Data Hoarding is Holding up AI Development

One of the more interesting applications of AI being developed is that of detecting human emotions. This is a remarkable idea, since computers being just a machine, they are totally lacking in emotion. So how can they detect emotion?

The answer is lots and lots of data. Indeed, high volumes of data is the necessary prerequisite to training any AI system. In the case of emotions, data can be gathered about the many human movements, gestures and actions that betray underlying emotions. Many are well known such as smiling, frowning, grimacing, grinning, folding arms, holding arms out, holding hands out, clenching them, hesitation in speech, vocabulary – the list goes on. Data can be gathered about these kinds of actions and related to the emotions being felt by the subject.

To build a useful AI application, and train the application to recognize the various emotions indicated by the physical displays, vast amounts of data must be gathered. A popular source of such data is social media, because it catches people in unguarded moments and reveals their true emotions. Not always, but that’s one reason for the need of large amounts of data.

Gathering good data takes a lot of work, and even when someone gathers enough to be useful for AI, they may not make it available, because they know how valuable the data is and hold out for a good payback.  As is often said, data is the new oil.

For a good summary of the role of data in AI and its usefulness for emotion detection and medical research, check out this link. https://informationweek.com/big-data/ai-machine-learning/how-ai-can-save-the-world-or-not/d/d-id/1340217?

Leave a comment

Comments

  • No comments found