Facebook is stepping up its strategy against undesirable websites and ads that are inundating the internet with offensive material. FB is rolling out new upgrades in its News Feed AI algorithms that are designed to recognize posts and ads that link to pages with offensive or “low quality” content, according to Marketing Land.
The new algorithms are developed with machine learning as it works to cull out the undesirable “low quality” ads and posts before they are approved for viewing.
Pages that link to ad-heavy sites or other offensive material are being targeted on FB, Instagram, and FB’s Audience Network, but excludes organic posts on Instagram because the safeguards already in place for that site have been effective.
The AI programs have already been “trained” to identify the traits of deleterious content by running them through a series of known bad-ads (many thousands of them), allowing the machine learning process to sniff them out like a drug dog searching for heroin.
There are any number of variant signs that might indicate a “bad-ad” but one of the major clues is pop-up ads that intrude upon the viewer’s desired content: for instance, if the user clicks on a link that promisedinformation about a cure for cancer but instead the link brings up ads about better sex pills.
One positive result: legitimate ad sites will likely see more traffic than before as the undesirable detours are taken out of the user’s journey. FB stated that no plans are in the making for ad-testing to make sure the quality standards are met because spammers can use such a tool to get through the new filters.