Google turns to algorithms to create better-looking ads

google-teracent-ads

The constant push to make ads more visually appealing — reducing the effects of ‘ad blindness’ and helping them to convert better for advertisers and pay more to publishers — continues, as Google announces a pending deal with the Californian startup company Teracent.

Teracent specialises in machine-learning algorithms which have been put to use on the image ads of Google and its partners.

The example given in the announcement shows a standard, human created ad (left) and a similar ad that’s been picked in real time from thousands of different creative elements by these algorithms.

google-teracent-ads

[Read more…]

When Algorithms Go Bad: Facebook Recommends Exes and Deceased

FacebookI’ve already written today about Facebook’s new feature that shows you which of your friends might need a helping hand to gain more contacts or keep their profiles updated with photos or content, but apparently there are more disconcerting recommendations going on.

Think of it as the social network equivalent of receiving mail in the post for someone who has died (the exact reason for the problem is a bit different, but the distress is comparable).

The fact is, the algorithms that are supposed to help Facebook provide useful recommendations can only look at previous and current links between people, with no understanding of context.

That’s why Facebook has recommended that some users reconnect with ex-partners, or even with people who have died (presumably the deceased simply have dormant accounts on Facebook now).

Emotionally unsettling to say the least. [Read more…]