Tech companies expand AI use while tech leaders call for pause AI development
A call for AI development to pause
As companies continue to integrate AI into their workflow, some are questioning if it is being rolled out too fast. On Wednesday, an open letter asks for a pause on “Giant AI Experiments” for at least six months. The letter asks the readers what we should and shouldn’t let AI do. Their questions include: “Should we let machines flood our information channels with propaganda and untruth?” and “Should we develop nonhuman minds that might eventually outnumber, outsmart, obsolete and replace us?”
A number of noteworthy signatures are included in the letter, including tech billionaire Elon Musk and Apple co-founder Steve Wozniak. The open letter calls for a six-month pause on AI development so that developers can consider the risks associated with AI. It notes that “this pause should be public and verifiable, and include all key actors.” It asks that if this pause cannot be enacted quickly, governments should step in.
Adobe introduces Firefly
As their entrance into the realm of using artificial intelligence, Adobe has launched Adobe Firefly. Firefly is “a family of creative generative AI models” that will eventually be integrated into Adobe applications like Adobe Express, Photoshop, and Illustrator. Firefly has not yet been integrated into these apps, but can be accessed through a public beta on the Firefly website.
There are currently two main functions to Firefly: text to image and text effects. Text to image is similar to other visual AI projects out there, like DALL-E. The user inputs some detailed prompt and the AI will return a generated image that “matches” the prompt. Users will be able to select what kind of styles they want via tags, such as lighting and composition.
Text effects allows the user to input some text and a description, and the AI will generate the user’s text in the style of their description. For instance, the user can input “Fish” as their text and “rainbow fish scales” as their description, and it will produce the text “Fish” covered in rainbow fish scales. Users will be able to choose their base font, background color, and a few other options.
A number of features are planned to be added to Firefly as well. “Recolor Vectors” is listed as “Coming Soon.” A number of techniques are listed as “In exploration”: Using a brush to replace objects with a text prompt-generated fill, generated images based on the user’s own object or style, using text to make a vector, extending images, generating images from 3D objects, making patterns from a detailed text description, generating a brush from a text description, making a sketch into an image, and turning text into a graphic design template.
Notably, Firefly is designed “to be safe for commercial use.” To “learn” how to produce this content based on a text prompt, AI needs to be given training photos. There have been intellectual property concerns over some of the training sets used for AI-art generators, as many artists had not given permission for their works to be included in them. Firefly is trained on Adobe Stock images and openly licensed and public domain content that has expired copyright. Adobe is currently working on a compensation model for Stock contributors as well.
Spotify’s new DJ
In an effort to increase user personalization, Spotify has rolled out an AI DJ into user’s pockets. Though currently in beta, DJ generates a continuous stream of music and commentary for the user. It sorts through the user’s current and old listening habits to provide a unique experience to the user. As the user uses it more, DJ will update itself to better reflect their taste.
DJ is a combination of three technologies. The first is through Spotify’s personalization technology, which generates playlists for the user (like Daily Mixes and Discover Weekly) based on their listening preferences. The second is the use of OpenAI. According to Spotify, “With this generative AI tooling, our editors are able to scale their innate knowledge in ways never before possible.”
The last technology is the use of dynamic AI voice to create DJ’s interactions with the user. In June 2022, Spotify had announced they were acquiring Sonantic, an AI voice platform that “creates compelling, nuanced, and stunningly realistic voices from text.” They are utilizing this technology to create DJ’s voice, which is based on the voice of Spotify’s Head of Cultural Partnerships Xavier “X” Jernigan.
Users can find DJ on their mobile app. If users are not happy with DJ’s selections, they can “hit the DJ button at the bottom right of the screen to be taken to a different genre, artist, or mood.”