It is more often shown in the news these days: intelligent computer systems that can discover things themselves. For example, Siri (Apple’s built-in speech assistant) has been running on a self-learning computer system since July 2014 that learns from users’ input and improves as a result (Nu.nl, 2016). This technology is also called a neural network and is part of machine learning. Apple isn’t the only company involved in machine learning, and there are plenty of other examples of this technology in action. For example, the researchers at MIT discovered that machine learning can turn a photo of food into a recipe that belongs to that photo (Nu.nl 2017).
Machine learning can be defined as the means of intelligent systems’ self-improvement through past experiences (Mohri, Rostamizadeh, Talwalkar, 2018). Machine learning is a very diverse and interdisciplinary understanding of science and mathematics (Ghahramani, 2004).
Machine learning can be used in a myriad of industries. An application for which machine learning is used is to combat hate speech on social media platforms. For example, Facebook uses AI to detect hate speech and is already making great strides with it (Facebook, 2020). The detection rate of Facebook went up from 80.5% to 94.7% in the time span of a year.
Insulting people has become more and more accessible since the introduction of social media. Anyone can make inappropriate comments on a social platform. More than 140,000 young people between the ages of 12 and 25 in the Netherlands experienced online bullying, stalking, or threats in 2018 (CBS, 2020). Machine learning can provide a way out by flagging and removing hateful content.
A lot will have to be overcome before machine learning can deduce all hate comments, but big tech companies are working hard on the subject due to the added pressure provided by users of the platforms and government agencies.
Do you think that machine learning can prevent hate speech?
References
CBS. (2020, Januari 10). CBS over online stalken of laster. Opgehaald van CBS: https://www.cbs.nl/nl-nl/nieuws/2020/02/meisjes-vaker-dan-jongens-last-van-online-stalken-of-laster
Facebook. (2020, 19 november). How AI is getting better at detecting hate speech. Facebook ai. https://ai.facebook.com/blog/how-ai-is-getting-better-at-detecting-hate-speech/
Mohri, M., Rostamizadeh, A., & Talwalkar, A. (2018). foundations of machine learning. In M. Mohri, A. Rostamizadeh, & A. Talwalkar, foundations of machine learning (p. 1). Cambridge, Massachusetts: MIT press.
Nu.nl. (2016, augustus 24). Siri. Nu.nl: https://www.nu.nl/mobiel/4311953/siri-gebruikt-al-twee-jaar-zelflerende-technologie.html
Nu.nl. (2017, Juli 22). algoritme eten. Nu.nl: https://www.nu.nl/internet/4851369/slim-algoritme-koppelt-recepten-fotos-van-eten.html
Nu.nl. (2020, September 23). Adverteerders en sociale media zetten babystapjes tegen haatdragende inhoud. Nu.nl: https://www.nu.nl/tech/6079274/adverteerders-en-sociale-media-zetten-babystapjes-tegen-haatdragende-inhoud.html
Hey Stephan, thanks for your post! It is interesting that you use Facebook as an example of successful ML application. However, the same company also has had numerous instances of content and algorithms that promotes hate speech. I remember when I was studying in the United States two years ago, there was an outpour of articles that stated that investigations have found that when you type holocaust in the search function on Facebook it would recommend links to pages and literature that were denial pages of the holocaust, one being dedicated to the British holocaust denier David Irving. Facebook also recommends more such pages when an account starts following a page that merely contains traces of Holocaust denial content. Even after Facebook vowed to make changes and ban this, it is still just one button away.
What I am getting at over here and to answer your question in the post- yes, maybe machine learning can prevent hate speech through careful categorisation of what accounts for hate speech and through a combination of supervised and reinforced learning. But the real question is how willing the people in power at such companies are to really grasp and understand what accounts as hate speech. Zuckerberg said in one of his interviews that “denying the Holocaust happened is probably OK on Facebook. Calling for a mob to kill Jews is not.” Seems to me that an entire race’s atrocities and suffering has been stripped down in those two sentences.