Microsoft is taking a big step forward in the artificial intelligence (AI) space with the development of a new AI language model, named MAI-1, looking to rival the offerings from tech giants like Google and OpenAI.
This move was reported on Monday by the Information, citing sources familiar with the project. According to the report, MAI-1 is being spearheaded by Mustafa Suleyman, a recent Microsoft hire and co-founder of Google DeepMind, now heading Microsoft’s newly established consumer AI unit.
The MAI-1 model, is noted to be “far larger” than Microsoft’s prior smaller, open-source models. It will possess approximately 500 billion parameters, a big increase but still half of what OpenAI's GPT-4 model entails. Microsoft’s last month's launch of the Phi-3-mini, a smaller and more cost-effective AI model, highlights the company’s strategy to cater to a broader client base while also pushing the envelope with more sophisticated, expansive models like MAI-1.
Despite the anticipation surrounding its potential capabilities, the exact purpose of MAI-1 is yet to be determined, hinging on its performance during development stages. Microsoft might reveal more details about this model at its Build developer conference later this month, offering insights into how it might integrate into or enhance the company's suite of productivity software.
Microsoft’s venture into developing MAI-1 comes amidst its billions of dollars worth of investment in OpenAI, utilising the latter’s AI technologies such as ChatGPT across its products, thus securing an early advantage in the generative AI market. However, with MAI-1, Microsoft appears set to not just leverage partnerships but also enhance its proprietary AI capabilities, which could reshape its position and influence in the rapidly evolving tech landscape.
_20240605152515_original_image_42.webp)
