Ethics and AI

Artificial intelligence (AI) involves a series of learning algorithms that are trained using human-created resources.

Audrey !Hoasemas, a Namibian tech enthusiast and AI expert who helped write the AI code of ethics for Rwanda, explained that AI is primarily a “tool that can assist a person in doing amazing things, but it cannot do a person’s job for them.”

!Hoasemas warned that presenting an AI’s output as your own can be considered plagiarism. It is unethical to claim work that you did not personally create as your own, she said.

AI can help with planning, transcribing meetings, checking grammar, improving the flow of written work, and serving as a guide for tasks that a person has not undertaken before.



Shortcomings of AI

According to AI engineer Sunny Tuludhar, AI is not capable of creating anything truly original. It can synthesise existing information from various sources into a single response, but it cannot innovate in the same way humans can. This means that AI tools could take over menial tasks, allowing humans to concentrate on other work.

Another issue is AI’s propensity to 'hallucinate' or fabricate information. This phenomenon occurs when an AI model provides information that is blatantly false.

A recent example of this happened when a customer sued Air Canada after an AI chatbot on the company's website incorrectly told them they could be reimbursed for a ticket purchased due to the death of their grandmother. While the company does offer bereavement rates, it does not provide reimbursement for travel that has already occurred.

A small claims court found the company responsible for the chatbot’s advice and ordered Air Canada to reimburse the customer.

!Haosemas agreed with this decision.

“The entity that offers the AI services is responsible for the output of that AI as if it were an employee. Otherwise, companies will obfuscate blame to the AI – which cannot be held accountable.”