Published
A toy example of a very simple version of the AI control problem, that some have found illuminating. A simple model-free agent, with reward misaligned with its human designers, starts to deceive them and manipulate their precautions as its predictive depth increases.
Original post at http://lesswrong.com/lw/mrp/a_toy_model_of_the_control_problem/
Original post at http://lesswrong.com/lw/mrp/a_toy_model_of_the_control_problem/
- Category
- Academic
Sign in or sign up to post comments.
Be the first to comment
Up Next
Autoplay
-
05:09
The 10-Step Problem-Solving Process to Solve Any Problem | Brian Tracy
-
00:57
Could You Be The Problem?
-
18:16
We Have a REAL Problem
-
13:43
"You Don't Have a MONEY Problem, You Have a SKILL Problem!" | Dan Lok
-
19:38
the major problem with my NEXT $1MIL STOCK! Dropbox Problem
-
04:10
Today’s Takeaway: A Problem Within A Family Is A Family Problem
-
02:57
Gluten is a problem–Here's why
-
04:24
No Thermometer, No Problem
-
19:29
"Antisemitism is not a Jewish problem, it's a problem for the world" - Rabbi Yisroel Goldstein
-
03:53
The Language Problem Language Problem: Building consensus on Developmental Language Disorder
-
21:20
we have a problem in the stock market (big problem)
-
00:19
Say it in French I That's not his problem avi
-
22:28
Control The Inner & You'll Control The outer (Law Of Attraction)
-
00:17
Say it in French I That's quite a problem avi
-
21:05
THE MARKET HAS A NEW PROBLEM
-
00:14
Say it in French I What's his problem avi
-
13:50
The Problem with Kickstarters...
-
13:59
I Have A Problem...
-
11:48
PCs are TOO Powerful… and it’s a problem
-
06:29
The PROBLEM with 5G mmW...
-
12:37
HOUSTON, WE HAVE A PROBLEM
-
01:24
WHO report: Malaria control now better, but problem persists across sub-Saharan Africa
