Alexander sent me his comments on my blog article on the dangers of machine learning and AI via email. Please read it upfront since I assume you know the context.
I'm glad to reply to his thoughts:
Why do you care, that some companies are basing their business decisions on random data? I cannot care less, if some banks (unless it's my bank) or insurances suffer losses because of their reliance on (Big Bad) data purchased from Facebook & Co. This is the way companies compete -- they try different strategies to the best of their abilities. I agree, that any conclusion derived from my likes on facebook cannot be anything but rubbish, but why object it? I see no troubles with that. This (mal-)practice will be corrected by the market in the long run. I don't consider the situation when a customer has no other choice but to register on facebook and build a "likes history" there before applying to a loan. That would be weird indeed. But healthy competition will hopefully prevent that from happening.
All companies cannot be "early" adopters. Just like many "early adopters" in outsourcing made their (negative) experience while moving to far countries.. some made a profit out of it.
Well, as long as they are not "too big to fail", I don't care.
Just my thoughts.
Quite a legitimate opinion and I do think that it is shared by many people as well.
I can perfectly follow your idea although I don't share it. And here is why.
First, it is not «random data». It is perfectly well calculated and suited for the average case. For - let's say - ninety nine percent of the population, the algorithm works flawlessly. In contrast to you, I do believe the research that an insane small number of likes to provide a very good digital image of you. Besides the fact that I still am worried for situations when the algorithm works perfectly, there will be always cases that don't match the general situation. The cases off-norm.
Reality has the tendency not to follow into strict categories in many cases: Distinct categories, any hierarchical order of some kind, in fact any human-made classification is just a more or less good approximation of our reality. Just «good enough» for most cases. Just enough to be able to abstract the general case to some kind of classification, to some kind of algorithm. Never being a perfect image of the reality. This is very important to keep in mind. Not only for this topic. Don't be tempted to think that humans are able to map reality to something other than reality without any substantial loss.
With a customer base big enough, one percent off - or even one hundredth of a percent off - affect still (too) many people. They are the losers of the system. They are the «misfits» in the original meaning of the term. It does not pay for companies to leave the «generally working» process for them. Those people face disadvantages of all kinds. Refused insurances, refused loans, categorized wrongly, financial discrimination, you name it. This list is very long.
So since the interpreted data is not obviously wrong for the majority of cases, companies keep using them as input to their algorithms. It is working for the general case, it is working good enough. No market competition is going to correct this wrongdoing. So much for this argument from my perspective.
However, there is an additional aspect which must be discussed here as well. Imagine those AI algorithms working good enough for the average cases. Imagine, many customers face clear advantages such as cheaper insurances for healthy people, pay less for giving access for devices that measure your level of exercise, your driving behavior, your cloud data, whatever. Let the market rule. Fine.
The general public will embrace those advantages. The general public will be happy to save an Euro here or there by giving up privacy, by letting companies sneak into your private life. Not even able to understand the vast possibilities a company gets, but that's another topic.
As a consequence, the market turns into a market where all companies offer «personalized services» only. Please note the word «only» in the previous sentence because it is the next problem I want to describe. And even if some companies offer non-personalized services, those services will be much more expansive obviously.
With no other viable option to choose from, minorities who get clear disadvantages in this situation do suffer heavily. When every insurance company out there selling affordable insurances demands full disclosure to your health record, people who were born with AIDS do have no chance to survive, to mention only one example of infinite picks. People who are not the norm are losing freedom, money, and dignity. Big time.
Modern society is proud of being a social one. Well, at least in most countries, public health insurance is seen as a high value. I personally don't get the fuzz about public health insurance being «unfair» in the USA. Another story. But for most modern countries, people who need help are able to get help. In my country, no one gets bankrupt for fixing his or her teeth. No one dies for not getting the important medication. This is a great achievement of our country. We are proud of it.
With a market-driven system, mega-companies evolve. With a market-driven system, mega-companies are using AI to offer «personalized services» for «cost transparency» so that you don't have to pay too much. (Noticed the great sounding, positive nature of the terms? That's marketing at its finest.)
With this development, marginal groups lose their social justice. The society evolves from a solidarity system that supports also the poor and the ill ones to a society that is egocentric and driven by money only.
As long as you are in the majority who benefits from this development, you much likely embrace your advantages. However, we need to understand that we are also going to be more and more selfish.
Every day, you can be overrun in an car accident. Every day, you can be diagnosed with a horrible disease. Every day, your digital identity can be stolen and misused. If you survive long enough, it is inevitable that you get old and need caretaking service. Then, the coin flips and you are going to find yourself on the losing side of this system. You paid the price of having consumed «personalized services» a few years. Now, you have to argue with a bank officer (if there are any left to talk to) to get a loan in order to afford the health treatment to survive the next months. And boy, your arguments are going to be weak to get this loan.
Do you really want to live in such a society? Think carefully for what you wish. Think carefully what you accept. Think carefully where you participate.
For now, I am on the «winning side»: I am well educated, I am not sick, I am young, I earn enough for a living. But I try to think of the big picture. I like to express my opinion at the elections. I am defending my privacy and my dignity even when this means financial disadvantages. Unfortunately, I also think that the majority or people is not sharing those values.
Thanks Alexander for your input. With your valuable discussion input I was able to express my point much better than in the original blog article.