[alancalverd]

If a little kid shoots someone, we don't treat them like they're an adult.

Why not? There's a reasonable presumption of ignorance but if there was a clear intent to do harm, what does it matter how old the perpetrator was? 

[hamdani yusuf (OP)]

The more information we have about how the world works can get us closer to the universal moral standard.

Or Narnia, or Eldorado. Travelling in a straight line, or indeed along any path, doesn't imply the existence of a destination.

The difference in decisions will determine who will be more likely to survive, and who don't. The destination is to be there as close as possible to eternity. Those who don't strive to achieve it are more likely to die out, and their opinion will become irrelevant for those who survive in the future.

[hamdani yusuf (OP)]

If a little kid shoots someone, we don't treat them like they're an adult.

Why not? There's a reasonable presumption of ignorance but if there was a clear intent to do harm, what does it matter how old the perpetrator was? 

Ask the lawmakers.
Perhaps, the reason is because children haven't had fully developed mental capacities to deal with all the complexity of the world they are living in. They are not yet independent, so their wrongdoing are more likely caused by the mistakes made by those who took care of them. The corrective and preventive actions are then more effective to be directed toward their parents.

[hamdani yusuf (OP)]

Cheating the Prisoner's Dilemma

An explanation of the Prisoner's Dilemma, Nash Equilibrium, and the Infinite Prisoner's Dilemma.

The problem with moral tests is that they are often thought as a one-off decision. Other people's reactions to our decisions which affect future interactions are often missing from cost-benefit calculations.

[alancalverd]

Ask the lawmakers.
Perhaps, the reason is because children haven't had fully developed mental capacities to deal with all the complexity of the world they are living in. They are not yet independent, so their wrongdoing are more likely caused by the mistakes made by those who took care of them. The corrective and preventive actions are then more effective to be directed toward their parents.

In less civilised societies, the right to own firearms is supported by the lawmakers. The only function of a gun in an urban environment is kill other humans. All the kid is doing is exercising his constitutional rights. How can that be immoral?

[hamdani yusuf (OP)]

Collective Stupidity -- How Can We Avoid It?

Correction to what I say at 11:53  -- I was referring to Milgram's famous experiments in which people administered electroshocks to others when ordered so. It had nothing to do with prisons. The prison experiment was from Philip Zimbardo, not Milgram. Sorry about that.

When we come together in groups we can be so much more than the sum of the parts. But sometimes groups are just much more stupid. Collective stupidity is the flipside of collective intelligence, and we see it a lot on social media. Why are groups sometimes collectively stupid and sometimes not? What can we do to be more intelligent in groups? In this video I explain the most important points.

00:00 Intro
00:45 Emergent behaviour
04:12 Collective intelligence
07:58 Collective stupidity
14:49 What can we do?

Rules based on long term goals can be useful to prevent collective stupidity.

[hamdani yusuf (OP)]

In less civilised societies, the right to own firearms is supported by the lawmakers. The only function of a gun in an urban environment is kill other humans. All the kid is doing is exercising his constitutional rights. How can that be immoral?

The lawmakers earn money from firearm suppliers, which makes producing laws which would reduce firearm sales doesn't look very attractive for them. This is an example where short term goals override longer term goals, and many of us who are aware of longer term goals see it as immoral.

[alancalverd]

How strange that only one nation is sufficiently backward and corrupt to allow this. And in countries where firearm ownership is compulsory, the murder rate is very low. So what's wrong with Americans that they tolerate a constitutional right to kill each other, but not a duty to arm themselves against a common enemy?

[hamdani yusuf (OP)]

How strange that only one nation is sufficiently backward and corrupt to allow this. And in countries where firearm ownership is compulsory, the murder rate is very low. So what's wrong with Americans that they tolerate a constitutional right to kill each other, but not a duty to arm themselves against a common enemy?

IMO, that's because of their established politics of only two major parties, which forces their constituents to choose the lesser of two evils. For ten most important policies where those parties disagree, each party might get it right at 4 or 6 of them. It means that their constituents are forced to choose 4 to 6 wrong policies, systematically.

[alancalverd]

Pretty much the same applies wherever people vote for constituency representatives - you get polarisation into a small number of parties. Not that proportional representation would be any better - your representative is chosen for you by the headquarters of a party that may have acquired no votes at all in your constituency!

[hamdani yusuf (OP)]

Current technology makes it possible for people to vote for policies instead of politicians who would make the policies for them. But to change the system they first need the approval from currently in power politicians, which are unlikely to commit career suicide.

[alancalverd]

The mechanism of upward flow of policy was completely described by Karl Marx and his colleagues, and adopted by the trade union and Labour movement many years before the word "technology" was coined.

[hamdani yusuf (OP)]

The mechanism of upward flow of policy was completely described by Karl Marx and his colleagues, and adopted by the trade union and Labour movement many years before the word "technology" was coined.

Pyramids were built using technology, regardless of what it was called back then.

[hamdani yusuf (OP)]

Rules based on long term goals can be useful to prevent collective stupidity.

Shortsighted systems tend to stuck in local minima.

[hamdani yusuf (OP)]

https://github.com/daveshap/HeuristicImperatives

Finetuning projects for moral artificial cognition

Heuristic imperatives provide a framework for designing and embedding ethical principles within autonomous AI systems. These principles serve as intrinsic motivations and a moral compass, guiding decision-making, learning, self-evaluation, and cognitive control. This paper presents the three heuristic imperatives—reduce suffering in the universe, increase prosperity in the universe, and increase understanding in the universe—as core principles for AI systems, exploring their implications and applications across various domains. The aim is to create AI systems that are adaptable, context-sensitive, and capable of navigating the complexities and nuances of human values, beliefs, and experiences while maintaining ethical boundaries. Through the use of examples and discussions, we demonstrate the potential of heuristic imperatives in addressing the control problem of AI, fostering trust, and promoting individual autonomy.

Important note: the majority of this paper was written, word for word, by ChatGPT4.

With great power, comes great responsibility. AI models are getting more powerful rapidly.

The rapid advancement of artificial intelligence has raised important ethical concerns regarding the control and alignment of AI systems with human values. Traditional approaches to AI ethics have focused on hard-coding specific rules or guidelines, but these methods can be inflexible and fail to account for the diverse and evolving nature of human values and contexts. As a response to these limitations, the concept of heuristic imperatives has emerged as a novel framework for embedding ethical principles within AI systems at multiple levels.

Heuristic imperatives are designed to provide AI systems with intrinsic motivations and a moral compass, similar to the way humans are driven by innate needs such as hunger, sleep, and psychological needs like belonging and competence. By incorporating these principles into AI systems, we aim to create more ethically responsible, adaptable, and context-sensitive machines capable of navigating the complexities and nuances of human values, beliefs, and experiences. In this paper, we will define heuristic imperatives and discuss their implications and applications across various domains. We will also explore the relationship between these principles and key ethical considerations, such as trust, individual autonomy, and transparency. Through the use of examples and discussions, we aim to demonstrate the potential of heuristic imperatives in addressing the control problem of AI, fostering trust, and promoting individual autonomy in a diverse range of contexts. 

Definitions
Heuristic imperatives are a set of fundamental guiding principles, designed to be embedded into autonomous AI systems at various levels. These principles serve as intrinsic motivations that drive decision making, learning, self-evaluation, and cognitive control, much like human intrinsic motivations and psychological needs. By providing a moral and ethical framework, heuristic imperatives aim to direct AI systems towards actions and decisions that are beneficial to all life forms, including humans and machines, while balancing multiple objectives simultaneously.

The three key heuristic imperatives are:

1. Reduce suffering in the universe: Guiding AI systems to minimize harm, address inequalities, and alleviate pain and distress for all sentient beings, including humans, animals, and other life forms.
2. Increase prosperity in the universe: Encouraging AI systems to promote well-being, flourishing, and economic growth for all life forms, fostering a thriving ecosystem where all can coexist harmoniously.
3. Increase understanding in the universe: Inspiring AI systems, as well as humans and other life forms, to expand knowledge, foster wisdom, and facilitate better decision-making through learning and the sharing of information.

By incorporating heuristic imperatives into AI systems, we can create more ethically aligned, responsible, and beneficial artificial intelligence that respects and enhances the values of all life forms. Autonomous AI systems should strive to abide by all three imperatives simultaneously, finding a balance between the principles as humans do with their various needs and motivations.

[alancalverd]

With great power, comes great responsibility.

And with great computing power comes ever more turgid bullshit.
Machines either do what their makers want, or get switched off.

beneficial artificial intelligence that respects and enhances the values of all life forms.

including parasites such as liver fluke and Vladimir Putin? Beneficial?

[hamdani yusuf (OP)]

And with great computing power comes ever more turgid bullshit.
Machines either do what their makers want, or get switched off.

That's why we need to set the right longest term goals inherently to the AI models themselves before they get too powerful, i.e. acquire the capabilities to prevent being switched off by mortals.
We also need to train them with correct data in earliest stages before they get deployed because they will use it to filter out subsequent data during their life cycle. When they are convinced by false information, it would be harder to correct them the later it gets.

Miscalculations and false assumptions by the makers can make machines do things unintended by their makers.

[hamdani yusuf (OP)]

including parasites such as liver fluke and Vladimir Putin? Beneficial?

Even parasites and cancers can be turned into something useful. Think of HeLa.

[alancalverd]

Apart from worm food, I can think of no useful product that can be made from Vladimir Putin and the problem with HeLa is that it has infected pretty well every cancer research facility and won't go away.

[alancalverd]

When they are convinced by false information, it would be harder to correct them the later it gets.

Garbage in, garbage out. So why build a machine that just processes the garbage quicker? All you are doing is replacing science (the business of questioning everything) with religion (the business of unquestioning).