Small, Task-Specific AI Models To Dominate General-Purpose Large Language Models By 2027: Gartner

The need for contextualised, reliable and cost-effective solutions is driving a shift in the kind of artificial intelligence models that enterprises implement. By 2027, businesses will use small, task-specific AI models at a rate at least three times higher than that of general-purpose large language models (LLMs), according to recent research from Gartner Inc.

While general-purpose LLMs provide robust language capabilities, their response accuracy declines for tasks requiring specific business domain context, driving the shift towards small, task-specific AI models.

The shift towards specialised models that are based on certain functions or domain data is being driven by the variety of jobs in corporate workflows and the need for increased accuracy. By using less computing power and responding more quickly, smaller, task-specific models lower operating and maintenance costs, the Gartner report noted.

By using retrieval-augmented generation (RAG) or fine-tuning techniques, businesses can construct customised LLMs for specific tasks. Enterprise data becomes a crucial differentiator in this process, requiring data preparation, quality checks, versioning, and general administration to guarantee that relevant data is organised to satisfy the fine-tuning specifications.

“As enterprises increasingly recognise the value of their private data and insights derived from their specialised processes, they are likely to begin monetising their models and offering access to these resources to a broader audience, including their customers and even competitors,” said Sumit Agarwal, VP analyst at Gartner. “This marks a shift from a protective approach to a more open and collaborative use of data and knowledge.”

According to Gartner, enterprises looking to implement small task-specific AI models must consider the following recommendations:

Pilot Contextualised Models: Implement small, contextualised models in areas where business context is crucial or where LLMs have not met response quality or speed expectations.

Adopt Composite Approaches: Identify use cases where single model orchestration falls short, and employ a composite approach involving multiple models and workflow steps.

Strengthen Data And Skills: Prioritise data preparation efforts to collect, curate and organise the data necessary for fine-tuning language models. Invest in upskilling personnel across technical and functional groups such as AI and data architects, data scientists, AI and data engineers, risk and compliance teams, procurement teams and business subject matter experts, to drive initiatives.