According to a report from Heise Online, an executive from Microsoft Germany casually mentioned the upcoming release of GPT-4 at the KI im Fokus (AI in Focus) Digital Kickoff event.
The event was designed to answer questions about large language models, including Dall-E 2, GPT-3.5, and Codex, available as a part of the ChatGPT preview in Azure OpenAI Service, used by over 1,000 customers.
Specifically, they wanted to answer questions related to upcoming AI developments and what they mean for company leaders, executives, and employees.
During this event, Andreas Braun, CTO and Lead Data & AI STU at Microsoft Germany, was quoted as saying, “We will introduce GPT-4 next week…” with multimodal models and video generation capabilities.
Braun considers this to be a game changer as LLMs will allow people to ask a question in one language and get an answer in another, creating global collaboration opportunities without language as a barrier.
CEO of Microsoft Germany, Marianne Janik, joined Braun to explain AI disruption for companies, calling it an “iPhone moment” that will revolutionize the way repetitive tasks are handled.
Janik does not believe this will lead to job losses, but rather opportunities for employees to learn how to use AI to improve their processes and output.
As far as customer data is concerned, Janik noted that models were not trained with customer data.
Clemens Sieber, Senior Specialist for Artificial Intelligence and Machine Learning, and Holger Kenn, Director of Business Strategy, explained the ways that AI can make companies more efficient.
One example, shared in Sieber’s GitHub repository, shows how call centers can save hundreds of working hours per day by having support calls automatically summarized and entered into their customer support system.
The project that could save hundreds of working hours daily was created in two hours, demonstrating how AI allows you to dramatically improve processes in a short amount of time.
Sieber also noted that, due to concerns about the accuracy of AI-generated responses, Microsoft is working on “confidence metrics” to allow thumbs up or down feedback.
Kenn described the opportunities that come with multimodal models, such as text prompts translated to images, music, and video.
While reported mentions of GPT-4 from the AI in Focus event have not led to an official announcement, the timing does seem fitting.
Microsoft recently added VisualChatGPT to GitHub. It would allow you to send and receive images while chatting. The paper explaining the use of Visual Foundation Models with ChatGPT was submitted on March 8th.
Next week, Microsoft will host a livestream event on March 16th, The Future of Work with AI.
Satya Nadella, CEO, and Jared Spataro, Corporate Vice President, Modern Work & Business Applications, will explore how AI will reinvent productivity for every person and organization, changing the way we all work.
The event is rumored to cover AI integrations in Microsoft 365 products like Teams Premium, Viva Engage, Loop, Outlook, and Word. According to the Microsoft blog, new developments would be coming in March.
Microsoft reinforces their commitment to responsible AI development with application-level protections, technical protections, process and policy protections, and documentation on design guidelines and transparency on what has been tested.
I have reached out to Microsoft Germany for confirmation and additional comments.
