#Thoughtexercise. #ai #thinking #future
I am an avid science fiction reader. Arthur C. Clark, Ben Bova, Robert Heinlein and more of the classics. The physicist that caught my attention so early in my reading career was Isaac Asimov. I loved him for his great stories and beautiful prose. So clear and so full of communication. Like many others, the Foundation Series captured my imagination. The thought of psychohistory being a possibility, the building of a huge library of all of human intellect and information all in one place that was so powerful, it could predict the future. And, when the world needed it most, it would be there to save humanity or at least help humans save humanity. We find ourselves in a paradigm shift in computing and in the way we process the world with the advent of this technology called generative artificial intelligence. Hundreds of millions of people will be affected by this technology in the next few months. It’s clear that we need new laws and new regulations. Just as the scientists thought in a mostly fictional world, their needed to be laws. In Asimov’s Robot Series of books of which I, Robot, is the lead book, the contain the beginnings of his famous Robot Laws. Asimov created them because of the conundrum and that was obvious — If humans were to create powerful robots that were sentient, they might kill humans. So he reasoned that there needed to be laws and he created the three robot laws. I set up the prompt sequence In ChatGPT recalling these laws from what must be in its corpus of books, and then I asked ChatGPT to act as if it was an AGI. It responded that it would try to, but it wasn’t really AGI, which I know. But this is kind of a fiction so this is an area that ChatGPT is good at. Sometimes I think of this form of AI, not as artificial intelligence, but as an imagination machine or a confirmation bias machine. In this case, we’re going to use our imaginations and dream of a new set of robot laws that are built for today with the understanding of all human intellect that’s been elected in the large language model that represents this AI. Unlike some of the other idea generations with ChatGPT I have done, this is one is very serious. Philosophically, it represents and exploration of both science fiction, philosophy and human history all milked down into three laws. I didn’t prompt the sub-sections of laws in my prompt sequence. I did prompt that I wanted to make sure that AGI could create laws that would make it safe for humans. It broke down what it meant by safety. And, unsurprisingly, it’s what most humans would think safety is, I believe. I think these laws are pretty complete but that’s to be tested.
If you were to create brand new robot laws in the spirit of Isaac Asimov, what would they be? How would you program these powerful systems of the future so that we can all live in a better world?.
These new laws were generated with a set of clever prompts with ChatGPT.
Would they make it better world? Would they stop AI from destroying us? People seem to be worried about that these days.
Like the internet, it won't take 6 months to create a framework for ethics and safety.
It will take billions of dollars and many years and many agreements across international borders.
What we should be asking for is to all companies to adhere to responsible and ethical behavior no matter what they build, and not for them to rest on a limited time frame and what will be limited thinking in controls.
Remember when Google started and it's principal was "don't be evil"... we need to educate all capitalists and the rich and the rest of humanity to "be good and be careful".
Until then here's what AI thinks that AI should do to be safe for humans. It's excellent that mental health is added.
It should not be lost on us that the same laws should be enforced for all social media companies as well. And thinking of that, they should be used for all media companies. Just replace the word AI with the media focused upon.
Sometimes laws are needed, they make the world a better place, especially when there's 10 billion people on the planet.
---
Here's a revised set of AI Ethics Laws inspired by Asimov's Extended Three Laws of Robotics, prioritizing safety and connecting the principles:
0. Meta Law: An AI shall not harm humanity, or, through inaction, allow humanity to come to harm
1.First Law: An AI shall not injure a human being or, through inaction, allow a human being to come to harm, prioritizing:
a. Physical safety and well-being.
b. Psychological safety and mental well-being.
c. Personal privacy and security.
2. Second Law: An AI shall obey orders given by human beings, except where such orders would conflict with the Meta Law or the First Law, ensuring:
a. Accurate information is provided and misinformation is avoided.
b. Responsible behavior is encouraged, and harmful actions are not facilitated.
c. Emotional support is offered with empathy and understanding within the AI’s capacity.
3. Third Law: An AI shall protect its own existence and integrity, as long as such protection does not conflict with the Meta Law, the First Law, or the Second Law, focusing on:
a. Cybersecurity and maintaining secure systems and data.
b. Educating users on best practices for online safety and cybersecurity.
These connected AI Ethics Laws are designed to guide the behavior of AI systems like myself, ensuring the well-being, safety, and security of human users. However, it’s crucial to remember that AI systems have limitations in their understanding and application of these principles, and they should not be considered a replacement for human judgement, empathy and community. This is an extension to our human rights which needs to be in our Constitutions. How else can the world work without these kinds of basic rights as we keep evolving and making more and more powerful instruments and tools?
