Instagram‘s algorithm is recommending COVID-19 and anti-vaccination misinformation to potentially millions of users, according to new research.

The Center for Countering Digital Hate (CCDH) used test accounts to investigate the recommendations on Instagram’s Explore page and new Suggested Post feature.

They found that the tools encourage users to view misinformation and then push those who engage with the posts towards other extremist content:

The researchers generated the recommendations by creating 15 Instagram profiles and following different lists of accounts, from health authorities to anti-vaxxers.

They logged into the Instagram accounts every day and recorded the recommendations they received.

As the Suggested Posts feature doesn’t trigger for new accounts that haven’t interacted with posts, the users scrolled through their feeds and the Explore section, and liked random posts to generate suggested content.

They then screenshot the recommendations they got between 14 September to 16 November 2020.

In total, Instagram recommended 104 posts containing misinformation. More than half of them were about COVID-19, while a fifth were about vaccines, and a tenth about the US election.

Users also received recommendations for posts promoting the QAnon conspiracy theory and antisemitic content. The only profiles that weren’t recommended misinformation exclusively followed recognized health authorities.

Imran Ahmed, the CEO of CCDH, said that Instagram is purposefully promoting extremist misinformation:

CCDH has published an open letter to Mark Zuckerberg, the CEO of Facebook, which owns Instagram, urging him t o disable and fix the “broken algorithm.”

In response to the report, a Facebook spokesperson said the research was five months out of date and based on “an extremely small sample size” of 104 posts.

However, Instagram users have pointed out that Instagram is still recommending misinformation today.

You can read the CCDH report here .

After years of testing, Netflix confirms it’s launching a shuffle play feature this year

If you’re struggling to find something worth watching from Netflix’s vast catalog and you don’t trust the recommendations of your friends, the streaming service has some good news for you.

The company is finally launching a “shuffle play” feature for users who would rather let an algorithm pick their next series to binge on.

The streaming service confirmed it will roll out the feature globally during a Q4 investor interview on Tuesday.

“Our members can basically indicate to us that they just want to skip browsing entirely, click one button, and we’ll pick a title for them just to instantly play,” said COO and chief product officer Greg Peters.

Peters didn’t specify a launch date for the feature, but Variety reports that it will arrive in the first half of this year.

Netflix has been testing the function since at least April 2019, when viewers started spotting a “Random Episode” label in their playback controls.

Further rounds of testing followed last year. In August, a number of users noticed a Shuffle Play button pop up on Netflix’s TV app, promising to “find things for you to watch based on your tastes.”

Surrendering your viewing choices to an algorithm might sound risky (or even dehumanizing). But Netflix has already shown the value of its AI-powered suggestions.

The company says more than 80% of the shows watched on the service are discovered through its recommendation system, which analyzes your viewing habits to find new shows you might like.

In an interview with Wired, Todd Yellin, Netflix’s vice president of product innovation, compared the system to a three-legged stool:

The new shuffle feature remains unnamed now. Netflix Co-CEO Reed Hastings jokingly suggested calling it “I’m Feeling Lucky,” a nod to an old Google button that took users directly to the top result for their search.

New AI tool accurately diagnoses COVID-19 just by the sound of your coughs

Scientists from Essex University have created a COVID-19 screening tool that can accurately diagnose the virus by analyzing the sound of a person’s cough.

The researchers say it could be used in a smartphone app to provide a more comfortable form of detecting the virus than eye-watering swab tests.

The tool, named DeepCough3D, uses AI to analyze audio samples of coughs in frequencies that humans can’t hear.

The researchers tested it on over 8,000 samples of people coughing in hospitals in Spain and Mexico since April 2020. Around 2,000 of the patients were COVID-19 positive, while the remainder had tested negative.

DeepCough3D proved 98% accurate at identifying whether the samples were positive or negative.

Lead researcher Dr Javier Andreu-Perez said the tool could “prove a real game-changer” in how we combat the pandemic:

DeepCough3D is one of numerous attempts to diagnose COVID-19 by listening to coughs. Notably, MIT researchers recently developed an algorithm that successfully detected around 98% of COVID-19 infections by people with COVID-19.

However, the Essex University team say their research stands out from other studies because it’s proven highly accurate at detecting the infection in thousands of clinically-validated sample that were tested by certified laboratories.

They say that previous studies used mostly crowdsourced samples found online or only a small quantity of clinically-validated samples.

The researchers also used the tool to classify coughs into three severity levels, which could help healthcare professionals allocate resources such as ventilators.

They now plan to conduct interventional studies with the tech and work towards a wider release and certification of the tool.

You can read the study paper in the journal IEEE Transactions on Service Computing .