Pinterest uses AI to protect users from damaging content

As part of its ‘Getting better at helping people feel better’ campaign, aimed at increasing the well-being and safety of its users, Pinterest has announced the roll out of wider spread AI interaction across the platform.

Intended to remove and reduce exposure to harmful content, this is an expansion on the powerful Pinterest AI tools already in place.

However, does this really mean pinners are safer?
Should we be asking bigger questions of Pinterest about how AI is managing our pins, and whether this open access to our data is really in our best interests?

Pinterest has announced the introduction of more robust screening tools and AI integration across the platform. Photo: Pinterest Newsroom Gallery

AI in social media

Artificial Intelligence is nothing new in the social media landscape, nor of course is damaging content. Any platform allowing public access without having a means of robustly screening individuals will inevitably attract some abuses. AI has long been hard at work in the background to implement filters, capture ‘trigger’ phrases and words, and more recently to attempt to direct users to support when search terms are flagged.

As reported on Expert, LinkedIn recently bought out a small AI enterprise called Bright to enhance its AI capabilities. Tech Crunch reported the Facebook investment into a research facility dedicated to developing AI functionality, and Google acquired DeepMind to focus on analysing user reactions and behaviours.

Pinterest is already a platform with a strong AI presence, with its Lens tool now being able to recognise over 2.5 billion objects.

How does AI serve social media platforms?

The uses of AI are widespread, from tracking and analysing consumer trends, creating pathways for referrals and redirects to support services, and plotting user behaviour and reactions to optimise user experience.

Of course, there is a cynicism for the way in which social media controls user content, collects personal data and information, and uses this to drive advertising. Nearly all social media platforms are profit making enterprises, and the more they know about what makes their users tick, the more targeted and slick their advertising mechanisms become.

Pinterest leverages powerful AI technology to identify over 2.5 billion objects, and create tracking and relatable imagery around them. Photo: Pinterest Newsroom Gallery

What changes is Pinterest making?

As part of their new campaign, announced in conjunction with World Health Day, Pinterest has introduced more robust AI tools which have an enhanced capacity to identify, remove and avoid directs to any content which may ‘display, rationalise or encourage self-injury’ (Pinterest Newsroom).

With a focus on self-harm, these augmented controls look to identify potentially harmful or triggering content.

The other expansion is to incorporate more sophisticated referrals, where pinners searching for or using key phrases, words or imagery highlighted as potentially harmful are automatically directed to support services and information.

What AI tools does Pinterest use?

The Pinterest AI works through PinSage, a neural network which creates a complex context for every image, allowing it to be grouped, mapped and categorised alongside other visually or thematically relevant pins. For further reading about how PinSage works check out Wired, which goes into greater detail.

Pinterest reports that new controls have seen an 88% reduction in access to harmful content. Photo: Pinterest Newsroom Gallery

So – is this a proactive initiative to be celebrated?

Pinterest seems to have taken a cohesive approach to rolling out their enhanced AI offering, working with the Samaritans, National Suicide Prevention Lifeline and Vibrant Emotional Health to develop and introduce these new safeguards (Pinterest Newsroom). This in itself speaks volumes and is a refreshing change from the often heavily guarded and insular attitude of social media platforms on the functionality and uses of their systems.

The impact has been a reported 88% reduction in access to harmful content, and a response rate three times faster than before (VentureBeat.com). These are impressive statistics.

All in all, this seems to have been a well thought out and considered roll out, which no doubt will have invisible impact on users so needing meaningful action to be taken to safeguard them; those who are often most vulnerable in our society.

Whilst action taken by social media platforms such as Pinterest to protect and look out for the well-being of their users has to be a positive thing, the thread of doubt still weaves in the maelstrom of controversy over the profitability and data harvesting of social media platforms – how much of this is real, and how much is well-timed PR…

0 0 vote
Article Rating
0 Shares:
Subscribe
Notify of
guest
0 Comments
Inline Feedbacks
View all comments
You May Also Like
Read More

How does WeWork make money?

WeWork has made quite a splash in recent months, achieving an eye-watering valuation, and generating plenty of headlines.…