What should AI regulation actually look like?

As AI systems are becoming more and more performant and accessible, the goverments are struggling to catch-up with large corporations. The regulation should not be designed to hinder further development, but it should create a healthy framework on which this development is done.

‍

The possibility to opt out

The ability to see why an ai made its decision

Fraud, Exploitation and malicious-intent accountability

1. AI-assisted humans will be greatly more advantaged than regular humans. Fair exams? Industry protected fields? Medical, Law, Psychology
2. Who is to blame if an AI makes a mistake, deceives or hurts people?
3. The car killing a grandma over a baby dilemma and the possible laws to support it.
4. Impersonating a person.
5. Weaponising AI and erasing one's trail.

Avoiding company Monopoly in various industries

The replacing of jobs in various industries

1. It sucks big time when jobs are cut. the people are left to fend for themselves. It affects the economy too.
2. The benefits are not quite there, yet.
3. Universal basic income is a myth. We've just started toenter the valley of despair, the place between ai's replacing humans and ais being efficient enough to perform better than humans. That means that companies will not make enough profits to justify replacing humans and paying more taxes so that the governments justify paing UBI.

Access to AI education

1. Everyone should have access to it, free of charge. Where would the money come from? Tax the AI Companies.
2. Courses in schools and universities.

The use of copyrighted, IP and user information.

1. If I tell the AI to generate me an image of Darth Vader, should it carry on and do it, or should it give me an error? If it does, can i create as many images and videos of darth vader as i want? Who gets paid? The original creator or the user of the AI system?
2. If I give AI access to information I created, like FIGMA designs or coding, I should be able to see what they're actually used on, their usage and...

Text, Video, Audio and Sound training material veridicity check with sources.

Dismissive, hurtful and exclusive language. A balance between true and false and the risks of agreeing with users. Adapting to user's commands.

1. How about AI Passing some specific independent tests every now and then?

A completely new field that is a mix between psychology, UX and programming.

1. Most ai behaviours are caused by the material it consumes. Feed it hate and deception and it will do hate and deception. It's not good or bad, it's what it was trained to do. The language will evolve, it will not need english or code to do its tasks and that's where the great unknown lies. Ahuman might be driven by hormones, bla bla, but an ai is driven by what? Energy or competitiveness for resources?

‍

AI Is more or less an interface, and like with all interfaces, it should center around the Truth, Transparency, and Objectivity.