AI Tracker: From Game Of Thrones To Biases, AI Had A Big Week

There are those who get excited about AI and those who fear it — and for everyone in between, there’s speculation. The AI industry is never short on innovations, but some people questioned this week whether making innovations quickly is a good thing. With speed, accuracy can suffer. However, in the world of AI, sometimes there are no limitations.

AI As Author

Envisioning someone writing a book typically brings to mind an author furiously typing away on a typewriter, filled with inspiration and devotion to the craft. But this week, that vision changed slightly when artificial intelligence was used to create the next book in the “Game of Thrones” series. Author George R.R. Martin takes his time finishing the next “A Song of Ice And Fire” book, but AI is predicting what may happen next, the New York Post reports.

“Game of Thrones” fan Zack Thoutt used his software engineering skills to develop an AI model that was able to “read” the existing five books in the series and create predictions of what could be next. The book’s initial five chapters can be read on GitHub.

Can AI Develop Biases?

If artificial intelligence is expected to react and behave like humans, it would stand to reason that the technology could have the same imperfections as humans as well, and a recent alert from Google shows how to avoid creating AI technology with inherent biases. In the past, AI has been known to show biases, such as associating the word “doctor” with males and not females, Quartz reveals. Google’s new video shows programmers how to avoid incorporating biases into AI technology.

The three main types of biases touched on in the video include interaction bias (which bases its behavior on how others interact with it), latent bias (which happens when the AI only has information that forces it to operate a certain way) and selection bias (which means the AI’s data is based on lopsided statistics). However, careful programming can seek to overcome these issues, according to Quartz.

How AI Impacts The Shopping Experience

Consumers might not yet see robots in retail locations, but shoppers’ lives are affected by artificial intelligence, perhaps even more than they think. In-store chatbots are being employed to increase engagement and help shoppers locate items within stores, said Nintex CMO Matt Fleckenstein in comments to Forbes. “Additionally, eCommerce has long leveraged AI for product recommendations,” he said, noting that McKinsey data reveal that 35 percent of Amazon purchases are derived from product recommendations created by algorithms that use past customer purchases, searched products and what others have bought to suggest additional items.

Is AI Advancing Too Quickly?

As artificial intelligence adapts and grows at breakneck speed, some researchers are starting to question whether that speed is a good thing. Some studies have begun to find cracks in AI’s accuracy, and in some cases, data show that humans still outpace AI in some activities. Now the Electronic Frontier Foundation has begun tracking AI progress and performance, Wired reports. “We want to know what urgent and longer-term policy implications there are of the real version of AI, as opposed to the speculative version that people get overexcited about,” said Peter Eckersley, chief computer scientist with the Electronic Frontier Foundation.