Trolls beware! Twitter will now prompt users to review and revise any ‘potentially harmful or offensive’ replies before they go live
- The feature uses AI to detect offensive language in drafts of newly-typed replies
- It sends users a notification asking them if they’d like to review their message
- Users are also free to ignore the warning message and post their reply anyway
- The feature was tested last year but has now been rolled out for Android and iOS
- Twitter’s improved AI detection means the feature accounts for ‘friendly banter’
Twitter has rolled out a feature that prompts users to review their ‘potentially harmful or offensive’ replies before posting them – although it accounts for ‘banter’ between two mates.
The feature, which was first tested last year, uses artificial intelligence (AI) to detect harmful language in a freshly-written reply to another user, before it’s been posted.
It sends users a pop-up notification asking them if they want to review their message before posting.
According to Twitter, the prompt gives users the opportunity to ‘take a moment’ to consider the tweet by making edits or deleting the message altogether.
Users are also free to ignore the warning message and post their reply anyway.
The feature was first tested for a limited period this time last year, and then reintroduced for iOS this February.
Upon typing out an expletive-strewn reply, Twitter users will see the pop-up, entitled ‘Want to review this before tweeting?’ It adds: ‘We’re asking people to review replies with potentially harmful or offensive language.’ Users can then tap on three options – Tweet, Edit and Delete. They can also click on an option to let Twitter know they think it was wrong to send the notification
HOW DOES IT WORK?
Upon typing out an offensive, derogatory or expletive-strewn reply, Twitter users will see the pop-up, entitled ‘Want to review this before tweeting?’
It adds: ‘We’re asking people to review replies with potentially harmful or offensive language.’
Users can then tap on three options – Tweet, Edit and Delete.
They can also click on an option to let Twitter know they were wrong to send the notification.
After the feature successfully reduced the amount of offensive tweets sent on the platform, Twitter has now rolled it out globally for an users using the English-language version on both iOS and Android.
‘People come to Twitter to talk about what’s happening, and sometimes conversations about things we care about can get intense and people say things in the moment they might regret later,’ Twitter said in a blog post on Wednesday.
‘We began testing prompts last year that encouraged people to pause and reconsider a potentially harmful or offensive reply.
‘Based on feedback and learnings from those tests, we’ve made improvements to the systems that decide when and how these reminders are sent.’
Twitter confirmed to MailOnline that the feature is currently just for replies – not for all tweets.
The ability for two people who have never met to send toxic messages to each other on Twitter is one of the main criticisms of the platform.
For this new rollout, Twitter has worked to improve the feature’s AI detection capabilities so it can differentiate nasty and ‘harmful’ messages, such as between two strangers, from ‘friendly banter’ between two mates.
What this means is two friends can continue to have expletive-strewn – albeit good-natured – exchanges on the platform without getting the pop-up notification.
‘In early tests, people were sometimes prompted unnecessarily because the algorithms powering the prompts struggled to capture the nuance in many conversations,’ Twitter said.
‘[They] often didn’t differentiate between potentially offensive language, sarcasm, and friendly banter.
‘Throughout the experiment process, we analysed results, collected feedback from the public, and worked to address our errors, including detection inconsistencies.’
When a Twitter user sees the notification and taps ‘Did we get this wrong’, users can answer by tapping ‘yes’ or no’
Despite this, the initial tests of the feature on iOS since last year proved successful, Twitter said.
If prompted, 34 per cent of people revised their initial reply or decided to not send their reply at all.
And after being prompted once, Twitter users composed, on average, 11 per cent fewer offensive replies in the future.
In the blog post, Twitter said it will ‘continue to explore’ how prompts and other forms of intervention ‘can encourage healthier conversations on Twitter’.
‘Our teams will also collect feedback from people on Twitter who have received reply prompts as we expand this feature to other languages,’ it said.
In a similar vein, Twitter is working on an ‘Undo Send’ timer for tweets that will give users five seconds to rethink your message, it previously confirmed to MailOnline.
Twitter is working on an ‘Undo Send’ timer for tweets that gives you five seconds to rethink your message
The blue bar is a timer that runs out after five seconds. App researcher and tipster Jane Manchun Wong posted the animation of the leaked feature to her Twitter account
Twitter has confirmed it is working on an ‘Undo Send’ feature that will give users a chance to cancel their tweets if they have second thoughts.
The feature leaked online in March 2021 in the form of an animation, showing a blue timer with the word ‘Undo’ that counts down after a user has posted a tweet.
Twitter users will have just five seconds to tap the timer before the tweet is made public.
Some Twitter users pointed out that it would just mean there would be a frustrating five-second delay before their tweets actually appear.
Twitter confirmed to MailOnline that it is currently ‘testing’ Undo Send, but declined to provide any further details.
It’s not known when the nifty new feature will be rolling out, or if it will come to Twitter for both desktop or mobile.
But it would be sure to reduce some of the most vociferous verbal abuse on the platform by giving users a chance to think twice before they post.
Twitter still isn’t yielding to user demands to implement an edit button, however – its most-requested feature.
Source: Read Full Article
