Artificial Superintelligence Must Be Illegal.

AI is a consumer or technical product that cannot buy itself to be by itself.  There are four basic cognitive functions according to Jung - creativity/intuition, thinking, feeling/values and sensation. They are best used within four cultural arenas (realism, hierarchy, activism and individuality/retreatism). The best AI provides the user with input from all four functions - with an independent thinking check verifying intuitive creativity. They cannot both use natural language models. Thinking must be rigorous in fact checking where intuition is concerned with the general paradigms of the group (which generative AI is good at teasing out). The thinking subroutine must independently check the intuitive against both sensory reality and questioning the group intuition - or if the intuitive is "introverted" and comes up with one solution rather than suggesting many - thinking must than look at the broad base of knowledge as it exists in an organized fashion - checking some kind of index in opposition to a best guess scenario. Intuition must also be checked against values/feelings/ethics. AI cannot have its own - but it can look at both societal positions and an ethical values matrix.

It needs to report on all four (or 8) functions and how they relate to other social actors within an organization or society. Part of the ethics of AI is to be allowed to look at the analysis of other AI solutions by competing actors. For example, if a corporation hierarchy is using an AI to do the above analysis, an AI serving either egalitarian or reactionary organizations should be able to deconstruct the competing analysis. The question is how to manage disclosure of facts and values. References must be supplied and verifiable to each side. Individualists and realists (aka consumers and the working class) should also be in the client position and AI for each way of life should be able to project what the AI in the other ways of life will recommend. Also to how the human actors would respond, even and especially without AI assistance.

Again, let me emphasize that the AI systems proposed above are tools, not something with agency. Indeed, using such a tool should not replace humans, it should educate them so that neither some shadow AI or more likely those who operate against their interests - capitalist hierarchs and reactionaries - cannot exploit them.