Gartner: 1/3 of GenAI Interactions to Employ Autonomous Agents by 2028

Gartner forecasts that by 2028, approximately one-third of interactions with generative AI (GenAI) services will utilize action models and autonomous agents for completing tasks.

These autonomous agents, which function without continual human intervention, utilize various AI techniques to make decisions and generate outputs. They possess the capability to learn from their environment and improve over time, enabling them to handle complex tasks efficiently.

Arun Chandrasekaran, Distinguished VP Analyst at Gartner, predicts a potential shift in human interactions with GenAI, moving from users prompting large language models (LLMs) to direct interaction with autonomous intent-driven agents. This evolution could lead to increased autonomy and better alignment with human goals.

Autonomous agents are expected to have a significant impact across various sectors. They can perform tasks such as chaining different model types, verifying model outputs, and automating complex business processes based on human intent. These capabilities can streamline operations and reduce the burden on business users across multiple industries.

In healthcare, autonomous agents can assist medical professionals in disease diagnostics, treatment planning, and patient care. In education, they can provide personalized learning experiences tailored to individual student needs. Similarly, in gaming, autonomous agents can enhance player experiences by providing immersive and realistic interactions. Additionally, in the insurance sector, autonomous customer service apps can handle policyholder interactions and assist with claims, fraud detection, and policy management, significantly reducing resolution times.

Gartner emphasizes the importance of defining clear objective functions for autonomous agents to deliver meaningful value. Organizations are advised to identify use cases where autonomous agents can add value by reducing human effort and skill requirements. Building a supportive architecture and acknowledging that autonomous agents complement rather than replace prompt engineering efforts are also essential steps. Achieving a balance between autonomy and control through extended pilots and rigorous agent monitoring is crucial for successful implementation.