Spotify Launching OpenAI-Powered “DJ” That Talks About Songs Between Tracks
]
Streaming giant Spotify, arguably a horseman of the distilling-music-into-vibes-only-pocalypse, wants to bring back the radio DJ. They, uh, just don’t want those DJs to be human. More algorithms for all!
The music service just announced that it will soon be leveraging artificial intelligence to provide each user with an individual “AI DJ in your pocket.” You know, because that’s something we’ve all been asking for.
According to TechCrunch, Spotify says the system will share “culturally relevant, accurate pieces of commentary at scale.” In other words: bringing the radio DJ back en masse, but without having to employ humans to do the job.
“If you’re not feeling the vibe, just tap the DJ button and it will switch it up,” reads a Spotify press release explaining the tech. “The more you listen and tell the DJ what you like (and don’t like!), the better its recommendations get.”
“Think of it as the very best of Spotify’s personalization,” it adds, “but as an AI DJ in your pocket.”
Of course, one might argue that Spotify’s inhuman pocket DJ goes directly against everything that a radio DJ has always been: a curator, sure, but a curator who uses their individual human taste and sensibility to bring music to listeners. Spotify’s AI DJ is seemingly the exact opposite, regurgitating — like so many music algorithms do — a listener’s patterns back at them, without any genuine flair (re: human) of its own.
Just another AI cog in the vibe vacuum, baby. But we digress.
Spotify says in the release that the AI DJ is powered by two AI components: a Spotify-owned AI voice generation tool called Sonastic and, most intriguingly, unspecified OpenAI tech. Though it’s unclear what exactly the OpenAI device at hand is, it’s presumably some version of OpenAI’s Large Language Model (LLM), GPT — the same overly confident tech that’s not just constantly wrong about things, but has led to the chaos that is Microsoft’s Bing Chat/Sydney going off the absolute rails.
And look, we’re not hoping that the streamer’s artificial DJ starts spouting believable bullshit about musicians, transforming Mr. Dee-JAI into a PR nightmare. We’re also not hoping that it falls in love with users, suddenly begins to ponder its sentience, or starts to name its enemies, all of which OpenAI’s tech has already done. But it really would be something to behold.
Spotify does claim that it has editors in charge of making sure that the AI provides accurate information, but we’ve heard that tune before. (It’s also unclear how editors might keep up with the program, considering that the AI’s algorithm is allegedly crafting cultural commentary in real-time.)
“The expertise of our editors is something that’s really important to our philosophy at Spotify,” reads the press release. “We have experts in genres who know music and culture inside and out. And no one knows the music scene better than they do.”
“With this generative AI tooling,” it continues, “our editors are able to scale their innate knowledge in ways never before possible.” Glad to see that the “tooling” rebrand is catching on.
And while the Spotify DJ probably won’t have the exact same problems as the interactive Bing AI, there’s another dark reality of GPT tech that’s lurking here: machine bias.
While OpenAI does have some guardrails in place for its products, those guardrails are very far from perfect, and racial, sexist, transphobic, or otherwise terrible biases are certainly embedded into any robust training data — in fact, GPT-powered products have already seen some pretty terrible stuff slip through the cracks. Utilizing a bias-ridden product to make humanless music commentary may ultimately prove to be an unnecessary risk on Spotify’s behalf.
But only time — and maybe, if this thing does follow in the footsteps of OpenAI integrations before it, not that much of it — will tell. In any case, if you’re an English-speaking Spotify user, a DJ should be arriving in your pocket soon.
READ MORE: Spotify launches ‘DJ,’ a new feature offering personalized music with AI-powered commentary [TechCrunch]
Spotify just launched a personalized ‘DJ’ powered by Generative and Voice AI
]
From the use of generative AI for lyric writing, to the use of voice-mimicking AI in songs, the use of artificial intelligence in music has become one of the industry’s key talking points of 2023.
Now music streaming giant Spotify is throwing its hat in the ring, with a new feature powered by its own personalization tech, as well as by voice and generative AI.
The company is launching a ‘DJ’ feature, which it says is like an “AI DJ in your pocket” and adds that it serves as “a personalized AI guide that knows you and your music taste so well that it can choose what to play for you”.
This feature is first rolling out in beta, and Spotify says it will deliver a curated playlist of music alongside commentary around the tracks and artists it thinks you will like.
This commentary, according to Spotify, will be narrated “in a stunningly realistic voice” using a so-called “dynamic AI voice platform” from its recent acquisition of London-based AI voice startup Sonantic.
Launched in December 2018 by Zeena Qureshi and John Flynn, Sonantic’s founders have backgrounds in speech and language therapy to Hollywood sound production.
Last year, Sonantic built a custom AI voice model for actor Val Kilmer, which was used in the most recent Top Gun film, Top Gun: Maverick.
To create the voice model for Spotify’s new AI DJ, the company says that it partnered with its own Head of Cultural Partnerships, Xavier “X” Jernigan, who previously served as one of the hosts on the platform’s first (and personalized) morning show, The Get Up.
According to Spotify, his “personality and voice resonated with our listeners and resulted in a loyal following for the podcast” and his voice has now been used as the first model for the AI DJ.
You can hear what that sounds like in the video below:
The new feature also uses generative AI to power the commentary, through the use of technology from US-based artificial intelligence research firm, OpenAI.
Spotify claims that it put this tech “in the hands of [its] music editors to provide you with insightful facts about the music, artists, or genres you’re listening to”.
Added Spotify in a blog post: “The expertise of our editors is something that’s really important to our philosophy at Spotify.
“We have experts in genres who know music and culture inside and out. And no one knows the music scene better than they do. With this generative AI tooling, our editors are able to scale their innate knowledge in ways never before possible.”
In addition to Sonantic’s voice tech and the use of generative AI, Spotify’s new DJ feature is powered by its ownpersonalization technology, which gives users a lineup of music recommendations based on their listening history.Music Business Worldwide
Spotify releasing artificial intelligence DJ in two countries
Spotify is adding an artificial intelligence DJ for certain listeners.
The audio streaming service said Wednesday in a press release that U.S. and Canadian users with premium subscriptions would first start getting access to the DJ that day. In the beginning, it will be “in beta” and in English, according to Spotify.
Headphones are shown in front of the Spotify logo.
The DJ will provide a “curated lineup of music alongside commentary around the tracks and artists we think you’ll like in a stunningly realistic voice,” the company said. The feature will apparently pull its selections from new tunes in addition to ones the user has previously listened to.
CLICK HERE TO READ MORE ON FOX BUSINESS
It is designed so that its music recommendations improve as it gets more and more feedback from the user, according to the press release.
Spotify said the DJ feature stems from several things, including its personalization technology and a generative AI “through the use of OpenAI technology.”
READ ON THE FOX BUSINESS APP
GET FOX BUSINESS ON THE GO BY CLICKING HERE
The audio streaming service also pointed to its 2022 purchase of Sonantic, an AI voice platform that it said creates “realistic voices from text.”
The Spotify logo is seen displayed on a smartphone next to a pair of earphones.
Spotify Head of Cultural Partnerships Xavier Jernigan served as the DJ’s first model for its voice, something the company indicated it will “continue to iterate and innovate.”
The Spotify logo is seen displayed on a smartphone screen near a pair of earbuds.
AI has been making headlines recently, with Google and Microsoft both recently making announcements about new AI-powered features for their respective platforms.
GOOGLE VS. MICROSOFT IN AI RACE
On Friday, Meta Platforms also unveiled “LLaMA,” a large language model “designed to help researchers advance their work in this subfield of AI” that the Facebook corporate parent is releasing.
Joe Toppe contributed to this report.
This Week in Apps: Meta’s paid verifications, Instagram’s founders’ new app and Spotify’s AI DJ
]
Welcome back to This Week in Apps, the weekly TechCrunch series that recaps the latest in mobile OS news, mobile applications and the overall app economy.
The app economy in 2023 hit a few snags, as consumer spending last year dropped for the first time by 2% to $167 billion, according to data.ai’s “State of Mobile” report. However, downloads are continuing to grow, up 11% year-over-year in 2022 to reach 255 billion. Consumers are also spending more time in mobile apps than ever before. On Android devices alone, hours spent in 2022 grew 9%, reaching 4.1 trillion.
This Week in Apps offers a way to keep up with this fast-moving industry in one place with the latest from the world of apps, including news, updates, startup fundings, mergers and acquisitions, and much more.
Do you want This Week in Apps in your inbox every Saturday? Sign up here: techcrunch.com/newsletters
Top Stories
Meta starts selling blue badges…but also security and customer service
In a stealth announcement over the weekend, Meta announced a radical change to Facebook and Instagram with news that it would offer to sell its blue verified badge to customers, taking a play from Elon Musk’s Twitter playbook. The paid subscription includes other features as well, including improved impersonation protection and direct access to customer support, plus more visibility through upranked posts. It’s initially rolling out to Australia and New Zealand.
Twitter’s initial attempt at paid verification proved problematic, as users bought the badge then changed their name and profile picture to troll other high-profile accounts (including Musk) and businesses. Twitter had to pause the service and readjust.
Seemingly learning from Twitter’s mistakes, Meta’s paid badge has a few more rules in place.
For starters, users must verify their identity with a government-issued ID card, and then won’t be able to change their profile name, username, date of birth or photo after paying for verification. If they later want to make a change, they’ll need to unsubscribe and then get reverified. This dramatically cuts down on bad actors, though could be a bit of a pain for creators who like to refresh their photos from time to time. However, it may not always be this way — Meta said it’s working on a feature that will eventually allow users to change these settings through a new verification process that won’t require them to cancel and resubscribe… it’s just not ready yet.
Also of note: Meta Verified won’t verify users across Facebook and Instagram — users will have to buy separate plans for the two apps, and Facebook’s subscription, for now, is only sold on the web. That means customers will be shelling out $27 per month at the current prices for access to this badge and other perks across Meta’s apps. (The subscription is $11.99 per month on the web and $14.99 per month on iOS or Android.)
The trend toward paid verification is a potentially fraught move for social networks like Meta and Twitter, as they’re now responsible for services that users believe should be free — things like safety, security and customer service. Being able to identify an account as authentic is seen as a feature the networks should provide to ensure that their users can trust who they’re interacting with. And being able to get help with problems like impersonation or other customer support issues is also considered something that should be a part of the social network’s core service. By stratifying these features into pay-to-play tiers, the networks are setting up a system where people with money have a better class of service than those with less to spend. But security and trust shouldn’t be sold as if they’re upgraded seats on an airplane, they should be baked into the core offering.
Instagram co-founders launch their new app… and it’s for news
Artifact, the personalized news reader built by Instagram’s co-founders, is now open to the public, no sign-up required. Last month, Instagram’s creators Kevin Systrom and Mike Krieger unveiled their latest venture as an invite-only experience, promising their news app would later evolve to include social elements, like being able to discuss the news with friends. With this week’s launch, Artifact is dropping its waitlist and phone number requirements, introducing the app’s first social feature and adding feedback controls to better personalize the news reading experience, among other changes.
In preparation for expanded social features, the company will now allow users to upload their contacts to see when articles are becoming popular with people in their network. But unlike a similar feature on Twitter, it won’t show you who is reading them.
Artifact will also now give users more visibility into their news reading habits with a newly added stats feature that shows you the categories you’ve read as well as the recent articles you read within those categories, plus the publishers you’ve been reading the most. But it will also group your reading more narrowly by specific topics. In other words, instead of just “tech” or “AI,” you might find you’ve read a lot about the topic “ChatGPT,” specifically.
The launch of a brand-new app from Instagram’s founders, and particularly one focused on news, was a surprise — especially given the difficulties of launching a news reader here in the U.S., where it would have to compete with offerings from the tech giants, like Google News, Apple News and, of course — from the founders’ earlier employer — Meta’s own News Feed. But Systrom believes that the underlying machine learning technology being used will help Artifact differentiate itself from others — it’s leveraging the transformer advances that are also powering new AI tools like ChatGPT.
While users are likely curious about the app because of its founders’ pedigree, it remains to be seen if there’s room for another news reader to carve out a niche under the tech giants’ shadow. Before the waitlist was lifted, the app had around 47,000 installs, according to data.ai. As of late this past week, it had climbed to No. 4 in the U.S. App Store’s News category, but hadn’t broken into the Top Free Charts. [Update: the most recent data now shows 66K installs and the app is ranking No. 5 in the News category in the U.S., No. 5 in the UK, and higher in some EMEA markets including Malta, Spain, Israel, and Hungary.)
Spotify launches an AI DJ
Ah, what a time to be alive! Music streaming service Spotify this week launched an AI DJ to personalize the music listening experience for its users. Similar to a radio DJ, Spotify’s DJ feature will deliver a curated selection of music alongside, in its case, AI-powered spoken commentary about the tracks and artists you like, using what Spotify says is a “stunningly realistic voice.” (The voice is based on Spotify’s Head of Cultural Partnerships Xavier “X” Jernigan, who had hosted Spotify’s morning show podcast, “The Get Up.”)
To access the DJ, you’ll head to the Music Feed on the Home page of Spotify’s iOS or Android app, then tap Play on the DJ card to begin. The DJ will then begin to play a lineup of music and short commentary. As listeners engage with the DJ feature, they’ll be presented with a personalized stream of songs that will include both newer tracks and old favorites, and a variety of genres. But it’s not a long-running playlist. After you move through one style of music or selection (like your summer throwbacks), you’re then presented with another (like your favorite hip-hop tracks). This experience feels more like Spotify tied its personalized playlists together, then interspersed them with commentary.
The interesting thing here is that Spotify said it’s leveraging Generative AI through the use of OpenAI technology to create the commentary, which is meant to scale its in-house music experts’ insights about music, artists and genres. Meanwhile, its AI voice comes from its 2022 Sonantic acquisition. Spotify has led the market for years with its personalization tech for crafting playlists, but now its rivals have their own versions of this type of experience. By adding an AI DJ, Spotify hopes to attract and retain users who want a more lean-back experience while introducing a new feature that can’t be quickly copied by the competition.
Platforms
Apple
Apple is offering new sessions with App Store experts February 28 through April 13 that will focus on measuring with App Analytics and growing a subscription business using App Store features.
that will focus on measuring with App Analytics and growing a subscription business using App Store features. The Apple Store app added the ability to share a saved list with friends and family and more ways to access the saved lists from within the app. It’s also now offering more info about your local stores and their surroundings.
Google has begun to support third-party payments in the Play Store in India to comply with local regulations. When users opt for an alternative payment method, Google will reduce the commission by 4%.
App Updates
Social
Media & Entertainment
Spotify is planning to launch a TikTok-style feed for music discovery in its app, according to Bloomberg, which said the news would be announced at the company’s upcoming Stream On event in March. Spotify previewed the feature at its Investor Day last June.
according to Bloomberg, which said the news would be announced at the company’s upcoming Stream On event in March. Spotify previewed the feature at its Investor Day last June. Podcasts are coming to YouTube Music. YouTube announced that ad-supported podcasts would be made available on YouTube Music, with support for background listening included for free. The feature will include both audio and video podcasts, initially for users in the U.S.
YouTube announced that ad-supported podcasts would be made available on YouTube Music, with support for background listening included for free. The feature will include both audio and video podcasts, initially for users in the U.S. YouTube Music’s redesign brought a new feature that lets users create their own automatically generated radio stations by picking up to 30 artists and then applying mood filters. The stations can also be further refined with other specific filters like “new discoveries” or “chill songs,” for example.
Nexstar Media Group launched a free NewsNation app for streaming devices, including Apple TV, Roku, Fire TV and others.
for streaming devices, including Apple TV, Roku, Fire TV and others. Spotify re-org’d again . After last month’s departure of Dawn Ostroff, who oversaw podcast content and advertising, Spotify’s head of audio talk shows and partnerships Max Cutler is also leaving the company on May 1 as part of a larger re-org. Cutler notably oversaw deals with top creators like Joe Rogan and Alex Cooper (“Call Her Daddy”), after joining the company when it bought his network Parcast. Julie McNamara, who oversaw originals, will now manage exclusives too as Cutler departs.
After last month’s departure of Dawn Ostroff, who oversaw podcast content and advertising, the company on May 1 as part of a larger re-org. Cutler notably oversaw deals with top creators like Joe Rogan and Alex Cooper (“Call Her Daddy”), after joining the company when it bought his network Parcast. Julie McNamara, who oversaw originals, will now manage exclusives too as Cutler departs. Clubhouse is adding a “Mutals” feature that lets you see who you know in common with other participants in one of its live audio rooms. The company said it could serve as a good icebreaker for chatting up new folks.
in one of its live audio rooms. The company said it could serve as a good icebreaker for chatting up new folks. Spotify began testing playlists that could only be unlocked by NFT holders. The feature was being tested by the metaverse band Kingship and communities like Overlord, Fluf and Kevin Rose’s Moonbirds.
The feature was being tested by the metaverse band Kingship and communities like Overlord, Fluf and Kevin Rose’s Moonbirds. YouTube launched a new multi-language audio feature that allows creators to add dubbing to their videos after creating the dubbed tracks with a third-party partner. The feature was tested by select creators, like MrBeast, and is now expanding to thousands more creators for use in long-form videos.
with a third-party partner. The feature was tested by select creators, like MrBeast, and is now expanding to thousands more creators for use in long-form videos. Celeb greetings app Cameo named Meta vet Matty de Castro as its GM of Enterprise Sales, Cameo for Business.
Cameo for Business. Xiaomi is shutting down its short-form video app Zili next month, citing an “operational adjustment.”
Gaming
ASO matters! Rovio said it’s delisting Angry Birds on Google Play and renaming it on the App Store because the older game is taking away attention and downloads from its newer versions, like Angry Birds 2, Angry Birds Friends and Angry Birds Journey. The older App Store game will be renamed to Red’s First Flight in order to redirect search traffic to the newer titles. The game will remain playable on devices it’s been downloaded to even after the rebranding and removal.
Please read below for an important announcement regarding the availability of Rovio Classics: Angry Birds. pic.twitter.com/a4n4bU5gQJ — Rovio (@Rovio) February 21, 2023
Unreal sneak peek ahead. Epic Games said it’s returning to the Game Developers Conference in San Francisco on March 22 to give a glimpse of the future of game development with its Unreal Engine. The company promises to “look at some new projects” and “dive into the latest Epic tech.” The keynote will be livestreamed on Twitch and YouTube.
Messaging
WhatsApp was spotted developing a “private newsletter tool“ which would expand on its existing broadcast functionality to allow users to broadcast via newsletter, as well. The reveal follows Meta’s launch of a broadcast channels feature that lets users send a one-to-many message to Instagram users. The same feature is also being tested in Messenger.
which would expand on its existing broadcast functionality to allow users to broadcast via newsletter, as well. The reveal follows Meta’s launch of a broadcast channels feature that lets users send a one-to-many message to Instagram users. The same feature is also being tested in Messenger. Messenger was spotted internally testing another BeReal clone, “Roll Call.” The feature asks users to add a photo or video to a prompt with a timed countdown to share what they’re up to at the moment with a group of friends in a chat.
The feature asks users to add a photo or video to a prompt with a timed countdown to share what they’re up to at the moment with a group of friends in a chat. Google rebranded its chat features in the Google Messages app as “RCS Chat” and now refers to a “Chat Message” as an “RCS Message,” among other changes. The subtle shift in branding is meant to highlight Google’s adoption of the next-gen communication protocol meant to replace SMS. Apple has steadily refused to implement it on its own devices, as RCS offers many iMessage-like features and would reduce its grip on the blue bubble-demanding market.
AI
Two weeks after launching the new AI-enabled Bing on desktop, the new Bing became available in the Bing mobile app and through Microsoft’s Edge browser for Android and iOS. Skype, Microsoft’s messaging app, also now allows you to bring Bing into a text conversation to add additional information with the @Bing command. Bing’s AI has seen some drama since its launch as users trolled and tested the AI’s limits, which pushed Microsoft to adjust some parameters around things like the length of conversations and other things. Unfortunately, that means the AI is now restricting users to six turns per conversation and 60 total queries today.
Google Photos made its AI-powered Magic Eraser photo editing feature available to Android and iOS users with a Google One paid subscription. The feature was previously Pixel-only. The company also rolled out a small handful of other editing tools, as well, like a new HDR video effect and exclusive collage styles.
Etc.
Amazon’s Alexa app was updated with a new feature that allows users to manage and move their music between multiple Echo devices or groups of speakers within the app instead of using voice commands.
within the app instead of using voice commands. Google said it will begin the big Google Tasks merger in March. This will allow users to manage all the tasks created across Google apps like Gmail, Docs and Chat in the Tasks app itself. On May 22nd, it will also move reminders from Calendar and Assistant into Tasks too.
This will allow users to manage all the tasks created across Google apps like Gmail, Docs and Chat in the Tasks app itself. On May 22nd, it will also move reminders from Calendar and Assistant into Tasks too. Samsung’s Bixby mobile assistant added a new feature that lets users clone their own voice with AI to answer phone calls, but it’s only available in Korean for now.
to answer phone calls, but it’s only available in Korean for now. Stripe’s Tap to Pay arrived on Android in six countries, including the U.S., Canada, the U.K., New Zealand, Australia and Singapore. The feature supports payment methods using Google Pay, Mastercard, Visa and American Express debit and credit cards. Last year, Stripe was Apple’s first payment partner for “Tap to Pay.”
Travel and Transportation
Security
Researchers found bugs that would have allowed attackers to bypass Apple’s sandbox on iOS and Mac, allowing them to access messages, photos and call history. Apple fixed the bugs before the disclosure was made public.
allowing them to access messages, photos and call history. Apple fixed the bugs before the disclosure was made public. Twitter dumbly made SMS 2FA a paid subscription feature only …which we suppose is in keeping with the new social networking model where security and customer support are only available to paying customers now.
…which we suppose is in keeping with the new social networking model where security and customer support are only available to paying customers now. Apple removed scammy authenticator apps from the App Store which couldn’t even scan QR codes until users subscribed to their service. Some also used dark patterns that should have never gotten through App Review — like tapping on the X to close the paywall would prompt a subscription confirmation.
Let me show you something interesting. This app was released on 02/19/2023 and ranks 5 for “authenticator app” in the US App Store. As you can see from the video, once you tap on “X” to close the paywall, you will get triggered in subscription confirmation. pic.twitter.com/JI7XBcAy1s — Kevin Archer (@IM_Kevin_Archer) February 21, 2023
Government, Policy and Lawsuits
ChatGPT Killed the Radio Star? These AI-Powered Chatbots Can Replace DJs
]
Taking a page from ChatGPT, a company in Ohio has created an AI-powered DJ that’s designed to read the news, talk about trending social media posts, and crack jokes during song breaks.
Futuri last week launched(Opens in a new window) “RadioGPT,” a system that can help radio stations localize their content to a market and phase out pre-recorded segments for live radio talk from an artificial DJ.
The technology taps OpenAI’s GPT-3 language model, which also powers ChatGPT, an AI-powered program that can generate human-like conversations on any topic. Futuri decided to use GPT-3 to create content from the company’s TopicPulse program, which can quickly identify news and events trending on social media. The generated content is then read outloud using an artificial but human-sounding voice.
Futuri created a demo(Opens in a new window) of RadioGPT, which shows the system using various artificial voices during song breaks. “Anything a radio human can do, I can do better,” the program proudly proclaims at one point.
(Credit: Futuri)
In the demo, RadioGPT talks about the news and social media topics trending in a city called Springfield. “By and large, the stories in the demo sample are actual stories, but from a variety of different markets, and the names may be changed (We used ‘Springfield’ because there are several in the US and Canada.),” Futuri told PCMag. “The sample is intended to be representative of the type of content one can hear on RadioGPT once launched by our partners.”
The website(Opens in a new window) for the technology says it can also report the weather, discuss artists being played on the radio station, and hold giveaway prizes to listeners. So it’s not hard to imagine RadioGPT representing a threat to human DJs when OpenAI’s own ChatGPT has sparked concerns about AI programs replacing some white collar work.
”Programming is available for individual dayparts, or Futuri’s RadioGPT can power the entire station,” the company says in the announcement. Futuri also notes the technology could be used to train the technology with “existing personalities’ voices” at a radio station.
The system is already live for the company’s beta partners, which include the Portland, Oregon, company Alpha Media in the US and Roger Sports & Media in Canada. But one challenge facing the technology will be preventing the artificial DJ from making on-air mistakes, such as spouting misinformation.
The news comes shortly after Spotify added a DJ card to Premium users' Music Feed in the Android and iOS apps. Tapping it triggers an AI DJ that serves up a list of music recommendations based on what Spotify already knows you like. That DJ is equipped with a “stunningly realistic voice” to offer “commentary around the tracks and artists.”
10 Bits: The Data News Hotlist
]
This week’s list of top data news highlights covers February 18, 2023 to February 24, 2023 and includes articles on launching a data dashboard to track prescription drug prices and using an AI system to recreate voices in phone calls.
- Improving Language Models
Meta has created and released a language model that researchers can use to study methods of preventing bias, toxicity, and hallucinations in language models. The company trained the model on the top 20 most-spoken languages and released versions with 7 billion, 13 billion, 33 billion, and 65 billion parameters.
- Riding in Robotaxis
Motional, a U.S.-based autonomous vehicle company, has launched three new upgrades to its autonomous ride-hailing service in Las Vegas, Nevada. First, the company will now offer rides at night instead of only during daytime hours. Second, vehicles that encounter unknown or difficult traffic patterns can connect to a remote operator to learn how to navigate through the issue. Finally, passengers will now have more control over interior functions, such as unlocking the vehicle and controlling the air conditioning.
- Playing Music
Spotify has launched an AI-powered DJ that uses a text-to-speech system to offer commentary on songs and artists and can pick songs according to users’ listening histories and musical tastes. Spotify’s music experts will use OpenAI’s generative AI systems to create the commentary. The DJ improves upon Spotify’s past personalization efforts as users can tap the DJ button if they do not like the selected music and receive a different song. Over time, the DJ will learn users’ preferences.
- Fighting Fires
Researchers at Rey Juan Carlos University and the Autonomous University of Madrid in Spain have created an autonomous robot that can navigate indoor environments during fire emergencies and help firefighters plan their rescue or intervention. The robot uses a camera and sensors to collect data on temperature, humidity, air quality, and the position of other objects indoors. Then, the robot uses an AI system to navigate its surroundings and map the shortest route to a target position, such as a victim’s location or a building exit.
- Playing Video Games
Sony has launched an update to the racing video game Gran Turismo 7 that allows high achieving players to race against GT Sophy, an AI system that has beaten several professional players. Players can compete against four versions of GT Sophy, each with different specifications.
- Simulating Climate Patterns
Researchers at the Department of Energy’s Argonne National Laboratory have used a supercomputer to simulate climate patterns in the United States for the next 50 years. The team found that the U.S. Midwest will likely experience prolonged droughts and short, heavy bouts of precipitation that could cause extensive flooding.
- Predicting COVID-19 Resistance After Exposure
Researchers at Johns Hopkins University and Johns Hopkins Medicine have created a machine learning model that can predict which patients have a high probability of natural resistance to COVID-19 after exposure to the virus. By better identifying which individuals are naturally able to avoid infection, the researchers may be able to learn which genetic, environmental, societal and behavioral factors influence defense against the virus.
- Tracking Prescription Drug Prices
The Minnesota Department of Health has launched a dashboard to improve prescription drug transparency. The dashboard contains data on prescription drug prices, price increases, and launch prices for nearly 700 drugs.
- Answering Phone Calls
Samsung has updated its Bixby Text Call tool, which allows users to respond to incoming phone calls with a text message, for Korean users. Now, users who speak Korean can send a text and the tool will convert the text to audio that sounds like the user, and respond to incoming calls. The tool generates the audio from vocal recordings of users saying specific sentences.
- Learning About Marine Life
Ikea has partnered with Meta and Warpin Reality, a Swedish mixed reality company, to launch a game that uses augmented reality to teach users about marine life. Shoppers in 21 stores across Sweden can scan QR codes with the Instagram app to view marine animals and learn about the effects of pollution on marine ecosystems.
Image credit: Flickr user Tim G. Photography
YouTube’s new head talks 2023 priorities, including AI, podcasting, Shorts and more
YouTube’s new head Neal Mohan penned his first letter to creators emphasizing that the company in the year ahead aims to continue supporting the community by giving them more tools to make money. He also touched on other 2023 priorities, including how YouTube is looking to experiment with generative AI and multiple formats like Shorts and podcasts, among other things.
Last month, YouTube’s long-time CEO Susan Wojcicki stepped down from the role and moved to an advisory position at Alphabet. That led to Chief Product Officer Neal Mohan being promoted to the company’s top seat. He is now in charge of spearheading YouTube, which competes in multiple categories, ranging from short videos to streaming services.
In the letter, the newly promoted executive quoted a study from Oxford Economics indicating that in 2021, more than 2 million creators earned the money equivalent of a full-time job across multiple countries. In the last few months, YouTube has begun experimenting with different methods that let creators make money, including shopping-related features and ad revenue sharing on Shorts. The company mentioned that people subscribing to individual channels have jumped 20% year-on-year to six million.
The new YouTube head also highlighted multi-lingual features, including dubbing clips into another language and auto-caption. He mentioned that the executives will be looking to meet more creators this year and offer more support to them.
“We’re also listening to creators through increased support. Last year, we more than doubled the number of creators and partners who can get live help through chat or email. Over half of these creators are located outside of the U.S. We’ve also significantly increased the number of creators who have a partner manager to give strategic tips for success on YouTube,” he said.
Mohan, who was previously the chief product officer at the company, said that the video streaming platform is experimenting with introducing more features for formats like connected TV experience, Shorts and podcasts.
Story continues
Notably, at a recent event, the company announced that podcasts are coming to YouTube Music with features like background play. Last year, YouTube hinted at its plans when a new podcast page appeared for U.S.-based users. The platform is now building RSS feed integration for podcasters so they don’t have to upload episodes to the service separately.
After announcing exclusive streaming rights for NFL Sunday Ticket last December, YouTube said that the company plans to let users view multiple matches at once through a feature rollout this year.
Google is also pretty keen to take some market share of short videos from TikTok and Instagram. The company has grown to 50 billion daily views for Shorts with the number of channels uploading short videos growing 80% year-on-year, the letter said. This year, it is going to introduce a side-by-side format that lets creators record Shorts with other Shorts or videos — essentially nudging them to make more reaction videos.
Given the recent popularity of OpenAI’s ChatGPT bot, multiple platforms are exploring product use cases of generative AI. Microsoft is embedding the tech in Bing search, the Edge browser and even Windows 11. In response, Google has announced its own solution called Bard. Snapchat has introduced a ChatGPT-styled bot for its paid user while Meta has formed a group that will look to build AI-powered features. So YouTube doesn’t want to stay behind — but the new company head was sparse on details in his letter.
“Creators will be able to expand their storytelling and raise their production value, from virtually swapping outfits to creating a fantastical film setting through AI’s generative capabilities. We’re taking the time to develop these features with thoughtful guardrails. Stay tuned in the coming months as we roll out tools for creators as well as the protections to embrace this technology responsibly,” Mohan said in the letter.
He also highlighted that YouTube is concentrating on making the platform safe — especially for kids with tools like parent-controlled playlists on Google TV.
In the letter, Mohan made a point that YouTube is engaging with various governments on policymaking. The company is involved in a case in the U.S. Supreme Court where plaintiffs have alleged that Google is responsible for promoting content on YouTube that was uploaded before the terrorist attack in Paris in 2015. The case hinges on how the court interprets Section 230, which absolves the platform from responsibility for user-generated content.
At a time when Google’s position as a search giant is being challenged by rivals incorporating AI in their search, the company will look toward YouTube to be a significant revenue driver in the future. The video platform already brought $29.2 billion in ad sales last year. Plus, more than 80 million people are paying for YouTube Premium and YouTube Music.
“This is a pivotal moment for our industry. We face challenging economic headwinds and uncertain geopolitical conditions. AI presents incredible creative opportunities but must be balanced by responsible stewardship. Creators, viewers, and advertisers have more choices about where to spend their time than ever before and platforms like YouTube need to deliver across a range of formats while investing in the policies that protect platforms from real-world harm,” Mohan said.