Today is very much about AI models because everyone made moves.
Google is rebranding its AI model Bard as Gemini and launching Gemini Ultra, its most capable large language model. Gemini Ultra will be available through a new $20 Google One tier, including 2TB of storage and access to Gemini in Google Workspace apps. The company is also launching a Gemini app for Android and bringing it to the Google app on iOS. Gemini Advanced, the paid plan, will be available in more than 150 countries and territories, initially in English, with Japanese and Korean coming later. Gemini Ultra 1.0 is a multi-model that excels in complex tasks and understanding context. The $20 price point puts Google in line with competitors like Microsoft. This includes upgrades to Google Assistant to leverage Gemini.
Microsoft has announced a redesigned look for its Copilot AI search and chatbot experience and new built-in AI image creation and editing functionality. They have also introduced a new AI model called Deucalion. The redesign includes a cleaner landing page with more white space and a visual carousel of AI-generated images.. The company aims to position Copilot as a creative tool for content creation in film/TV, video gaming, and entertainment industries. The AI image generation features are currently available in select countries. Microsoft will also have a CoPilot AI ad during the Super Bowl.
OpenAI’s image generator DALL-E 3 will add watermarks to image metadata using Coalition for Content Provenance and Authenticity (C2PA) standards. The watermarks will be visible and include a CR symbol in the top left corner of each image. The watermarks aim to identify the provenance of AI-generated content and will not significantly impact latency or image quality. However, they can be easily removed, and watermarking is not foolproof in preventing misinformation.
Meta plans to label Facebook, Instagram, and Threads posts containing AI-generated images. The technical solution for automatic labeling is not yet ready, but Meta will use metadata to identify AI-generated images from various sources. Users can disclose and label AI-generated or digitally altered content, and penalties may be applied for failure to do so. The move comes after Meta’s Oversight Board pressure and criticism of its manipulated media policy.
Apple has developed a new AI model called MGIE, allowing users to edit images by describing them in plain language. The model, developed in collaboration with the University of California, Santa Barbara, can perform tasks like cropping, resizing, flipping, and adding filters to images based on text prompts. Apple has made the MGIE model available for download on GitHub and released a web demo. MGIE can handle various editing aspects, including Photoshop-style modification, global photo optimization, and local editing.
Bard didn’t last long, replaced by Gemini. Gemini is across the entire Google suite, similar to CoPilot. Depending on your ecosystem, you’ll be learning one or both.
The Super Bowl ad signifies Microsoft’s commitment to making AI a mainstream tool, further cementing its position in the market. Don’t be dismissive. This is laying the groundwork for customers to ask, and remember, those questions are sales questions.
The other big trend here is watermarks, right on the heels of growing concerns about the authenticity and misuse of AI-generated content.
Finally, thinking broadly, MSPs and IT services companies must be versed in the various models available. This is the same relationship that sommeliers have with wine, matching customers’ needs and tastes to the right tools and models. Sommeliers are well-trained… and well-paid.