Artificial Intelligence
New AI reads news and removes bias from stories- Representational ImageCreative Commons

Knowhere, a tech startup, has built an artificial intelligence-powered news source designed to remove any color, spin, and political bias from news stories. It brings together human journalism and runs it through machine learning algorithms to deliver only facts on viral news stories.

The AI works by picking up a viral news story on the internet and comparing it to different other providers of news. It scans thousands of articles that cover the story from various angles, both liberal and conservative, and looks at as many possible sources as it can find, notes a report by Futurism.

After reading through several thousand stories, in as little as 60 seconds, the AI proceeds to "write" its own version of the event. It removes all leanings that most stories, based on where they are from, inherently carry. The AI's version of the same story will be delivered as only cold, bland facts, notes the report.

However, because the AI has to find a middle ground in reporting news, it is designed to take into account the trustworthiness of each news source. This was done by the makers of the AI to remove any chances of historically reputable news sites from being overshadowed by sites that might be loud and loose with facts.

For political stories, the AI even produces two additional versions of every news, apart from its own unbiased version, notes the report. They will be labeled "Left" and "Right" so that people who read them can understand why news is easy to twist.

An example provided in the report was based on the census collection issue in the US:

Left: "California sues Trump administration over census citizenship question"

Impartial: "US to add citizenship question to 2020 census"

Right: "Liberals object to inclusion of citizenship question on 2020 census"

Non-political stories like the Facebook scandal receive two additional versions that remove the positive/negative spin often applied to them.

Positive: "Facebook reveals that it scans Messenger for inappropriate content"

Impartial: "Facebook scans things you send on messenger, Mark Zuckerberg admits"

Negative: "Facebook admits to spying on Messenger, 'scanning' private images and links"

The AI also found that the way news is perceived by readers also depends on the header images. The images show a sad Zuckerberg for the negative story, but the image with the positive story has him smiling, setting the stage for what readers would feel when reading it.

Knowhere has just raised $1.8 million in seed funding, notes the Futurism report.