[op-ed snap] Rooting AI in ethics


From UPSC perspective, the following things are important :

Prelims level : Nothing much

Mains level : Ethics of AI

Most commercially available AI systems are optimized using the teleological perspectives and not the deontological perspective. 

Ethical issues – a case study

An AI system introduced in 2015 in the U.S. failed to recognize the faces of African Americans with the same accuracy as those of Caucasian Americans.

  1. From a teleological perspective, this flawed AI system gets a go-ahead because Caucasian Americans constitute 72.4% of the country’s population
  2. From a deontological perspective, it can be rejected as its intention was not to identify people from all races. 
  3. Digital platform companies, whose markets span many countries should aim to identify faces of all races with equal accuracy.
  4. AI facial recognition systems are used for law enforcement. Someone can be labeled a threat to public safety just because of limited data based on one’s skin color was used to train the AI system.
  5. The bias in the data used to train the algorithm stems from flawed historical and cultural perspectives and they contaminate the data.

NITI Aayog has a ₹7,500 crore plan to build national capability and infrastructure. The transformative capability of AI must be rooted in an egalitarian ethical basis. 

Inline Feedbacks
View all comments