Making sense of things

Making sense of things with limited context 

Iris.AI tries very hard to figure out the key concepts embedded in all the great science-related content we are continuously feeding her.

But she is a young AI, and we still have a long road ahead before she can enlighten users with a perfect grasp of a given text-based input.

Take this example. Judging by

Judging by previous TED speakers’ accounts, French mathematician Cédric Villani probably put a lot of effort crafting this very interesting TED Talk: What’s so sexy about math?

What did Iris.AI think of it when she read the transcript? What was the talk about, in her mind?

Weeelll. Sexy? Really? The comparison between pleasure derived from finally understanding the right reasoning to solve a mathematical problem and sexual pleasure was not picked up. In fact, it went completely over young Iris.AI’s head.

The comparison between pleasure derived from finally understanding the right reasoning to solve a mathematical problem and sexual pleasure was not picked up. In fact, it went completely over young Iris.AI’s head.

Iris.AI today struggles with elements such as metaphors, figurative speech or ironic remarks built within scripts. But then, didn’t you back in the day?

Through overlaying supervised training efforts on top of the currently deployed unsupervised NLP algorithms, our goal is for her to get better at these things over time. Much like for humans, the more context the brain is exposed to, the better its ability to make sense of things.

If you haven’t yet to joined our AI Trainer program, do so now.