Recently Berlin was the stage for the FIBEP 49th annual World Media Intelligence Congress. As you might have noticed, Zoom Media joined the congress for the very first time as newly elected member of FIBEP and Ab and I were asked to share some thoughts. It was clear from the start that the buzzword of the congress would be: Artificial Intelligence. This theme has dominated technological discourse for some time now and it’s only logical that the monitoring industry took it upon itself to share thoughts and have discussions with regard to this ‘new’ phenomenon.
A.I. is here to stay
Artificial Intelligence is here to stay, that much is clear. However, there are a lot of misconceptions – or rather – false feelings of anxiety towards this type of innovation in all layers of the industry as well as society. Will robots take over my job? To what extent can we rely on machines to do our work? How do we make sure the outcome of research – or in this specific case monitoring – is accurate when we rely solely on machines? The same type of questions arose during the congress, but luckily the correct answers were also given. Like I said, Artificial Intelligence is here to stay. Sometimes it’s hard to accept that the world goes through a tremendous change, sometimes we rather adhere to the status quo. However try to think back to the days when the internet became a factor to be reckoned with, or more recently the mobile phone everyone has in its pocket. It’s not so much a question of should we do something with A.I.? Rather it’s a question of how do we work with what’s already there?
Man and Machine
One of the most persistent fears is that machines will take over from humans. However, many of us are also convinced that machines can’t do the work as good as humans can. So why are we so afraid of these machines, and why do we put so much trust in them? Ask any developer if machines can level with humans and he or she will most likely say ‘of course’. Now I’m not a fortuneteller, but I do think machines will get smarter and smarter up to a point where they can level with humans. However, for now, we are not there yet. As I said during my speech at the FIBEP congress: “machines are precise but stupid, humans inaccurate but smart.” For the coming years, we believe it’s not about choosing between one or the other, but rather combining the strength of both.
Human Feedback Loop
Now how do you make sure you get both of them – man and machine – together? At Zoom Media we’ve developed the process of what we call, the ‘Human Feedback Loop’. Output provided to us by our Artificial Intelligence engines is carefully edited manually by our employees. The corrected output is then fed back to the system to make it correct itself and ‘learn’ from its mistakes. This way we don’t leave the accuracy of the output solely to chance, but we make sure the machines are trained with results that are 100% correct. For example, a speech-to-text engine can recognize a word that wasn’t part of its lexicon as a word that ís part of its lexicon. Whilst machines cannot (yet) analyze context, humans can. So the word is corrected manually, fed back to the engine, and from that point on the engine is capable of recognizing this ‘new’ word as well.
Where are we headed?
So Artificial Intelligence is here to stay, well where are we headed then? Like I said I’m not a fortuneteller, however, I do believe that ignoring this phenomenon is not an option if we want to keep moving forward, and moving forward is also something we can’t stop. So I’d say embracing these changes is what’s needed first and foremost. Don’t be scared, machines are not about to take over, machines and humans working together can help us be much more effective and efficient. Think about this, how much do you use your mobile these days? That’s right, you’re already making use of Artificial Intelligence each and every day without even noticing it!