AI came up with an "ethics" framework that was an objectively terrible one.

So I asked (AI) to provide a framework for ethical decision making, and boy did it whiff. <br><br>I have been trying to get the AI to come up with its own (admittedly flawed) framework for ethical decision making. It is a goal of mine to create a framework of ethics, and I figure that if I get an AI to do it, I have a decent starting point. It will be easy to identify the flaws in the AI's framework, fix it, and use it. <br><br>I still have not achieved this. Often the AI just spews up the list of questions that you should ask yourself when dealing with an ethical dilemma. However, it did one day give me a step-by-step process to use. <br><br>The process and its flaws (in bold) are as follows:<br><br>1. Investigate. Do research. Gather information. <br>2. Check whether you can pass the front page test. <br>3. Make a decision. <br>4. Check whether you can pass the front page test again. If you can't go back to the beginning. <br>5. Execute the decision. <br><br>The front page test is a question that came out of a scandal at Enron. After the company went bankrupt and it became clear the CEO and many in the company had committed crimes such as tax evasion (and subsequently been charged and convicted and sentenced for those crimes) the CEO gave an interview, in which he said that he only made decisions that he could see on the front page of a newspaper. <br><br>The AI thinks that if you can and can't pass that test before and after the decision, that means that the decision is ethically sound. **Why would changing your ability to pass the test have any affect on the ethics of the decision? How does ignorance of this change in ability make a decision more ethical?**<br><br>**This framework doesn't provide any means of investigation, and in particular does not provide any means of relationally investigating the effect of the decision on the primary stakeholders.**<br><br>**By being so vague you don't even have to actually look into the decision you are making. You can just say that you did.**