Deloitte Study Suggests AI Agents More Effective Than GenAI for Enterprise Productivity

New Delhi: A new study by Deloitte, a British professional services firm, suggested that Artificial intelligence (AI) nts could be a more effective tool compared to large langu models (LLMs) or GenAI applications. se AI nts can help open up new possibilities to drive enterprise productivity and program delivery through business process automation.

AI nts are automous systems that use AI to interact with ir environment, gar data, and complete tasks without needing human help. study explains that tasks once considered too complex for GenAI can w be handled securely and effectively with AI nts.

study also highlights difference between GenAI and AI nts. While LLM-powered chatbots can only follow "input-output" pattern of tritional apps, y often struggle with tasks that need multiple steps. 

"y (LLM or GenAI) conform to 'input-output' parigm of tritional applications and can get confused when presented with a request that must be deconstructed into multiple smaller tasks. y also struggle to reason over sequences, such as compositional tasks that require consideration of temporal and textual contexts. se limitations are even more prounced when using small langu models (SLMs), which, because y are trained on smaller volumes of data, typically sacrifice depth of kwledge and/or quality of outputs in favour of improved computational cost and speed," study said.

GenAI is typically used in simple applications like creating personalized s based on user search history or reviewing contracts. In contrast, AI nts can handle more complex tasks and use specialized digital tools for job.

"AI nts equipped with long-term memory can remember customer and constituent interactions--including emails, chat sessions and phone calls--across digital channels, continuously learning and justing personalised recommendations. This contrasts with typical LLMs and SLMs, which are often limited to session-specific information," study ded.

study also points out that businesses will need multiple AI nts working toger, as a single AI nt has its limitations. However, AI nts come with new risks, such as bias in decision-making and vulnerability to data breaches or cyberattacks, which could compromise sensitive information.

"A significant risk is a potential bias in AI. Algorithms and training data, which can le to inequitable decisions. ditionally, AI nts can be vulnerable to data breaches and cyberattacks, compromising sensitive information and data integrity," study concluded.

(With ncy inputs)