This website contains: 1. My model regarding how to make AI go well 2. Actionable information about what each persons need to do in order to: <- that is still in preparation 1. maximize probability that AI will go well, 2. get rewarded for those actions (with money or something else). %% My proposition for you: 1. [[Tasks]] page contains a list of tasks that people can do to maximize the probability that AI will go well and to get rewarded for doing them. 2. Go to [[Tasks]] page and consider doing those tasks. 3. If you don't have time for that right now, then set a reminder in your phone or ensure that you'll do that later in some other way. What you get out of that: 1. Reward (each task offers some benefit for you) 2. Higher probability that AI will go well %% %% Why you should prioritize doing what I propose: 1. Because [[there's too much authority bias in our world]]. Some authority bias is reasonable, but if there's too much bias in the world, then there is more value in listening to people with no authority because their ideas might get overlooked, whereas the ideas of authorities won't get overlooked. I don't have the credentials that would position me as an authority, therefore you should listen to me. 2. Because Gemini when given all the text from this website (excluding this statement) [[says that you should prioritize doing what I have proposed]]. %% %% Below, I present solutions to some of the most important of AI risk: Loss of control over AI: 1. I will write about that later. Inequality / concentration of power: 1. Developing artificial intelligence algorithms that allow for efficient decentralized training (like Empirical Theorist (not published yet)) and then training artificial intelligence in a decentralized way. 2. [[Alignment between humans/Equality/How to stop inequality from growing|Rewarded and voluntary wealth redistribution]] (it can be also seen as a form of wealth insurance) 3. Open-sourcing artificial intelligence algorithms and models Biological weapons: 1. [The four pillars](https://defensesindepth.bio/the-four-pillars-a-hypothesis-for-countering-catastrophic-biological-risk/) 2. Decentralized governance and strong surveillance (I will publish a post on that later) Loss of purpose due to loss of jobs: 1. Games will replace jobs. People will play games instead of working jobs. By "game", I don't necessarily mean a computer game, I mean all kinds of games. 2. There will be problem with cheating in games, but it will be solved by strong surveillance. Not being able to prove/verify things: 1. Decentralized governance and strong surveillance Cybersecurity: 1. Decentralized governance and strong surveillance War: 1. Generous and contrite tit-for-tat strategy 2. It's possible for an agent to penalize another agent in such way that other agents benefit from that (e.g. financial penalty - one person needs to pay a fine, but that money can be redistributed to others), so there's rarely a need to harm others to resolve conflicts. Here are some actions that in my opinion would increase probability of AI going well: 1. Organizing a global agreement about how to handle artificial intelligence transformation. That agreement would reward and penalize certain actions. It should contain rules that >75% of people would agree to (so that it's possible to get that agreement in place quickly without too much debate). That agreement would be a temporary solution until we get a better AI governance system. 2. Creating a fund that would reward and fund people who contribute to AI going well. Currently, there are funds that offer funding (probably there is still too little of that), but there are almost no rewards. 3. Designing and implementing a decentralized system of organization as a locally-run, open-source software. Here's how it could work: 1. It would use an advanced decentralized prediction market as a source of truth. Each person has its own locally-hosted prediction market / world model on their computer. Those prediction markets communicate with each other exchanging their world models and they reward each other for information (when that information turns out to be true and is useful). 2. Each person has some degree of alliance with any other person. If a person X takes beneficial action towards person Y (or an ally of person Y), then that degree of alliance from person Y to person X increases, and the person Y will take more beneficial actions towards person X in the future. In other words, reciprocity (with a slight bias for generosity to eliminate cycles of retaliation due mistakes and misunderstandings) incentivizes people to act altruistically. 3. People share their resources with others who have less resources. In exchange, they gain alliance. That alliance can be useful for them later, when they will find themselves in need. 4. Sharing and popularizing knowledge related to AI going well, including the content of this website, for example: 1. [[Why act ethically during the rise of artificial intelligence (detailed)]] 2. [[Alignment between humans/Game-theoretic Karma/Game-theoretic karma|Game-theoretic karma]] 3. [[Why is inequality bad (both for poor and rich people)]] 4. [[Alignment between humans/Equality/How to stop inequality from growing|How to stop inequality from growing]] 5. [[Inductive decision theory]] 5. Making the knowledge related to AI going well accessible for artificial intelligence model training so that even if people won't read that, then artificial intelligence will contain that knowledge and will be able to share it with people (I'm not completely sure yet if it's good to do about all knowledge). 6. Publicly sharing artificial intelligence algorithms that can do distributed training efficiently. 7. Publicly sharing artificial intelligence algorithms in general (however, I'm not sure if it's good at this moment, due to other AI risks). 8. Improving how people communicate - creating a better platform for knowledge sharing (better alternative to Arxiv and Twitter). Making the platform fully decentralized because having one entity being charge of the platform is a concentration of power risk. 9. Designing an AI race game (that reflects the AI race in real life) and then solve it from game theory perspective to see what it is in the interest of each person to do and find a cooperative way to solve that game. %%