SAFETY - AI Dynamics

Alert: Superposition of Waluigi Simulacra

By

@goodside

–

10 March 2023 22h27

oh no look out it's a superposition of waluigi simulacra

→ View original post on X — @goodside

10 March 2023

ChatGPT’s Dangerous Nuclear Stability Assumption Highlights AI Safety Risks

By

@willknight

–

10 March 2023 22h00

When ChatGPT was asked (WSJ essay by Schmidt et al) to write about AI and arms control, it's funny that it assumes giving AI the nuclear codes would lead to "unprecedented stability". Great example of a coherent answer that is, in fact, completely insane! https://
wsj.com/articles/chatg
pt-heralds-an-intellectual-revolution-enlightenment-artificial-intelligence-homo-technicus-technology-cognition-morality-philosophy-774331c6
…

→ View original post on X — @willknight

10 March 2023

Glass Walls Cannot Harm Advanced AI Systems

By

@parrysingh

–

10 March 2023 20h45

Nope. A glass wall neither freezes me, nor scalds me.

→ View original post on X — @parrysingh

10 March 2023

Research Paper Methodology Flaws Spark Important AI Discussion

By

@jeffdean

–

10 March 2023 20h12

I definitely like the paper in that it focused attention on the potential issue and started a good (& still ongoing) discussion about the right methodology. I just wish it hadn't had flaws that caught people's attention w/eye-catching numbers that turned out not to be correct.

→ View original post on X — @jeffdean

10 March 2023

EU Addresses Real Risks Posed by AI and Algorithms

By

@ronald_vanloon

–

10 March 2023 18h53

‘Risks posed by #AI are real’: EU moves to beat the #Algorithms that ruin lives
by @kalyeena @guardian Read more: https://
buff.ly/3A4zJZH #ArtificialIntelligence #MI #Digital #DataScience cc: @ronald_vanloon @bigdata @ravikikan

→ View original post on X — @ronald_vanloon

10 March 2023

Robot Dogs With Guns: AI and Robotics Innovation

By

@ronald_vanloon

–

10 March 2023 14h50

#Robot dogs with guns
via @EvanKirstel #AI #ArtificialIntelligence #MI #Robotics #MachineLearning

cc: @maxjcm @ronald_vanloon @pascal_bornet @marcusborba pic.twitter.com/vZ5kkCQfGW
— Ronald van Loon (@Ronald_vanLoon) 10 mars 2023

#Robot dogs with guns
via @EvanKirstel #AI #ArtificialIntelligence #MI #Robotics #MachineLearning cc: @maxjcm @ronald_vanloon @pascal_bornet @marcusborba

→ View original post on X — @ronald_vanloon

10 March 2023

Lifeguard Drones Now Saving Humans from Drowning

By

@ronald_vanloon

–

10 March 2023 0h16

Lifeguard #Drones are now saving humans from drawing
by @TechInsider #AI #ArtificialIntelligence #Tech #Technology #Innovation #TechForGood #FutureOfWork

cc: @maxjcm @ronald_vanloon @pascal_bornet @marcusborba pic.twitter.com/WsUf5hnTSx
— Ronald van Loon (@Ronald_vanLoon) 9 mars 2023

Lifeguard #Drones are now saving humans from drawing
by @TechInsider #AI #ArtificialIntelligence #Tech #Technology #Innovation #TechForGood #FutureOfWork cc: @maxjcm @ronald_vanloon @pascal_bornet @marcusborba

→ View original post on X — @ronald_vanloon

10 March 2023

Academic AI Tasks Bias Toward Short Length Outputs

By

@yitayml

–

09 March 2023 22h25

Read the flan limitations section of the blogpost. Most of the academic tasks are short. It creates a bias for short length outputs.

→ View original post on X — @yitayml

9 March 2023

The Hot Mess Theory of AI Misalignment Framework

By

@karpathy

–

09 March 2023 17h55

"The hot mess theory of AI misalignment"
a favorite talk from a recent alignment workshop turned article; offers a unique and imo fairly realistic framework for superintelligent system futures that departs from your stock paperclip maximizers.

→ View original post on X — @karpathy

9 March 2023