TechCapsules

AI-summarised brief · reviewed before publication

According to a recent prediction by IT consultancy Gartner, more than 40 percent of agentic AI projects are expected to be cancelled by the end of 2027. This is attributed to factors such as rising costs, unclear business value, or insufficient risk controls. Despite this, a notable 60 percent of agentic AI projects are expected to be retained. This is somewhat remarkable, given that the rate of successful task completion for AI agents is only around 30 to 35 percent for multi-step tasks. This rate was measured by researchers at Carnegie Mellon University and Salesforce. Further complicating the matter is Gartner's contention that most of the purported agentic AI vendors offer products or services that do not actually qualify as agentic AI. Agentic AI is a type of AI that uses a machine learning model connected to various services and applications to automate tasks or business processes. These AI agents operate in an iterative loop, responding to input using applications and API services. A prime example of agentic AI in action would be a task such as, "Find all the emails I've received that make exaggerated claims about AI and see whether the senders have ties to cryptocurrency firms." An AI model authorized to read a mail client's display screen and access message data would be able to interpret and carry out the natural language directive more efficiently than a programmatic script or a human employee. In theory, the AI agent would be able to formulate its own definition of "exaggerated claims," whereas a human programmer might find the text parsing and analysis challenging. A human employee could identify the AI hype in a given inbox, but it would likely take longer than a computer-driven solution. The concept of software that simply accepts orders and executes them efficiently, correctly, affordably, and without fuss is a recurring theme in science fiction. This ideal is exemplified in the famous Star Trek: The Next Generation command, "Tea, Earl Grey, hot," where the computer promptly responds with the requested beverage.