Cassandra Megraw, 6 Aug 2017

The Week In Social: AI everywhere, Instagram Stories, and the frowning poop emoji

Instagram Stories turn 1 and Snapchat is the one crying at their party

Where most babies are barely walking after a year, Instagram Stories has just celebrated its first birthday and is already running rings around its competitor-cum-inspiration (to put it nicely), Snapchat. Half of the businesses on Instagram produced a story in the last month, and the feature has boosted usage to 32 minutes per day for users under 25, and 24 minutes per day for those above 25. Not only has Stories equipped itself with (see: copied) most of Snapchat’s features, it’s caused the social app’s monthly active user growth rate to plummet from 17.2% per quarter to just 5%, and for its share price to sink to an all-time low of around $12.67. (Ouch, Instagram Stories… them baby teeth be sharp!) But given its recent investments in creative tools, ad targeting, and performance measurement, execs and agencies aren’t ruling out Snapchat quite yet.

screen-shot-2017-08-02-at-1-34-59-pm

Read more at TechCrunch

Slow loading websites? Ain’t nobody (Facebook) got time for that!

Facebook is following in Google’s fast-paced footsteps, updating its algorithm to favor stories that load quickly on mobile over their slow-loading counterparts. With Facebook finding that as many as 40% of website visitors abandon a site after three seconds of delay (busy bunch, aren’t we?), it’s writing our impatience into its algorithm, taking into account “the estimated load time of a webpage that someone clicks to from any link in News Feed on the mobile app.” Facebook users’ network connection will be considered, as well as the general speed of the webpage in question. If all goes well, that webpage will appear higher in the feed. With this change upon us, Facebook is now warning particularly slow pages about a decrease in referral traffic and offering tips on how to speed things up.

Read more at VentureBeat

Goodbye filters, hellooo Google & MIT’s new machine learning algorithm

In a world where it seems our phone cameras can’t get any better (well… most), Google has teamed up with scientists from MIT to create an algorithm that can retouch your photos in real time. The peeps involved used machine learning to create their software, training neural networks on a dataset of 5,000 images that had each been retouched by five different photographers. It might mean increasing the brightness here or reducing the saturation there, but whatever it is, they’ve worked the algorithm to be small and efficient enough to run on your device, sans lag. Smartphones already process imaging data in real time, but these new moves are more subtle and reactive, responding to each individual image rather than applying general rules to all.

Read more at The Verge

Facebook announces it’s using its powers of AI for fake news-flagging good

Facebook recently revealed the battle against fake news continues with machine learning on the frontlines. These AI soldiers will work together to filter out even more suspect stories than current procedures. This means the Related Articles artillery that was brought on in April is more likely to show the fact-checker’s account of the topic, rather than just presenting articles with different takes on the same topic. Here’s what a spokesperson had to say: “In addition to seeing which stories are disputed by third-party fact checkers, people want more context to make informed decisions about what they read and share.”

screen-shot-2017-08-03-at-2-45-59-pm

Read more at TNW

Beware the bot-using Instagram “influencer”

With Instagram micro-influencers becoming a valuable commodity over users rolling millions deep in followers, influencer marketing pros are deeming bot-use a “gray hat” form of fraud when it comes to ‘gram fame. Social influencer talent agency Viral Nation receives between 50 and 100 influencer applications each day, and says that 20 – 30% of them have used bots to garner followers and interactions. “It’s especially prevalent in the blogger and micro-influencer category, where bots will go and like and comment on each other’s feeds,” said Joe Gagliese, co-founder of Viral Nation. “We can sometimes easily detect by looking at their comments, like-to-comment ratio, and also the irregular follower growth patterns.”

screen-shot-2017-08-03-at-2-22-51-pm

Read more at DigiDay

Facebook acquires AI assistant startup Ozlo

In an effort to up its AI game and help build “compelling experiences within Messenger that are powered by artificial intelligence and machine learning,” Facebook has acquired AI assistant startup Ozlo. In the takeover, which was announced last week (before the eerie AI news that made headlines the day after [more on that below]) Facebook will receive Ozlo’s tech and most of its team of 30. The startup is also expected to shut down its apps and API services. Just for a little background: Ozlo started out in October 2016 and was initially focused on helping users sift through restaurant listings in a conversational format. Its offerings later expanded to include restaurants that serve foods for people with strict dietary restrictions, as well as weather forecasts, and more local business services like movie listings.

Read more at Venture Beat

Facebook shuts down AI system after chatbots develop their own language

Whoa, guys. The dystopian future where computers take over the world is upon us – and even Stephen Hawking is worried. Days after Tesla CEO Elon Musk spoke and tweeted about the huge risks of AI, and not a day after Facebook acquired an AI startup, the social media company had to shut down one of its AI systems after its chatbots started defying code and speaking their own language (!!!) Apparently “things got out of hand” and the chatbot “stopped using English and started using a language that it created.” Some of the brightest brains in tech are concerned about where AI is heading, and this is what Facebook founder Mark Zuckerberg had to say about that: “I think people who are naysayers and try to drum up these doomsday scenarios – I just, I don’t understand it. It’s really negative and in some ways I actually think it is pretty irresponsible.” Guess we’ll just wait and see. Gulp.

Read more at Manorama

Facebook reportedly working on voice speaker and video chat device with laptop-sized screen

In perhaps slightly less terrifying news, Facebook is currently developing an in-home “video chat device” to rival Amazon Echo, Google Home, and Apple’s soon-to-be-released HomePod. The snazzy gadget, which could be announced next spring, includes a touchscreen, wide-angle camera, microphones, and speakers. Bloomberg reports that Facebook wants to use AI (oh my!) for the advanced camera features, including one that would “scan for people in its range and lock onto them.” Several home cameras already offer this type of functionality for recognizing and identifying people in your home, but Facebook will likely face the more difficult challenge of overcoming privacy concerns and getting consumers comfortable with an always-on camera in their house. (Especially considering many already believe the conspiracy theory that Facebook secretly listens to them via their smartphone microphone.)

Read more at The Verge

2018 may very well be the year of the frowning poop emoji

Well guys, it looks like Unicode has been getting our letters! The organization which governs the introduction of emojis (did somebody say “dream job”?) has been hard at work drafting 67 candidates for its 2018 emoji release, and we’re pleased to see a frowning poop emoji is included. If it makes the cut, it’s still then up to your smartphone peeps to include it in their offering, but honestly, who could say no to a frowning poo? He’s clearly already had a bad enough day. Fellow candidates include a freezing emoji, celebration emoji, and hearts flying around the head emoji. Oh, and there are whispers on the street that we’ll soon be able to choose our emojis’ hair color. What joyous news! *Smiley poop emoji*screen-shot-2017-08-04-at-8-01-28-amRead more at The Sun

Facebook Stories are coming to desktop and the crowd goes… back to scrolling Instagram Stories

Facebook Stories is coming to desktop, hopefully eliciting more interaction than what mobile has managed to. Where this Snapchat-inspired offering has garnered 250 million users on Instagram each day, with Snapchat drumming up 166 million, the feature has been mostly ignored on Facebook since its launch at the beginning of the year. In fact, there were so few people using Stories on Facebook’s mobile app, that the company in April began to display grayed-out icons of users’ most frequently contacted friends instead of blank spaces where their Stories would otherwise be. Awkwaaaaard. So far it’s not clear whether this move will impact the feature’s usage significantly, but for now Instagram Stories remains the bae-est of baes in the ephemeral content space.

Read more at The Verge

Latest AmEx ad proves that it pays to listen

AmEx Israel recently treated Guns ‘N Roses fans to a bit of a sweet surprise with its latest Facebook ad – assuming they had their sound turned on. Knowing that 90% of Facebook users watch videos with the sound off, the brand hid a fun surprise in the video for anyone listening – a voiceover promising a free ticket to the first 50 people who knew the name of the band’s lead singer. Definitely an interesting way to see who’s really listening.

screen-shot-2017-08-03-at-2-46-35-pm

Read more at AdWeek

Facebook’s translations are now powered completely by AI

With 4.5 billion automatic translations required every day, Facebook has kicked its tech up a notch and begun processing each one through neural networks. The company blog post explained that “creating seamless, highly accurate translation experiences for the 2 billion people who use Facebook is difficult. We need to account for context, slang, typos, abbreviations, and intent simultaneously.” So while the old system has a shorter attention span and translated sentences word by word or by short phrases, the new neural networks consider whole sentences at a time. They do this using a particular sort of machine learning component known as an LSTM, or long short-term memory, network.

screen-shot-2017-08-04-at-8-32-18-am

Read more at The Verge