SxSW 2017: AI and the boys from Silicon Valley
Diversity. This is an important topic during the largest tech conference in the world. No idea what the exact numbers are, but this year it feels like the SXSW has succeeded in filling only half of the programme with white males in their forties. And, taking a quick look at the programme, it quickly becomes clear: diversity continues to be an important theme. Good news, because something else is going on.
We now live in a time where intelligence is all around us. As my colleague Danijel Bonacic said during his SXSW talk: “You can see it present in your Spotify Daily Mix, or your Google Mail that drafts a reply to an e-mail for you.” At this time, they are still small, sporadic stratagems. But with the help of the cloud, we are getting closer and closer to an all-encompassing AI. It is crucial to understand that AI must be trained in order to truly be of added value. We start with rules, after which the system makes itself smarter by using all available input. An illustration: Google now knows if someone is laughing or crying on your photograph but, in order to learn this, at one point, someone had told the AI what the features of a laugh were and thus had also determined the requirements for that laugh. By subsequently feeding in millions of photographs of laughing people, the system learns to better recognise this pattern.
Thus AI is all about data sets. But if these datasets are not ‘broad’ enough, errors occur in the system. Right now, for the most part, the ‘rules’ are determined by these 40-year-old white males. What’s more, it is often the early adaptors – in general, the highly educated target groups – that initially use and feed the AI services. In itself, it is not surprising that the system subsequently acquires a rather narrow view of the world.
Already in 2015, scientists discovered that women saw fewer advertisements on Google for well-paid jobs. The complexity of search engines makes it difficult to say why this happens. Either the advertisers have a preference for males, or this is an unintended result of the Google algorithms. But it is a fact that this is not necessarily very beneficial for the women looking for these well-paid jobs.
With all the #fakenews discussions, we need not look very far to see that this is happening abundantly around us. Our Facebook timeline is a reflection of what we feed the algorithm, and this makes it all the more difficult for us to remove our blinders and to show understanding for people outside of our bubble.
AI is everywhere. And it is less and less about code and more and more about strategy. About how you as a brand can better serve your users and can integrate those little stratagems that already make our lives easier now. To that end, we are required to listen and to closely study those users, all of them. And not just to the boys from Silicon Valley.