We're still in the basement of whatever curve this is, so zero to zero.
@esyudkowsky
-
GPT Evolution: Beyond Stochastic Parrots with Scaling
By
–
GPT-3, GPT-3.5, GPT-4. How blind would you have to be, to look at whatever has increased with the version number, and see only a "stochastic parrot" with a bigger memory?
-
Modified Physics Problem Transformed into a Gotcha Question
By
–
The problem is a modified version of the first problem in Thinking Physics, transformed by me into a gotcha by changing "total miles" to "total hours". I should be quite surprised if that exact problem and its answer was in GPT-4's training set to be memorized.
-
Missing the Point of AI Safety Concerns About Superintelligence
By
–
Sure, but also, utterly missing the point of notkilleveryoneism which is the concern about an AI that is smarter than all the little humans.
-
AI Systems Imagined Without Internal Mechanisms or Properties
By
–
This post is an utter case in point. The supposed AI is imagined to have absolutely no internal properties or machinery, just a blank causeless property of faithfully predicting what the programmers hoped it would predict.
-
DeepFloydFilter Issue: Silent Processing and Motion Parameter Adjustment
By
–
…it was actually DeepFloydFilter silently eating everything; took me a while to notice there was a call in there. 27 does seem too low for motion, once the filter is off. Proposed fix, set DeepFloyd to verbose.
-
Stable Video Diffusion notebook failed with blurry output images
By
–
Tried it on https://
colab.research.google.com/github/mkshing
/notebooks/blob/main/stable_video_diffusion_img2vid.ipynb
… and it straight-up didn't work. Input image had requisite height and width, output images were giant blurs of same. -
AGI Ruin: A Comprehensive List of Lethalities and Counterarguments
By
–
It's not great, but it sure beats the detail levels of any counterclaims: https://
lesswrong.com/posts/uMQ3cqWD
PHhjtiesc/agi-ruin-a-list-of-lethalities
… -
Heat Transfer Limitations in Advanced Computational Systems
By
–
Terrible heat transfer. Dumping all your entropy into Earth's oceans as a negentropy sink doesn't last but it gets you a valuable early burst of computation before you start disassembling planets to enclose the Sun in solar panels, which Earth also doesn't survive.
-
Alignment Concerns with Internal Health Reward System Design
By
–
The first comment (kudos for open review) links to a post that says some of this at greater length, but to repeat my own reaction: "There's nothing in there about alignment. The proposed motivational system is internal-system-health reward with nothing about caring for humans."