Microsoft eyes new in-house AI model to match rivals

Getting your Trinity Audio player ready...

Microsoft (NASDAQ: MSFT) is designing a new artificial intelligence (AI) model to compete with offerings from other industry players as it seeks to pull away from the rest of the pack.

The incoming large language model (LLM) will be an in-house offering, according to a Seeking Alpha report. Dubbed the MAI-1, the model could have as much as 500 billion parameters, putting it in the thick of things with offerings from Google (NASDAQ: GOOGL) and OpenAI.

Generally speaking, AI models with larger parameters can learn complex patterns from training data compared to smaller ones. For context, OpenAI’s ChatGPT-3 comes loaded with 175 billion parameters, while Meta’s (NASDAQ: META) Llama 2 has 65 billion parameters.

With half a trillion parameters, MAI-1 is by far Microsoft’s largest open-source model, and it is expected to rival Google’s large Gemini models.

Although details are sparse, sources say Microsoft could unveil a preview of the in-house model during its Microsoft Build developer conference scheduled this month.

Microsoft has the edge to compete with other AI firms as its expertise in the tech continues to grow, leading to the development of several products laced with AI functionalities, including Copilot.

“AI models are used in almost every one of our products, services, and operating processes at Microsoft, and the teams making and operating things on occasion need to do their own
custom work, whether that’s training a model from scratch, or fine-tuning a model that someone else has built,” said Microsoft CTO Kevin Scott.

The company found success with lightweight AI models with its Phi models, with the third iteration containing only 3.8 billion parameters but still demonstrating advanced coding, math, and language abilities. Microsoft’s internal AI capabilities have seen OpenAI rely on its supercomputers to train its widely popular generative AI chatbots.

“There will be more of this in the future too,” added Scott, hinting at the new internal model. “Some of these models have names like Turing and MAI.”

New hires get to work

In March, Microsoft confirmed a hiring spree of new executives to head its AI and other
emerging technologies unit, poaching staff from industry rivals in a strong show of strength. Going forward, new hire Mustafa Suleyman will lead the design of MAI-1, with former employees of Inflection AI forming the rest of the new team.

The new move makes it the first time that one person will spearhead Microsoft’s consumer AI offerings after years of experimentation. Previously, the company’s AI efforts employed a staccato approach that saw it fall down the pecking order in the grand scheme of things.

In order for artificial intelligence (AI) to work right within the law and thrive in the face of growing challenges, it needs to integrate an enterprise blockchain system that ensures data input quality and ownership—allowing it to keep data safe while also guaranteeing the immutability of data. Check out CoinGeek’s coverage on this emerging tech to learn more why Enterprise blockchain will be the backbone of AI.

Watch: Blockchain & AI—there should be confluence between these tech