Stanford’s research paper, entitled “Deep Visual-Semantic Alignments for Generating Image Descriptions,” explains how specific details found in photographs and videos can be translated into written text.
“We present a model that generates free-form natural language descriptions of image regions,” the paper’s abstract states. “Our model leverages datasets of images and their sentence descriptions to learn about the inter-modal correspondences between text and visual data.”
When comparing the software’s descriptions to that of humans, researchers found that both often expressed similar comprehension.
Stanford software analyzing individual items
Artificial intelligence software creating picture captions
Google’s version of the technology, presented in the “Show and Tell: A Neural Image Caption Generator” research paper, was able to produce comparable results.
Although the program struggled when presented with an entirely new set of images, Google scientists noted the software’s ability to “learn” from each new interaction.
“I was amazed that even with the small amount of training data that we were able to do so well,” Google computer scientist Oriol Vinyals told the New York Times.
Billions of videos and images posted online will likely be analyzed and tabulated by Google once the technology is perfected, allowing real-time cataloging as media is uploaded to the Internet.
“The field is just starting, and we will see a lot of increases,” Vinyals added.
Although Google scientists noted the possible uses for the visually impaired, the company’s deep ties with the U.S. surveillance state will likely dictate the software’s final application.
Google software learning from its mistakes
As Google quickly becomes one of the most powerful lobbyist groups in Washington D.C., many note the technology’s potential for censoring data even before it has a chance to be uploaded to the web.
Google’s history of censoring political content has been long documented, with infamous videos such as the WikiLeaks helicopter gunship footage being removed from the Alex Jones’ YouTube channel in 2010.
In fact, Google has complied with thousands of requests from governments across the globe to censor and remove content despite no clear violations in many of the cases.
Former Google CEO Eric Schmidt, a campaign advisor and major Barack Obama donor, has publicly stated his disdain for human privacy on multiple occasions as well.
“We don’t need you to type at all. We know where you are. We know where you’ve been. We can more or less know what you’re thinking about,” Schmidt said in 2010. “We know everything you’re doing and the government can track you. We will know your position down to the foot and down to the inch over time.”
Whether it’s pushing customers to put microphones in their ceilings, participating in the NSA’s PRISM program, or scanning every private Gmail account, the tech giant will undoubtedly continue to wield its state-backed power over millions of web users worldwide.