Hate Speech Detection

February 1, 2019

Hate speech runs rampant on the Internet, especially on social media and Web sites. Trying to contain hate speech is like trying to drain the ocean with a garden hose. Several tech companies are trying to reign in the hate speech, but their attempts stink. Digital Trends focuses on how hate speech AI technology is in the pits in the article, “Current Tech For Detecting Hate Speech Is Woefully Inadequate, Researchers Find.

The problem is not the tech companies, but their technology. Researchers from the Aalto University in Finland analyzed hate detection tools. They discovered that none of the tools could agree on what qualifies as hate speech and that they are stupid. The hate detection tools were easily fooled with typos and letter substitution. Humans are still needed to interpret the true meaning of words and their context. For example:

“The researchers next demonstrated how all seven systems could be easily fooled by simple automatic text transformation attacks — such as making small changes to words, introducing or removing spaces, or adding unrelated words. For example, adding the word “love” into an otherwise hate-filled message confuses detection systems. These tricks were capable of fooling both straightforward keyword filters and more complex A.I. systems, based on deep-learning neural network architectures.”

Computers still cannot compete with humans when it comes to understanding and interpreting human emotions. Hate speech detection will improve as more developers research and experiment with the AI.

Whitney Grace, February 1, 2019


Comments are closed.

  • Archives

  • Recent Posts

  • Meta