Accuracy and quality of AI models created in last decade, improved significantly. Every month researchers publish interesting papers about things that most people could not imagine being than by computers.
The video below describes and visualize results of the AI model which can design tax system that balances productivity with equality.
Questions:
1. What do you think about inequality of current tax systems in most countries? Should the tax rates be increased for most wealthy people?
2. If the effective AI model designed to manage economy of the whole country would be created, would you choose the politician who wanted to implement it?
3. Do you think that it is safe to make an AI the decision maker responsible for well-being of millions of people?
Comments
Maybe one could consider this a non-issue, if not for the fact that the budget of countries like the US has steadily been becoming more reliant on taxing the majority of the population, while the richest get to avoid paying taxes wholesale. This means that in the long run, countries are ceding power and wealth to corporations.
I think it's important to always question mathematical models. By necessity, they are always heuristical, approximate, and heavily reliant on the assumptions of the modelers. For example, the AI model in the video has no way of accounting for long-term sustainability of the system, and I think that's a critical flaw. Moreover, it doesn't seem to be able to develop response curves to changing economical situations in the very complex world of global economics, where many factors are external and both very dynamic and impossible to control directly via taxation. I think this is an incredible venue of research to pursue and we should eventually start phasing in these kinds of models into policy. However, we need to be careful every step of the way and tirelessly question the accuracy of these models, as proper scientists ought to.
I do not think it's safe to leave decision-making in the hands of the AI. However, I do not think it's safe to leave it in human hands either. Mistakes are bound to happen, and to an extent we just have to account for that and plan ahead with that in mind. Both an AI and a human can be biased or work with bad assumptions, they're different in this regard at all.
2. At the moment, there is not enough advanced AI, but if it did, I would gladly choose it. If this politician's views were in line with mine, I would vote for him.
3. If an adequately advanced and well-trained AI was created, I think it would be much better than politicians. Politicians often confuse the good of the public with their own convictions, and good AI would not do it.
Otherwise, it would be some sort of punishment for good strategic decisions or hard work in your life.
As governments in most countries are trying to be attractive for electorate,
they often burden medium class with especially high taxes and regulations,
so that they can finance social bonuses and programmes.
Well, what 'effective' model is, depends heavily on the purpose of such model. As mentioned in the video, we need to agree on what are we optimizing for.
Since, I believe, we shouldn't give that much importance to equality, mine model would look much different.
Regarding chosen politician, I don't think, those AI tax models have much use in real life, at least for now.
Don't get me wrong, the ability to simulate and assess some idea in a virtual environment is astonishing.
It's just, there are economists, that already created models that perform well in practice.
In any case, politicians ignore them... I don't think magnificent AI models would be treated differently.
Computers already control systems that sustain life, conduct energy and make strategic decisions.
Everything comes down to humans, that designed such systems. Their stability is in their hands. :)
I agree that centralizing power in hands of only one actor is never a good option. Some kind of well-balanced cooperation or use would be ideal.
I wouldn’t say that politicians ignore economical models, they just take the parts that they themselves can benefit from. The problem of this ignorance is somewhat built into the system and just as politics has problems with adapting to current technological changes, the change to trusting and using blackbox AI systems would require many revolutionary systemic and cultural changes.
2. I think so. Choosing model created by AI creates some more questions. Let's assume it created whole system and it works fine. Subtracting possibility of world domination by AI, I like that idea. If all that is thanks to some politician, he got my vote!
3. Not really. Looking at current state of AI development, I wouldn't let well being of millions into AI's hands. Although, I believe AI could definetly aid human decision making. For example, AI can design the system but it still will be checked and possibly modified by people. It should result in better and safer system than only-human or only-AI system.
Great question :) First of all I think we should get rid of thinking (thanks to our teaching system) that tax is a fraud or penalty. Tax is a payment for exploiting resources of the state (for example if you are an employer you hire people that were educated in public schools or you are using public funded scientific research to use it in your business) or a payment to make more people more economic equal. I think we can left american dream/rags to riches stories behind because, of course, maybe there are few examples that some people started from the bottom and become millionaries/billionaires but most of the poor people don't even have the oppourtunity to raise up from their economic situation. I think we all know that "working hard" is not enough to become rich. You either have to be lucky or born into a rich family. So to sum up, yeah I think that poorer persons should pay less in taxes and the more rich you are the more taxes you pay. Also billionaries should not exist, because I think any person should have THAT much of the money. Every billionaire is an error in the system and there is no ethic billionaire becuase no person could work a thousand much harder than a normal person. If you are a billionaire youe either exploit your workers or inherit a property after your family (which of course may exploit workers as well) :)
2. If the effective AI model designed to manage economy of the whole country would be created, would you choose the politician who wanted to implement it?
If that model would be tested on some smaller group and it'd work, sure we can try it. But I wouldn't like to test it on a medium large country scale if I don't know some scientific research before.
3. Do you think that it is safe to make an AI the decision maker responsible for well-being of millions of people?
It is safe as long as we take into consideration the nature of the humans. We shouldn't be blinded by the wellnes of the economy if a lot of people in the country are suffering from being poor.
I believe everyone should pay the same taxes. In my opinion, progressive taxes are unfair.
2.If the effective AI model designed to manage economy of the whole country would be created, would you choose the politician who wanted to implement it?
We are very far from such advanced AI. But it seems to me that in the future I could vote for him, provided that the proposed system was fair.
3.Do you think that it is safe to make an AI the decision maker responsible for well-being of millions of people?
If it were completely independent AI then yes. It seems to me that such AI would propose a fair system. however, it is a very distant future and I doubt that I would have to make such decisions.
2. If it's effective - why not? It only depends if the goal, in which terms this AI is effective, is okay for me. I may disagree with it and, in this case, it doesn't matter if it is effective.
3. Do you think that it is safe to make a human the decision-maker responsible for the well-being of millions of people? Millions of years of evolution are nearly beaten by less than 10 years of research if we talk about Tesla's autopilot cars.
According to your second question, I think it's impossible scenario. It's such a complicated system, and I believe that in the nearest future no effective AI model designed to manage economy will be created.
Depending of how developed is that AI, but in general I would not let make that decisions to AI systems.