2020 18:29:37 No.11923436 Report

Quoted By:

AI is millions of line of code in Python. An AI will require scripts for everything they do, they will never be conscious or have free will, as they are dependant of human coding.

Anonymous

Anonymous Tue 21 Jul 2020 19:06:58 No.11923525 Report

Quoted By: >>11923638

>>11923399
Computers are better than humans at logic. But how about creativity and expression? That's where they fall flat

Anonymous

Anonymous Tue 21 Jul 2020 19:33:53 No.11923638 Report

Quoted By: >>11925580

>>11923525
define creativity

Anonymous

Anonymous Tue 21 Jul 2020 19:43:28 No.11923677 Report

Quoted By:

>>11923399
Neurologists and programmers do not like eachother.

Anonymous

View Same Google iqdb SauceNAO 1595024109077.jpg, 164KiB, 901x1200

Anonymous Tue 21 Jul 2020 20:37:55 No.11923910 Report

Quoted By: >>11923917 >>11924813

it doesn't have instincts
give it a prime directive + reproduction and death if if doesn't meet it's goals

Anonymous

Anonymous Tue 21 Jul 2020 20:40:52 No.11923917 Report

Quoted By: >>11923932 >>11924792

>>11923910
goal: convince thot with text to send nudes
thot nudes == food
not getting nudes after a week == death
reproduction every 2 weeks

Anonymous

View Same Google iqdb SauceNAO 1592237333654.gif, 486KiB, 960x720

Anonymous Tue 21 Jul 2020 20:42:26 No.11923921 Report

Quoted By: >>11925773 >>11926387

>>11923425
Man. That hits home.

Idk, maybe we could make a bot that uses a very large neural network and VR chat with a fail criteria being set for being caught out as a bot? But really we just need someone willing to lend us a very big computer. I heard some guy running a Minecraft PvP not needed like 3k in Microsoft azure just to make it go.

Anonymous

Anonymous Tue 21 Jul 2020 20:46:20 No.11923932 Report

Quoted By: >>11923957

>>11923917
That might work if you can do a whole brain emulation of a thot. But more idealistically, if you maybe made a bot designed to "catfish" men on tinder and use date offers * conversation length as your formula you might get somewhere

Anonymous

Anonymous Tue 21 Jul 2020 20:47:46 No.11923940 Report

Quoted By: >>11926399

>>11923399
It's almost there and called gpt-3 http://lacker.io/ai/2020/07/06/giving-gpt-3-a-turing-test.html

Anonymous

Anonymous Tue 21 Jul 2020 20:51:40 No.11923957 Report

Quoted By: >>11923974

>>11923932
you can Chadfish women on tinder

Anonymous

View Same Google iqdb SauceNAO (You).jpg, 74KiB, 742x745

Anonymous Tue 21 Jul 2020 20:53:19 No.11923963 Report

Quoted By:

>>11923418
>Because we are no closer to true AI than we were decades ago. All the AI you see mentioned in the media is pure marketing hype. It's all simply Machine Learning, a bunch of algorithms to process and analyse large amounts of data. It's not intelligence or thought, it's quicksort()

Anonymous

View Same Google iqdb SauceNAO hqdefault.jpg, 8KiB, 480x360

Anonymous Tue 21 Jul 2020 20:56:25 No.11923974 Report

Quoted By:

>>11923957
Bruh.

My mind is fucking blown.

Somebody help this lad! We have the technology, we can get free nudes on an industrial scale!

Anonymous

Anonymous Wed 22 Jul 2020 02:21:11 No.11924792 Report

Quoted By: >>11926397

>>11923917
>reproduction every 2 weeks
Have fun training it for thousands of years before getting any results.

Anonymous

Anonymous Wed 22 Jul 2020 02:29:17 No.11924810 Report

Quoted By:

>>11923399
Algorithmic computing is a dead end. We need new hardware.

Anonymous

Anonymous Wed 22 Jul 2020 02:31:30 No.11924813 Report

Quoted By:

>>11923910
how does a machine learn how to survive? It would have to go through some sort of long evolutionary process as humans did.

Anonymous

Anonymous Wed 22 Jul 2020 06:26:18 No.11925283 Report

Quoted By:

>>11923399
>input module
stream of data from sound and vision
>processing module
pattern recognition on data streams
>amalgamator module
connect patterns across data streams
>output modules
audio generator
>loop module
positive and negative feedback loops
>memory module
layered smart memory storage

Anonymous

Anonymous Wed 22 Jul 2020 07:48:07 No.11925384 Report

Quoted By:

>>11923399
>What's missing?
We don't know

Anonymous

Anonymous Wed 22 Jul 2020 09:47:15 No.11925572 Report

Quoted By:

>>11923418
I have never seen a more intentionally misleading post in my life.
We are absolutely closer to AGI now compared to 10 years ago. OpenAI and Deepmind are a testament to that.

Anonymous

Anonymous Wed 22 Jul 2020 09:51:51 No.11925580 Report

Quoted By:

>>11923399
Human intelligence
Why don't you help?
why do you want it?
>>11923638
Who knows
possibly some kind of meaningless abstraction

Anonymous

Anonymous Wed 22 Jul 2020 10:02:24 No.11925602 Report

Quoted By:

>>11923399
I honestly think we're very nearly there. Things missing right now are memory, better priming and another factor thousand in number of parameters. GPT-5/6/7 (who knows?) will probably be general AI in at least its ability to converse with humans. It will still struggle with topic that require in-depth analysis, but it'll be like talking to regular, if shallow, person.

Anonymous

Anonymous Wed 22 Jul 2020 11:55:34 No.11925773 Report

Quoted By:

>>11923921
Who's the guy in the picture?

Anonymous

Anonymous Wed 22 Jul 2020 12:15:09 No.11925804 Report

Quoted By:

>>11923399
We abandoned Lisp.

Anonymous

Anonymous Wed 22 Jul 2020 16:37:40 No.11926387 Report

Quoted By:

>>11923921
You are vastly underestimating the amount of training steps needed to train an agent like that with current techniques.
That's why agents are usually trained in simulated environments at many times faster than real time. Training in real environments would be unfeasibly slow.

In this case, training in a simulated environment is also not possible either unless "caught out as a bot" is more rigorously defined

Anonymous

Anonymous Wed 22 Jul 2020 16:40:06 No.11926397 Report

Quoted By:

>>11924792
This
Training on real environments is unfeasibly slow

Anonymous

Anonymous Wed 22 Jul 2020 16:41:02 No.11926399 Report

Quoted By:

>>11923940
You are overrating GPT3
https://www.youtube.com/watch?v=SY5PvZrJhLE

Anonymous

Anonymous Wed 22 Jul 2020 16:41:09 No.11926400 Report

Quoted By:

>>11923399
>Why isn't artificial general intelligence available yet?
Leftism

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