Ethics in AI

1. Awareness of ethical impact of AI is crucial — “The privileged are processed by people, whereas the poor are processed by algorithms” Cathy O’Neill, in her book Weapons of Math Destruction (Crown). It was an eye opener for me to learn the number of ways AI implemented in the wrong way can do harm to society. At times, harm is done unknowingly. I dont want to list all of them but having your own bubble in Youtube is an obvious sign. It’s a must, that people are aware of these issues so they can demand accountability and change.
2. The scope of ethics in AI — Everything beyond the modeling core falls in the realm of ethics. From deciding what inputs go into your model to how the model results are used in practice, all have ethical implications.
3. Maintaining an ethical standard for AI will continue to increase in importance, and complexity - Just as in other technological revolutions, the focus initially is on the technology. As implementing AI becomes easier (think Microsoft Lobe, AutoML tools) it becomes more important to figure out how to implement this technology so that the right outcomes are achieved. This is not just a technology problem but needs involvement of experts in several domains.
4. Is a model biased against a smaller customer ? — Although there is a lot of focus on societal impact, what about the impact for a business? Will AI favor a larger customer over a smaller customer? Will an existing money making but dwindling product be prioritized over a newer and potentially groundbreaking product ?
5. AI practitioners need to take responsibility — Before legislation catches up, before processes are put in place, AI practitioners need to take responsibility.