Harry Potter, Hamlet, Catch-22 or The Great Gatsby all this are world famous novels, but where they all came from ? The simple answer is, our brain. Brain is an excellent artist which knows how to stitch thoughts and imagination in a life remembering stories.
But what if there will be someone else who also knows this secret art, what if a machine starts to tells us stories ? From childhood we have experience thriller, sad, horror and all types of stories, can a machines offers us something new ?
A recently developed neural network is capable of telling stories from a series of images, in a way that imitates human storytelling. The thing which makes it more different then other is it’s ability to make preferences about what’s happening inside a picture. It not only identifying and describing the objects, but pair them in a story.
It is created by team of researchers from UC Santa Barbara.
“Different from captions, stories have more expressive language styles and contain many imaginary concepts that do not appear in the images,” said the team on their whitepaper.
Recommended : AI Solve’s Rubik’s Cube In Just 4.22 Seconds
The neural network developed by researcher is called Adversarial REward Learning (AREL) framework. AREL is a special type of framework, it learn an implicit reward function from human demonstrations, and then it optimize the policy search with the learned reward function. It does not rely on an automatic evaluation system, thus it avoids cloning (and regurgitating) human efforts.
To test the AI, the team employed the humans of Amazon’s Mechanical Turk to conduct two separate tests. First, a Turing test, which simply asked the Turk workers to determine whether a story was created by a human or a computer. According to the researchers, AREL passed the Turing Test three our of five times.
In second test, the researchers asked Turk workers to pick between AREL, a human story, and one created by previous state of the art AI. Nearly half the time the human workers chose AREL. The result was about half of the human workers chose AREL.
The results show how neural network can be designed to better align with humans, and this should pave another straight path into the advancements of language processors. Developers figure out how to make the outputs generated by a neural network better align with human-thinking.
If we talk about the implementation of this story telling AI, Sports referees, for example, could either be replaced or augmented with an AI capable of understanding and explaining a series of events. Scientists can also use this to understand brain thought process and it’s ability of explaining decision and reasoning.
AREL isn’t quite ready for prime time yet though, the research clearly lays the groundwork for future endeavors to create a better neural network.
“We believe there are still lots of improvement space in the narrative paragraph generation tasks, like how to better simulate human imagination to create more vivid and diversified stories.” according to the researchers.”
More in AI :