In 2020, Microsoft’s headline-grabbing USD 1 billion investment in Sam Altman’s startup Openai signalled that the Satya Nadella-led company was entering a technological arms race with Alphabet. OpenAI was Microsoft’s way of acknowledging that the company intended to stay on top of cutting-edge AI technology. A couple of years in, Microsoft’s efforts to push AI development and leverage its partnerships with OpenAI have become more apparent. This year’s Microsoft Ignite event held on October 12 is evidence of their increasing reliance on OpenAI. 

While AI is all-pervasive in announcements from events held by tech companies, Microsoft has capitalised on OpenAI inventions to power some of its core products. Besides integrating OpenAI’s tech into Azure, which opens up new advancements in AI to the public and in turn helps it curate its models, these modalities enhance Microsoft’s tools. Teams can now summarise a meeting by listening to the recording, Powerpoint presentations can now use DALL.E to create designs that aren’t readily available – the possibilities are endless. 

Applications of DALL.E 

Last year, Nadella had invited Altman on the stage with him to announce the Azure OpenAI service, which gave users access to OpenAI’s API through the Azure platform. This allowed customers to use a variety of the company’s AI models including GPT-3 models, while being secure and compliant with other enterprise-grade capabilities that are a part of Azure. This year, Nadella announced that the Azure OpenAI service offering will now include access to DALL-E 2, the AI text-to-image generator that pioneered a surge of similar tools in the area. 

Microsoft was also ‘massifying’ DALL.E 2 using other important releases. The company launched Designer, a long-awaited web app that will compete with Canva. The application will help users create designs for posters, presentations, digital postcards and invitations using DALL.E 2. The application will have preset templates, photos, icons, headings and shapes that users can choose from to post directly on Instagram, LinkedIn and Facebook ads. Designer will be free for an initial period, after which it will be included in the Microsoft 365 personal and family subscriptions. Post this, certain functions will be free for non-subscribers. 

Another new application buttressed by DALL.E 2 is Image Creator that will be available on Bing and Edge in a few weeks. Users can send requests to DALL.E 2, which will be a frontend for the app to create art. Due to its similarities with DALL.E 2, Image Creator will be rolled out in a more calculated manner. Microsoft will launch the app in certain areas and collect feedback before expanding it to other geographies. 

Applications of GPT-3

GitHub Copilot, a language model fine-tuned to turn natural language prompts into coding suggestions, is powered by OpenAI Codex. Microsoft stated it will integrate Codex with Power Automate, a platform that automates organisational processes. Power Automate is a part of the company’s Power Platform, a home to the company’s bunch of low-code products like Power Apps and Power Pages. Power Automate will now use natural language processing capabilities to help users describe what function they want to automate with ease. This will quicken the workflow by pushing away the need to search or guess. 

“Creating flows is the foundation of Power Automate, and now you can describe what you want to automate in a sentence, and an AI-based copilot will build your flow within seconds,” Microsoft stated in a blog. “No searching or guesswork, just describe it and let the copilot work its magic,” the post added. 

“This new way to create flows runs on OpenAI Codex, an AI model descendant of GPT-3 that can translate natural language to code, in this case, Power Automate cloud flows. This is the same technology behind GitHub Copilot and Power Apps Ideas (preview). At launch, we’re supporting key Microsoft 365 connectors and in the coming months, we’ll expand to include more.”

Microsoft also rebranded Syntex adding a host of AI-based content management applications and relaunched it at Ignite. Syntex uses GPT-3 to perform a range of functions like reading, tagging and indexing content especially in the case of high volumes of text like customer transactions, invoices and contracts. Syntex reads the content, summarises it and highlights the essential portions of the text automatically. The tool also processes images and has the ability to tag approximately 10,000 automatically objects.

Microsoft also introduced Intelligent Recap, which acts like a virtual assistant at meetings. Recap will replay the meeting to users, assign tasks after the meeting, automatically create separate chapters from the meeting and offer personalised notes from the recording. The application is an addition to a number of a new range of features that have been included as an add-on to Microsoft Teams Premium.

Applications of Whisper 

Microsoft Teams premium has also added a live translation feature that delivers real-time translations for 40 spoken languages so that participants can read the captions in their own language. To avail the feature, the organiser of the meeting must have Teams premium. The feature will be temporarily available in preview until it becomes generally available. The feature is expected to help employees working remotely communicate easily. 

Microsoft’s speech recognition model Whisper, released in September is what underpins this feature. Whisper was trained on 680,000 hours of multilingual data scraped from the web, and improves understanding of accents, cuts through background noise and simplifies technical jargon. 

The post Does Microsoft’s Future Depend on OpenAI? appeared first on Analytics India Magazine.