AI – An Unbiased View
Technology is used for the welfare and advancement of civilization and at the same time can be grossly misused for creating economic and financial harm, execute cyber-crime, create division and even accelerating possibility of wars between nations. Advancement in Technology unfortunately does not guarantee freedom from misuse, in fact newer technologies present many more such possibilities. Along with innovating and building newer technologies that make our lives easier, we need to continuously devise systems and tools that minimize possibilities of misuse. One of the classic examples in our present times would be the use of Artificial Intelligence. We could either use or misuse the unending capabilities extended through Artificial Intelligence.
Businesses using AI for increasing sales, understanding customer behaviour, building customer base and diversifying would most often use factual and objective data thus ruling out biased outcomes based on bad or inappropriate data. However, when AI is used to create opinions, replace surveys, manage civil society, apply law and order, and run political campaigns and elections, there is a greater risk of the outcomes being highly unreliable primarily due to the source of the base data. The responses or solutions created using AI is ultimately based on the data that has been used or is being collected on an ongoing basis. What is this data, what or who is it based on (how broad was the sample), and the judgement of the humans driving the data mining and analysis and creating the algorithms can hugely sway the direction of the outcome.
It is therefore extremely important to have a system that will ensure that the data is devoid of biases whether due to human intervention or sampling error. It will be interesting to see if the tools that exist now, that detects biased data or code including Google’s What-if, PWC’s Bias Analyzer or IBM’s AI Fairness 360 are doing complete justice. Unless we have tools that have high precision rates in detecting and controlling these biases, it will be prudent to avoid depending on AI outcomes in areas that affect the civil society at large.
The fact that humans will continue to be on the driver’s seat and will be responsible for the advent of newer technologies is most certainly evident by the different threats and possibilities each new discovery throws up and the continuous need to counter them through different solutions.
By
Shiny Varied Ajimon – Founder