Algorithms are biased, predictable,dumb, incorrect, unethical and much more. Now just replace the word 'Algorithms' with 'Humans'.
Thanks to analytics ,we now know that most of us order a few food items ( less than 10) in each location . While Baskin Robins has 30 flavors ( one for each day of the month), best selling are not more than 5. Bulk of the hit movies have the same formula for decades but we still like them again and again ( be it the comic super heroes, revenge dramas or situational comedies). Every grocery subscription on Amazon reinforces the fact that inspite of 100s of brands available, we stick to the same month after month. Watch any tv debate and you will find one or two panelists who spew the same stuff that popular ethics deplore. News/TV/Social media contain the same inaccuracies or lies that everyone criticises algorithms about.
Is it any surprising that algorithms which learnt from the same human behavior are biased, predictable,dumb, incorrect, unethical and much more? Why do we hold them to a different standard from their creators' ?
Ask 10 people the capital of a country and chances are most would be wrong. Unless someone corrects the humans of their incorrect answers, they carry the same wrong information to their graves. An algorithm once corrected ( which is entirely possible through feedback loops), never repeats the mistake. We cannot guarantee that humans don't repeat their mistakes. If humans become infallible by learning, we wouldn't have so many traffic signs!!!
When automobile industry started with assembly lines, it had lots of people doing repetitive tasks in a specific order and to a specification. Today most of the lines are replaced by robots with a very small number of humans addressing exceptions/errors. Cars made by these robots are running for tens of thousands of kilometers/miles and world seems to be fine with them. While recalls are costly, they are much fewer today than earlier.
Predictable, sometimes quirky human behavior lends itself so well to feed the algorithms. We are more predictable /biased than we think and we are more times wrong than we care to admit. If anything, algorithms repeat the same behavior tirelessly.
Benefits of any technology are almost always under-estimated or over-estimated. Before judging them, it might be worthwhile looking into the mirror.
