The artificial intelligence arms race is intensifying, with reports that tech giant Microsoft is building a new, monumental Large Language Model (LLM) named MAI-1. According to information from The Information, MAI-1 is said to perform comparably to some of the leading LLMs today, including OpenAI’s powerful GPT-4 model.
“I’m not sure why this is news, but just to summarise the obvious: we build big supercomputers to train AI models; our partner Open AI uses these supercomputers to train frontier-defining models; and then we both make these models available in products, and services so that lots of people can benefit from them. We rather like this arrangement,” stated Microsoft CTO Kevin Scott on his LinkedIn profile.
This groundbreaking development marks a significant shift for Microsoft, which has previously focused on investing heavily in OpenAI and leveraging its models, such as the GPT-4 family, to power various Microsoft Copilot products. With an investment of over $10 billion in OpenAI to utilize its AI models, this is the first time Microsoft is training an in-house AI model of such a massive scale that it could potentially compete with models from industry leaders like Anthropic and Google.
“Each supercomputer we build for Open AI is a lot bigger than the one that preceded it, and each frontier model they train is a lot more powerful than its predecessors,” Scott wrote, acknowledging Microsoft’s ongoing collaboration with OpenAI while also hinting at the company’s independent AI research efforts.
According to the report, Microsoft’s in-house AI is currently being trained, and the model, referred to as MAI-1, is said to be much larger than any of the smaller, open-source models that Microsoft had previously trained. This colossal size essentially means that the model would require significantly more computing power and training data, making it more expensive to develop and maintain.
Reports suggest that MAI-1 will have around 500 billion parameters, a staggering number compared to other models in the industry. For context, OpenAI’s GPT-4 is reported to have over a trillion parameters, while smaller models from Meta and Mistral come with 70 billion parameters. The work on this groundbreaking model is overseen by Mustafa Suleyman, CEO of Microsoft AI and a pioneering figure in the field of artificial intelligence.
Suleyman, who co-founded DeepMind and Inflection AI before joining Microsoft in March this year, is spearheading this ambitious project. Reportedly, Microsoft may preview MAI-1 close to the Build developer conference later this month, offering the world a glimpse into the potential of this cutting-edge AI model.
While the exact purpose of MAI-1 remains unknown, some reports claim that Microsoft is dedicating significant computing resources, such as Nvidia GPUs and vast amounts of data, to train this AI behemoth. The Information’s report states that “MAI-1 will be ‘far larger’ than the previous smaller, open source models Microsoft had previously trained, which means it will be more expensive.”
Microsoft’s clarification on the matter, provided by CTO Kevin Scott, sheds light on the company’s overarching strategy. Scott affirmed that “we build big supercomputers to train AI models; our partner Open AI uses these supercomputers to train frontier-defining models; and then we both make these models available in products, and services so that lots of people can benefit from them. We rather like this arrangement.”
However, Scott’s post also acknowledges Microsoft’s independent research on AI, stating, “Each supercomputer we build for Open AI is a lot bigger than the one that preceded it, and each frontier model they train is a lot more powerful than its predecessors.” He added that Microsoft will continue on this path, “building increasingly powerful supercomputers for OpenAI to train the models that will set the pace for the whole field – well into the future.”
While Microsoft has not officially confirmed the development of MAI-1, the reports surrounding this AI model have garnered significant attention, particularly in light of the company’s recent acquisition of talent and technology from Inflection AI. The tech giant’s foray into developing such a large-scale, in-house AI model could potentially shake up the industry and position Microsoft as a formidable competitor to the likes of Google and OpenAI in the realm of cutting-edge artificial intelligence.
As the AI arms race intensifies, the emergence of MAI-1 could mark a pivotal moment in Microsoft’s AI strategy, signaling a shift from relying solely on OpenAI’s models to developing its own powerful AI capabilities. With the potential to rival industry leaders, MAI-1 could propel Microsoft to the forefront of the AI revolution, positioning the company as a key player in shaping the future of this transformative technology.
Copyright©dhaka.ai
tags: Artificial Intelligence, Ai, Dhaka Ai, Ai In Bangladesh, Ai In Dhaka, Microsoft, Google, Claude, Future of AI
