Feb 20

The (problematic) Power of Algorithms

Twitter, Youtube, Google, the App Store - algorithms determine what users see with all the consequences, but without any regulation.

The rise of algorithms

With the huge amounts of content, apps and products on the various platforms and the need for monetization, companies tried to find ways to increase the income flow. Platforms like Twitter, that did not make any money for a long time, decided that ads are the way to go to earn money with online services. They could have tried to just ask their users to pay for the service, but for reasons unknown advertisement was the way to go for most platforms. While ads might be annoying, there seems not much drama in this decision on the first sight.

Unknown consequences

If ads are your business model, you want the users to stay on your platform as long as possible. If you would just charge your users with a subscription, you could be much more relaxed about how long a user spends on your platform. So someone in charge of finances tells a data scientist, please write an algorithm that makes the user spend more time on the page. The manager does not think about fake news or hate speech, the data scientist does only think of the technical implementation. In the end those platforms are no news companies, but tech companies.

What I see as a user:

Your mileage may vary, but my feeling is, those algorithms and the ad business did not have a good impact on society. A feeling, you might say - prove it!

Well I can't, and that is the problem.

Intransparency

Every platform makes a huge secret of their algorithm, they even use it in court (Google) to hide behind it, like it was some uncontrollable, life like thing that does whatever it wants (and to give the user the best experience, of course /s). Well, sounds like a neural network.

These neural networks depend a lot on the data you feed them and their decisions are hard to understand. Even if you know exactly how the algorithm works, training it with different data can completely change the results. Face detection systems that do not detect black faces, because the developers where probably white are just one case.

It is funny that you can often find comments on Youtube videos like "May the algorithm be with you", or "Left a like for the algorithm". Well it is funny to read as a viewer, less funny as a creator who tries to make a living out of it.

Will the problem solve itself?

I myself notice fatigue when beeing on Twitter or Youtube, because all the content is somewhat polarizing. I could imagine that users turn their back on services that they consider toxic in the long run. The rise of meditation apps somewhat proves the point. But I cannot think of any case where self regulation of a company is working. In the end these are all businesses trying to make money, and they will probably not make less money just out of bad conscience. Just look how long it took until the former US president got canceled..

How to solve

In my opinion, transparency about the business models would be the first step. Algorithms are complex and hard to predict, but it would be nice to know what metrics go in there to know as a user, why you see content. That would also give content creators the possibility to understand why their website is on page two on google, or their video does not get any views etc.

In my case it would be simple - We use algorithms to detect similar articles like so:

Powered by nooshub.com.

Turns out AI ethics were a hot topic today.

The consequences of algorithms on society make clear, that ethics are an important factor. If algorithms replace human decisions, they must hold up to the same standard. A journalist cannot put fake news on the start page without consequences. Tech companies cannot hide behind being just tech companies, when they use algorithms to decide which content is shown. I could understand the argument, if they would just output stuff that users put in as a technical service, but this is not the case.

The control and transparency of algorithms is a complicated and important topic. In my opinion it needs to be solved by politics and not by self regulation, better sooner than later!