Microsoft Debuts Smallest AI Model as AI Eludes Small Businesses

Microsoft has introduced its smallest artificial intelligence (AI) model, aimed at businesses with limited resources. The company’s Phi-3-mini can perform tasks like content creation for social media while using less data, Reuters reported Tuesday (April 23), while outperforming models twice its size. The report notes that Microsoft argues that because smaller AI models are built […]

The post Microsoft Debuts Smallest AI Model as AI Eludes Small Businesses appeared first on PYMNTS.com.

Microsoft has introduced its smallest artificial intelligence (AI) model, aimed at businesses with limited resources.

The company’s Phi-3-mini can perform tasks like content creation for social media while using less data, Reuters reported Tuesday (April 23), while outperforming models twice its size.

The report notes that Microsoft argues that because smaller AI models are built to handle simpler tasks, they are easier for companies with fewer resources. The company gives the example of a business using Phi-3 to summarize a long document or extract industry trends from market research reports.

Microsoft says Phi-3-mini will be available immediately on its cloud service platform Azure’s AI model catalog, as well as the machine learning model platform Hugging Face, and Ollama, a framework for running models on a local machine.

The launch comes on the heels of a report by Stanford University showing that small and mid-sized businesses (SMBs) will struggle to keep pace with tech giants like the Microsoft-backed OpenAI in creating their own AI models.

In an interview published here last week, Nestor Maslej, editor of that report, spotlighted the study’s findings on the growing AI divide between large and small companies. As large companies pour billions into AI research and development, smaller firms lack the resources and talent to compete. 

“A small or even medium-sized business will not be able to train a frontier foundation model that can compete with the likes of GPT-4, Gemini or Claude,” Maslej said. 

“However, there are some fairly competent open-source models, such as Llama 2 and Mistral, that are freely accessible. A lot can be done with these kinds of open-source models, and they are likely to continue improving over time. In a few years, there may be an open, relatively low-parameter model that works as well as GPT-4 does today.”

Meanwhile, Microsoft earlier this week teamed with Cognizant to expand the adoption of  generative artificial intelligence (GenAI) solutions by enterprises, with the companies hoping to make Microsoft’s GenAI and Copilots available to millions of users.

“We’re committed to helping them harness the power of generative AI at scale, and Microsoft Copilot is a proven tool that can bring transformative gains, unlocking talent and potential in ways we can only imagine,” Ravi Kumar S, Cognizant’s CEO, said in a news release.

The post Microsoft Debuts Smallest AI Model as AI Eludes Small Businesses appeared first on PYMNTS.com.