So I was randomly selected for a drug swab test at the airport, I know that its not a person who is selecting people and judging them based on their race, color, nationality and age, and that its done by an algorithm (AI) which in turn learns to select people randomly based on the past data which has been fed in it for the AI to learn what kind of people were searched in the past and have higher probability of using/smugging drugs. I do understand that.
What if that past data being fed in AI itself is baised because past data does include searches done by humans who were in some or all ways baised. What if the past bais continues to seep into our present.
Some people would say, but hey the past data contains the profile of people who were caught, and maybe AI is doing the right thing. Well lets look at it this way, if in the past you checked 5 women for every 100 men, then naturally more men will be caught, simply because the sample size is significantly bigger. Maybe had they checked more females, they would have caught more females, but that didn't happen and now our dear AI thinks well women are not high risk based on the past data, why should i check them more now! I would continue to pay more attention to males.
This logic can be applied for gender, age, color, nationality or anything that you might think of. If you never looked in a particular direction, how would you find anything there.
Its a vicious cycle. AI learns to be biased, quite unconsciously, based on the previous data, then all the actions it takes, based on that baised past data, are also fed into the AI for it to continue learning, which kind of further cements that bias in the algorithm.
Artificial Intelligence is not really intelligence, its just imitating behaviours to increase the probability of success in whatever that algorithm was built for, based on the data fed into it.
It is critical that we understand that AI, no matter how good, is not at all aware. AI isnt aware how racial profiling was wrong or that we as humans want to change it now. AI isnt aware of apartheid or colonialism or any of the social issues which we has human race have faced and have tried to correct. AI lacks that insight. AI has no sense of wrong or right, or compassion, its just a monkey which imitates whatever you do, only faster and very accurately. It makes us more efficient and accurate but it doesnt make us more correct.
While it is good that AI in several areas has been quite useful, and I am not writing this article to blame anyone or the AI for the bias existing in the system, I believe the next version of Artificial Intelligence would be Artificial Awareness something which would be able to, unlike humans, think rationally and will be able to distinguish between good and bad, would be aware of the history, would take into consideration the collective intellect of humans before it makes a decision. I know it sounds impossible, but then whatever humans have achieved at some point looked impossible. Someday it will become a reality and oh boy that would be something!
