Slowly but surely, artificial intelligence (AI) is advancing. Where it stands in comparison to human intelligence is difficult to say because people excel across a broad set of intellectual tasks, while AI tends to be narrowly focused. But machines keep getting better at tasks like sifting through vast quantities of data, understanding natural language, and recognizing objects in images.
Over the weekend, scientists from a Silicon Valley company called MetaMind described their efforts to advance the state of the art in a research paper, "Dynamic Memory Networks for Visual and Textual Question Answering."
The paper explains improvements in memory and input modules for a system called a dynamic memory network (DMN), a general architecture for answering questions about text and images. A DMN might, for example, parse text describing a series of events in which a person passes through a series of rooms and drops an object in one of them. If asked about the location of the object, the DMN could infer the correct answer from the story.
MetaMind's researchers have devised a way to have their data model learn the facts it needs to reason without labeling those facts in a training session. Their approach involves creating a component that considers text in a bidirectional manner, rather than analysis that flows one way from beginning to end.
By allowing both past and future sentences to be considered when attempting to reason about a specific situation, the researchers see more accurate results. The technique also can be used to improve the identification of objects in pictures.
Such domain-specific competency has become increasingly interesting to companies as they see ways to enhance business processes through machine learning, neural networks, and related disciplines.
Companies like Apple, Facebook, Google, Microsoft, and IBM have made it clear that they consider AI a source of future growth. Enthusiasm for the technology can also be seen in AI investments from firms like Goldman Sachs and JPMorgan. According to research firm CB Insights, the funding of AI companies has grown from $45 million in 2010 to $310 million in 2015.
"Through visual question-and-answering, manufacturing companies can monitor and maintain equipment remotely to address problems before they exist," said Michael Machado, senior product manager at MetaMind, in a blog post.
"Enterprises are looking to leverage our tools to streamline customer service operations, saving money and improving response time. Forward-looking decision makers strongly believe that AI solutions that find obscure, sometimes hidden signals are quickly becoming their most important strategic tool."
While incremental advances in AI may matter mainly to computer scientists, business customers can still benefit without fully understanding every algorithmic improvement. MetaMind offers AI to customers in the form of cloud service APIs.
The company's technology allows developers to create apps that can perform feats like labeling objects with their names translated into a foreign language, evaluating road congestion from webcam images, or responding to sad Twitter posts with messages of encouragement, to name but a few.
Other companies also offer AI on tap too. Facebook has offered developers access to the voice recognition API it acquired through Wit.ai. Google provides a Prediction API, TensorFlow, and the Cloud Vision API. Microsoft has Azure Machine Learning.
AI represents the technology industry's solution for data pollution. With all the data that businesses and individuals generate, there's a pressing need to improve machine intelligence to sort the mess and make sense of it.