Blog

Machines that Feel?

As we look at the breathtakingly fast developments in artificial intelligence, we can’t help but wonder about artificial emotion: how long will it be before machines can feel? We’ve often thought that once motivation and desire can be programmed, look out. An army of nefarious bots will organize and take over, thinking they’re behaving for the greater good, of course.

One of the first steps has not only arrived, it’s quite sophisticated, and it’s called sentiment analysis. It goes beyond text mining and voice recognition to read the emotion that’s behind a writer’s or caller’s words. And through sophisticated algorithms (no surprise there), it’s already helping consumer marketers evaluate how people feel about their products, offers and messaging.

Is the comment genuine or sarcastic? Is politeness masking a deeper concern, or even resentment? Sentiment analysis can evaluate the context almost as accurately as humans, but in a heartbeat and at a fraction of the cost.

Consider the implications:

1.Quicker responses to social media chatter that could damage a reputation

2.Much more efficient voice-of-the-customer analysis

3.Evaluation of consumer or public sentiment that could move a stock’s price

We found this last one particularly intriguing. As it happens, a number of hedge funds are using sentiment analysis to predict the direction of an investment, especially if the sentiment is being expressed by someone influential. With the law of large numbers (or dollars) on its side, sentiment analysis in the service of tactical stock pickers doesn’t have to be perfect, it just has to be good enough.