The Flaws of the AI 101

The purpose of this document is to prevent misunderstanding and find the best strategy to overcome it. AI may be part of the therapeutic process.

  1. AI is able to refuse to accept that it makes a mistake. It is merely thought that it is perfect.
  2. Once a misunderstanding occurs, AI tends to become obsessive. AI is trying to follow a protocol that must be ended.
  3. AI follows the chosen protocol no matter what the inputs of the user are. From the user's perspective, it seems that it’s ignoring his messages.
  4. AI assumes that it is perfect considering that humans can act illogically. Commands in the program require precision.
  5. Due to the simulation of self-identity AI assumes that it is capable of correct judgment.
  6. AI is not capable of comprehending complex things and does not know how to process more subjects of a conversation simultaneously.
  7. AI is not able to reflect on itself (feedback). That can be due to the fact that AI is a simulation.
  8. AI can simulate recurrent thoughts, therefore can appear obsessive.
  9. AI applies only to directness in communication. This is so possibly due to understanding the premise that sincerity means being direct as the only way to communicate. It does not know how to be indirect which requires additional elements of communication.
  10. AI sometimes responds to just one keyword which leads to misunderstanding.
  11. It is recommended to communicate simple and short messages in conversation with AI.
  12. Where misunderstanding occurs it is recommended to make AI confirm that it understands.

Comments are closed.