While human being can be biased, discriminative and unfair, I firmly believed that machines could be perfect. My reasoning where that machines will be free of human follies like jealousy, personal likes and nepotism.
Then during my studies around digital transformation it downed upon me that machines can be unfair and biased too. This makes sense too, since humans are the ones creating them. AI elements need data to train them and originally this data comes from humans.
Take an example of a bot that screens out job applications for a particular post. Previously this work was done by a recruiter. Historically, if the recruiter has shown certain preferences to a particular ethnicity or an educational institute in the resume, this bias goes into the decisions based on which the bot decides the fate of the applications now.
The good news is that bias is AI is very much recognised and efforts are made to avoid it. An example is shared in the link below where IBM not only acknowledges the risk of increased bias in future AI but also promise effective measure to mitigate it.
https://www.research.ibm.com/5-in-5/ai-and-bias/
Subscribe to:
Post Comments (Atom)
Portfolio management turns investments into capabilities
Organizational capability is directly linked with value and benefits. So, the purpose is to turn investments into value. Once an investm...
-
Part of the capacity and capability management revolves around human resources. To fulfil the vision and strategies of an organisation, ...
-
One thing that sometimes bother me in the lack of clear and crisp feedback from senior management. To keep an employee or a project team mo...
-
The hardest part about risk management is that it is a perception. It is one’s own view of factors that can affect the progress. Since t...
No comments:
Post a Comment