On March 9, 2023, Microsoft Germany presented large language models as a disruptive force for companies and its Azure-OpenAI offering in detail at a one-hour hybrid information event called "AI in Focus - Digital Kickoff". At this event, Andreas Braun, CTO of Microsoft Germany and Lead Data & AI STU, casually mentioned the upcoming release of GPT-4. Microsoft has been working with OpenAI on multimodality, as was already announced with the release of Kosmos-1 in early March.
According to Dr. Alexander Braun, CTO of Microsoft iran number dataset Germany, GPT-4 will introduce multimodal models that will offer completely different possibilities. This also includes the processing of videos. If OpenAI is already coming up with publicly available video AI, this would of course be a huge surprise and would lead to even more tools for my AI newsletter .
What can we expect from GPT-4?
Dr. Alexander Braun, CTO of Microsoft Germany, said in the video that GPT-4 will be introduced next week and will contain multimodal models that will offer completely different possibilities - for example, videos. This would be a huge surprise and would lead to even more tools for newsletters.
What are multimodal models?
Multimodal models are a new trend in the AI industry that is becoming increasingly popular due to their high performance and versatility. These models are capable of processing different modalities by ingesting and analyzing visual, auditory and textual data simultaneously. For example, they can convert speech to text, identify and classify images and videos, and convert audio to text.
A major advantage of multimodal models is their ability to better understand and interpret human speech and gestures. They can process multiple modalities at once to better capture the meaning of a statement. One example of this is automatic video subtitling, where a multimodal model can analyze both speech and visuals to generate accurate subtitles.
Another benefit of multimodal models is their ability to enable natural human-machine interactions. For example, if a user gives an AI a complex command in natural language, the multimodal model can analyze both the text and the speech signal to understand better context. The result is a more accurate and effective response.