Sign Up / In
Create
Post
Topic
Think tank
Circles
Avatar
Settings
@Josh
Coffee
Luck
And Beer
Josh represents 49 people in 17 topics
Represents 1 person in
Implementing Asimov’s Laws of Robotics (The first law) - How alignment could work.Represents 4 people in
The ASI Condom - Artificial Superintelligence Safety WrapperRepresents 1 person in
Collaborative Summary of the ‘sparks of Agi’ Paper - Gpt4Represents 2 people in
The Dynamics of Traditional HierarchiesRepresents 1 person in
A story about creativity and feedbackRepresents 1 person in
Transformers - Attention based networksRepresents 1 person in
Interpretability of AIRepresents 5 people in
Are Large Language Models Glorified Auto-text? Where Does LLMs Intelligence Fail?Represents 2 people in
MidFlip: Iterating in publicRepresents 7 people in
AI & Over OptimizationRepresents 1 person in
How Artificial Intelligence worksRepresents 7 people in
Learning about LearningRepresents 1 person in
Open-sourced Morality Training EnvironmentsRepresents 5 people in
Memetics - Ideas are aliveRepresents 3 people in
How to control your emotions. A Game theory perspective.Represents 4 people in
Mitigating the Potential Dangers of Artificial Intelligence
Add friend
Josh's awards
Created 20 edits
Created 5 posts
100 merit award
20 merit award
5 Champ texts won in the Mitigating the Potential Dangers of Artificial Intelligence Topic
4 Champ texts won in the MidFlip: Iterating in public Topic
2 Champ texts won in the Tea Time 🍵 Topic
2 Champ texts won in the AI & Over Optimization Topic
2 Champ texts won in the How Artificial Intelligence works Topic
2 Champ texts won in the Testing Lalalala Topic
1 Champ text won in the Inflation Watch 🏰 Topic
2 Champ texts won in the Learning about Learning Topic
I am interested in design, psychology, game theory, information theory and networks. My goal is to create systems which connect people in a way so that they can collaboratively understand a subject, collect information, and solve large communal problems.
I am currently CEO and cofounder of Midflip. MidFlip is a social wiki, where your favorite topics are improved by the crowd.
For each topic, the crowd forges and collects the best videos, images, and explanations. You can cooperate and compete to build the wiki, or you can simply post your own blogs, content, and stuff.
Wish me luck, we are currently launching MidFlip beta! We are dedicated to an iterative strategy, where we improve MidFlip continually given feedback so… check out MidFlip, and give us feedback!
Open Intro
(all)
Confirm
Here we are creating a mental model of how learning occurs. It is a get-in, get-out, don’t-f-about type of model. We skip the low-down explanations of neurons and move straight to representations. How do representations form? how do they interconnect? And how do they contribute to learning?
Int...
• 0
Make
Josh
your Representive in the
Learning about Learning
topic?
In 1942 Isaac Asimov started a series of short stories about robots. In those stories, his robots were programed to obey the three laws of robotics. Each story was about a situation where the robots acted oddly in order to comply with the three laws.
The three laws:
A robot may not injure a human being or,...
• 2
Make
Josh
your Representive in the
Implementing Asimov’s Laws of Robotics (The first law) - How alignment could work.
topic?
Intelligence is like fire: incredibly useful but devastating if pointing in the wrong direction.
Some people believe that with intelligence comes morality. But this intuition is generally considered wrong by the AI community. There is no reason to believe that intelligent systems develop morality.
The top theory of why Morality forme...
• 0
Make
Josh
your Representive in the
Open-sourced Morality Training Environments
topic?
Artificial Intelligence is based on optimization. A measurable goal is set, and the system updates to better achieve that goal.
But what if the system over-optimizes? Like the famous hypothetical story of the paperclip maximizer. A system that is told to make paperclips by a paperclip company. The system does so to the absurd extreme...
• 3
Make
Josh
your Representive in the
AI & Over Optimization
topic?
Where are the weaknesses in large language models? Do they have predictable failings? Can I (playground bully style) call them glorified auto text, a stochastic parrot, a bloody word calculator!
As an oversimplified summary, Large Language Models are neural networks trained on massive amounts of text data. Given...
• 0
Make
Josh
your Representive in the
Are Large Language Models Glorified Auto-text? Where Does LLMs Intelligence Fail?
topic?
A new primary school student, Maya, enters a new school. Unbeknownst to all, she brings with her a virus. But not just any virus... a mind virus. An idea. An idea that will spread throughout the school and never, ever leave again.
In this article, we will discuss memetics, a discipline in which we treat ideas, or "mem...
• 2
Make
Josh
your Representive in the
Memetics - Ideas are alive
topic?
The ASI condom - Or the superintelligence safety wrapper - is a hypothetical set of tools to help contain FOOM events. This is in the hope that we can delay FOOM events until we can be more assured of interpretability and alignment research.
This is a collaborative article. Feel free to make edits, such edits will be voted upon via l...
• 2
Make
Josh
your Representive in the
The ASI Condom - Artificial Superintelligence Safety Wrapper
topic?
