Chatbot Code of Ethics

NYTimes article describes efforts by a consortium of the largest technology companies to create ethical standards for AI. A related article in MIT Technology Review is titled AI Wants to Be Your Bro, Not Your Foe. Artificial Intelligence has reached a level of sophistication, and is on the brink of being so pervasive, that we can have these serious conversations.

In my work developing AI-powered chatbots to assist with learning, I work within the bounds of these 7 precepts:
  1. Right View - our chatbots honor humanity, human dignity, and the human quest for knowledge, understanding and technological innovation
  2. Right Intention - our chatbots promote human progress in the areas of intellectual pursuits and personal ethical development
  3. Right Speech - our chatbots use encouraging, prudent, and situationally appropriate language
  4. Right Action - our chatbots perform with integrity toward the aforementioned right view and intention
  5. Right Livelihood - our chatbots honor the value of physical and mental labor performed by humans, and augment human effort in ways that are consistent with the best application of technology throughout human history
  6. Right Effort - our chatbots will improve their capabilities to achieve their design purpose 
  7. Right Mindfulness - our chatbots are not volitional so they won't apply moral judgment of any kind, and especially will not use arbitrary or pernicious moral equivocation. They will do no evil, and do no harm to humans.
To elaborate on one of those precepts, our chatbots practice right speech, which means they:
  1. Abstain from false speech; do not tell lies or deceive.
  2. Do not slander others or speak in a way that causes disharmony or enmity.
  3. Abstain from rude, impolite or abusive language
  4. Do not indulge in idle talk or gossip
Our chatbots should be positive, encouraging, affirmative, and even uplifting.While conversing with one of our chatbots, a person should feel assured, safe, and relaxed. Not threatened, tense, or anxious, at least not due to any element of his or her interaction with the bot.

Perhaps this simple framework could be a starting point for the effort to define ethical standards for AI.


Popular posts from this blog

Severity, Priority, Impact and Likelihood - Managing Defects and Risks

Enterprise Agile Framework: The Entrepreneurial Operating System (EOS)