Microsoft‘s Ignite 2023 event showcased numerous updates, putting a spotlight on AI product unveilings and expansions as the company presented its refreshed vision for IT professionals. We’ve gathered all the announcements from how Azure will impact businesses to how the Windows company plans to integrate AI features into all of its products.
Bing Chat is now Copilot
Microsoft has renamed Bing Chat, its AI-powered chatbot, to Copilot in Bing. The enterprise version, Bing Chat Enterprise, has also been rebranded as Copilot. This change is likely aimed at enhancing competitiveness with ChatGPT and other popular chatbots, separating the technology from the Bing search engine.
Copilot is now accessible on Windows, Copilot.Microsoft.com, and Bing is included in various Microsoft enterprise subscription plans at no extra cost. Come December 1st, it will also become part of Microsoft 365 F3 and will be available for $5 per month for other customers.
Microsoft Teams gets new AI-powered features
Teams now has a new, ‘Decorate your background’ feature, that can save users from having to tidy up their home office. The new video call feature can assess a user’s room, tidy up clutter, and even add plants or decorations to their background. Also, Microsoft announced an AI-driven noise reduction feature for Teams that’s capable of minimising repetitive background noise and other people’s voices.
The voice isolation feature is currently rolling out and will be widely available in early 2024 while the background option will be accessible in Teams Premium early next year. Microsoft is also exploring the potential of generative AI to enhance VR meetings to further amplify the immersive experience.
Microsoft in-house AI chips
The Windows company aims to reduce its reliance on GPUs by introducing two in-house AI chips: the Azure Maia 100 AI Accelerator for training and running AI models and the Azure Cobalt 100 CPU for general-purpose workloads which Microsoft CEO Satya Nadella calls, “The fastest of any cloud provider.” These chips are set to be deployed in Azure data centres early next year, initially supporting Microsoft AI services such as Copilot and Azure OpenAI Service.
AI copyright protections
The company has declared that customers facing copyright infringement lawsuits due to outputs from Azure’s OpenAI Service can now be protected and compensated against adverse judgments. This policy doesn’t automatically apply to all Azure OpenAI Service customers. Subscribers need to proactively implement, “Technical measures” and adhere to risk mitigation documentation to qualify for the AI copyright protection.
Windows AI Studio
In its plans to extend generative AI experiences from the cloud to Windows devices, Microsoft introduced Windows AI Studio. This toolkit merges AI tools and a catalogue of generative AI models that’ll allow developers to refine, personalise and deploy them for local and offline use in their Windows applications.
Windows AI Studio is set to incorporate more advanced local models, such as Meta’s text-generating Llama 2 and Stability AI’s text-to-image model, Stable Diffusion XL. Developers can access these AI models and tweak them to their preferences. It also offers an end-to-end, “Guided workspace setup” with model configuration UI and walkthroughs to fine-tune different small language models (SLMs), such as Meta’s Llama 2, and Microsoft’s Phi.
Microsoft also showcased a new tool that’s capable of generating what is seemingly a deepfake. The Azure AI Speech text-to-speech avatar creates a lifelike avatar of a person and animates it to articulate statements that the individual may not have uttered originally.
Acknowledging the potential for misuse, Microsoft is imposing restrictions on the tool. At launch, most Azure subscribers will have access to prebuilt avatars. Custom avatars with a limited access capability, are currently accessible through registration only and restricted to specific use cases.
You can view keynote highlights of the event from Microsoft’s website.