March 2, 2017
Will AI be able to moderate online discussions like humans?
Some artificial intelligence products have become so advanced in online discussion moderation that they will no longer be confused by colloquial language, neologisms or spelling mistakes. AI is able to take on routine human tasks, but cannot fully replace human intelligence.
Online discussions are abound with hate speech and off-topic comments, causing massive headaches for media companies. Legislation requires that illegal messages are removed, and users are more content if they can avoid becoming the target of inappropriate insults.
The volumes of comments posted on discussion forums and below news articles can be staggering, and their proper moderation may sometimes require infeasible amounts of manpower. Indeed, many news sites in Finland and across the world have closed down comment sectors of news articles as it has become almost impossible to moderate increasingly escalated discussions.
AI can learn from human decision-making
Until now, moderating has been carried out almost entirely by human hand, and traditional methods such as spam filters have not worked ideally in more automated moderation.
An AI product is capable of moderating massive amounts of content from discussion forums and news comments.
Finnish startup company Utopia Analytics has developed an AI product which, through machine learning, is capable of moderating massive amounts of content from discussion forums and news comments in a way that allows the automatic removal of inappropriate messages.
Sites have their own policies on what they consider inappropriate. Utopia Moderator learns from the individual moderation policy data of each site.
Utopia Analytics and Aller Media cooperated in a pilot study on AI moderation on the distinctly largest Finnish discussion forum, Suomi24.
Utopia Moderator acts as a sort of digital security officer supervising proper code of conduct on the site.
Previous moderation decisions by humans are used as learning materials for Utopia Moderator, and it is able to both memorize them and apply what it learnt in moderating new comments. Utopia Moderator acts as a sort of digital security officer supervising proper code of conduct on the site.
As the world constantly changes, so does the definition of appropriate behaviour. Utopia Moderator is brought up to speed on changes and additions to moderation policies by making a small sample of moderation decisions as new learning materials.
Utopia Moderator operates in any language and is not confused by the colloquial language commonly seen in social media, or by the constantly developing neologisms, nonstandard spellings or spelling mistakes.
It is able to take on routine human tasks but cannot replace human intelligence
The technology behind Utopia Moderator is based on statistical distributions: unlike a traditional spam filter or word list, it examines a comment as a whole and can identify its semantic content.
The background for this AI comes from the research and technology developed on machine learning in the academia, particularly in the Finnish Aalto University. In developing the technology, mathematical algorithms and data have been combined into a beautifully functioning moderation product.
Big data, artificial intelligence, machine learning and algorithms are are already affecting all of us indirectly.
Big data, artificial intelligence, machine learning and algorithms are part of the spearhead of technological development. They are already affecting all of us indirectly, but in the future, their role in an increasingly digitized society will be tremendous.
As it stands, AI has its own benefits and drawbacks – it will take at least decades for it to replace human intelligence, but it can be harnessed for completing routine tasks in data processing in countless ways and with high precision.
This development is not unlike that of the impact of factory assembly lines on the manufacturing of goods. The line produces standardized products in vast quantities: each purchased product is identical and in compliance with the customer’s expectations.
Similarly, AI is an assembly line of sorts, capable of processing data with the methods it has been taught. It clearly has its place in the moderation of massive volumes of online comments.
This article was originally published Wau.fi, a Finnish online publication covering all things digital.