Examination of technology –Multimodal Foundation Models

Digital Platform Regulators Forum

Working Paper

Digital Platform Regulators Forum, ‘Examination of technology –Multimodal Foundation Models’ (Working Paper 3, 19 August 2024)

---

Background (from WP website)

This working paper, the third in a series exploring digital platform technologies, examines multimodal foundation models (MFMs) and their implications for consumer protection, competition, the media and information environment, privacy, and online safety within the digital platform context. 

A multimodal foundation model (MFM) is a type of generative AI that can process and output multiple data types. Large Language Models (LLMs), as explored in our previous working paper, are an example of a type of generative AI that focuses on a single data type. 

MFMs can generate multiple outputs including text, images, videos and even audio. These AI models are trained on exceptionally large datasets comprising various formats, allowing them to process and generate outputs that combine these different forms.