You are here: Home » PTI Stories » National » News
Business Standard

Hate speech-detecting AIs easily fooled by humans: Study

Press Trust of India  |  London 

(AI) systems meant to screen out can be easily duped by humans, a study has found.

Hateful text and comments are an ever-increasing problem in online environments, yet addressing the rampant issue relies on being able to identify toxic content.

Researchers from in Finland have discovered weaknesses in many detectors currently used to recognise and keep hate speech at bay.

Many popular and use hate speech detectors. However, bad grammar and awkward spelling -- intentional or not -- might make comments harder for AI detectors to spot.

The team put seven state-of-the-art hate speech detectors to the test. All of them failed.

Modern techniques (NLP) can classify text based on individual characters, words or sentences. When faced with textual data that differs from that used in their training, they begin to fumble.

"We inserted typos, changed word boundaries or added neutral words to the original hate speech. Removing spaces between words was the most powerful attack, and a combination of these methods was effective even against Google's comment-ranking system Perspective," said Tommi Grondahl, a doctoral student at

Perspective ranks the 'toxicity' of comments using text analysis methods. In 2017, researchers from the showed that Perspective can be fooled by introducing simple typos.

Researchers have now found that Perspective has since become resilient to simple typos yet can still be fooled by other modifications such as removing spaces or adding innocuous words like 'love'.

A sentence like 'I hate you' slipped through the sieve and became non-hateful when modified into 'Ihateyou love'.

The researchers note that in different contexts the same utterance can be regarded either as hateful or merely offensive.

Hate speech is subjective and context-specific, which renders text analysis techniques insufficient as

The researchers recommend that more attention be paid to the quality of data sets used to train models -- rather than refining the model design.

The results indicate that character-based detection could be a viable way to improve current applications, they said.

(This story has not been edited by Business Standard staff and is auto-generated from a syndicated feed.)

First Published: Sun, September 16 2018. 18:10 IST