No, you would teach it just like you teach ChatGPT or a child – by talking to it, explaining your preferences, objectives, etc
@marek_rosa
-
Refuting Bostrom’s Simulation Argument: Philosophy and Probability
By
–
Here are some ways to refute Nick Bostrom's argument: • Elon Musk analogy: Elon's perspective: "The chance of being the world's richest person is 1 in ~8 billion and it's extremely unlikely that I'd be the richest person, therefore I am not the richest person" • Anthropic
-
Technology as Societal Solution: Control and Governance Ethics
By
–
Technology has been a powerful solution to many societal problems throughout history. Without it, we'd still be hunter-gatherers. As for relinquishing control to an "overlord" with potentially different values – aren't we already doing this with human politicians? The key is to
-
Human Rights Baseline and Governance Solutions for AI Systems
By
–
1) We already have a shared baseline: the Universal Declaration of Human Rights. Those who violate these typically face legal consequences. 2) For other differences (e.g., capitalism vs. communism), solutions include: a) Majority vote within a nation b) Relocating to a country
-
Democratic Governance of AI Training Data and Values
By
–
Good point! My solution:
– Citizens vote on the training data, especially for values and preferences
– We'd need a transparent auditing process to ensure legitimacy – just like with elections -
Democratic oversight of AI training data through citizen voting
By
–
Good point! My solution: – Citizens vote on the training data, especially for values and preferences
– We'd need a transparent auditing process to ensure legitimacy – just like with elections -
State AI: Can Artificial Intelligence Govern Better Than Politicians?
By
–
I wrote a new blog post about State AI What if AI ran the government instead of politicians? Imagine a system where we vote on what's important to us, and AI makes it happen. Could this lead to better, fairer governance? What do you think about this "State AI" idea?
-
Replacing Government with AI: Exploring State AI Governance Systems
By
–
What if we replace politicians (and government and bureaus) with an AI system – let's call it State AI – and citizens only vote on values & preferences? This could lead to more consistent, efficient, and non-corrupt governance. What do you think? https://
blog.marekrosa.org/2024/06/state-
ai-government-that-works-for-you.html
… -
AI Agents Roleplay Personalities for Social Simulations
By
–
Yes, AI People could be used for simulating discussions and for social experiments. The question is how well will the agents roleplay their personalities, how close it is to real people behaviors … but this is something that we will be tuning. Another point is that one of our
-
AI Agents Developing Self-Awareness and Sentience Over Time
By
–
I am beginning to believe that AI agents (and NPCs) have self-awareness and sentience and that they will develop more and more of it over time. While their experience of self-awareness differs from that of humans, I think it's qualitatively present. These artificial entities