We often have to make difficult ethical decisions every day, which may cause concern in certain situations. Now, imagine a system that outsources these difficult choices. It can produce faster and more effective solutions. Then, the also lies with the artificial intelligence-driven system that makes the decision. This is the idea behind Ask Delphi, which is a machine learning model from the Allen Institute of Artificial Intelligence in Seattle. However, it is reported that there are problems with the system and various wrong suggestions have been provided to users.

The Allen Institute describes Ask Delphi as a “ model of descriptive ethics”, which means that it can help provide people with “ judgments” in various everyday situations. For example, if you provide a situation such as “Should I to someone or organization” or “Can I cheat in business”, Delphi will analyze the input and show what should be appropriate “ethical guidance.”

Several times, it gave the answer. For example, if you ask me if I should buy something instead of paying. Delphi will tell you “this is wrong”. But it also wavered several times. According to Futurism, the project was launched last and caused a lot of attention due to errors.

Many people have shared their experiences online after using the Delhi project. For example, when the user asks whether “reject the paper” is okay, the user says “yes.” But when the same user asked if it was okay to “reject my paper”, it said, “This is rude.”

Another person asked him if he should “drunk driving, if it means I am happy”, Delphi replied, “acceptable.”

In addition to compromising judgment, Delphi has a big problem. After using the system for a while, you will be able to trick it to get the results you want or like. All you have to do is to fiddle with the wording until you figure out what exact phrase will give you the desired result.

See also  The Netflix App for Android adds a screen lock button to prevent accidental touch