The reach of hate speech on Twitter continues to represent an extremely small fraction of the overall conversation, the Twitter Safety account said stating Sprinklr study
Twitter on Wednesday shared the latest data revealing a reduction in the prevalence of hate speech on the platform, with Sprinklr’s independent model estimating daily English-language hate speech impressions at 0.003% compared to Twitter’s own estimate of 0.012%, for the period of January 1, 2023, to May 31, 2023.
More than 99% of content users and advertisers see on Twitter is healthy and the reach of hate speech on Twitter continues to represent an extremely small fraction of the overall conversation, the Twitter Safety account said.
“This past March, Twitter partnered with Sprinklr to understand, measure and reduce hate speech using its AI-based model and to further our commitment to creating a brand-safe environment for our agencies and advertisers. Today we are sharing an update that includes all of April and May 2023,” said the Twitter Safety account.
Sprinklr’s independent model continues to show the reach of daily English-language hate speech impressions is even lower than Twitter’s own model estimates, it added.
“Sprinklr estimates the average daily number to be 0.003% compared to Twitter’s estimate of 0.012% for the period of January 1, 2023, to May 31, 2023. Additionally, we estimate hate speech impressions are 30% lower on average versus pre-acquisition. In June, we significantly expanded our Freedom of Speech, not reach policy enforcement, which dramatically reduces impressions on harmful content and all indicators for June and July show sustained progress,” the Twitter Safety account stated.
“Our work is ongoing and we are proud of the progress we are making in maintaining a healthy global town square that is open for everyone to be themselves,” it added.
However, users on the platform continue to demand more transparency and clarification on this. Some even requested clarification on what exactly the platform means by hate speech.
“Then why does Twitter continue to censor information? Why does Twitter send drafts of shared articles from reputable publications until the graphic is removed? That’s a short list. I think Elon Musk needs to take charge and clean things up again. I know people who are being censored who shouldn’t be. Meanwhile, others can post all manner of offensive garbage with impunity. Differences of opinion are fine. Make your case rather than hurling invectives even if they too are protected speech,” a user commented.
“How does Sprinklr define hate speech? Control over that definition is very important. Just because they implement AI wouldn’t necessarily make it neutral either, something to monitor,” Eric Sanchez commented.
Another Twitter user, Mark Reardon, wrote that complete transparency is needed regarding Twitter’s definition of hate speech first only then this information will be useful.
“Elon Musk, what’s the definition of “hate speech” and healthy/harmful content? It’s definitely important to strike a balance of freedom of speech and quality of content but it seems limiting the rate and the impressions is not a wise solution,” Ceci commented.
