Is Agentic AI the next Big thing?
In today’s times where AI and its application are becoming commonplace, the mention of ‘Agentic AI’ begets the question “what’s the difference between AI and Agentic AI?” Simply put, while AI is typically reactive to input and follows the pre-defined rules, Agentic AI is capable of independent actions and decision-making.
For example, AI decision support systems can help with analysis and data presentation, but you control the final decision. On the other hand, Agentic AI can analyze situations, formulate strategies, and take action to achieve goals, with minimal human intervention.
Applications of Agentic AI/ Benefits:
Improving Business Operations:
Agentic AI agents can autonomously manage demand forecasting, handle complex logistics, optimize the supply chain, and check inventory levels. Businesses can easily process huge datasets and free themselves from real-time decisions, which would reduce costs and improve operational efficiency.
Transforming Customer Support Service:
Agentic AI can enhance customer support service by assisting software agents in providing personalized and 24/7 service beyond simple FAQs and automated responses. AI-enabled customer support agents would be able to understand written and oral queries, predict customer requirements, and resolve complex issues on their own.
Strengthening Cyber Security
AI agents can autonomously monitor network traffic, identify loopholes, detect anomalies, and run remediation measures against cyber threats in real-time without human supervision. It can help organizations enhance their security infrastructure and address complex security challenges.
Smart manufacturing
Agentic Workflow can coordinate production, quality checks, and logistics to enhance operational performance and productivity.
Education
AI agents could help personalize learning experiences by adapting content to each student’s needs, offering real-time feedback and supporting teachers with grading and administrative tasks. This allows educators to focus more on creative and interactive learning experiences.
What are the components of Agentic AI?
An AI agent has the following key components:
The user input or instructions from the human; the environment within which it operates, which can be physical or digital; Sensors, through which it perceives its environment; a control center, which involves complex algorithms and models; Percepts — the data center inputs it receives about its environment; Effectors — the tools it uses to take actions, which could be robotic arms or commands sent to other software systems. Actions, which represent the alterations made by effectors.
Types of Agentic AI
Simple Reflex Agents
Simple reflex agents follow predefined rules to map specific inputs to specific actions. These are simple rule-based machines or algorithms designed to provide static information and unable to change course
Examples are simple chatbots with predefined responses and automated email responders that send prewritten replies following specific triggers.
Model-based Reflex Agents
These are designed to track parts of their environment that are not immediately visible to them. They do this by using stored information from previous observations, allowing them to make decisions based on both current inputs and past experiences. These are more adaptable than simple reflex agents
Examples are — Smart thermostats that optimize energy usage by adjusting to current and historical temperature data, as well as user preferences, and modern irrigation systems that use sensors to collect real-time data on factors such as soil, moisture, temperature and precipitation, to optimize water dispensation.
Goal-Based Agents
Goal-based agents can take future scenarios into account. This type of agent considers the desirability of actions’ outcomes and plans to achieve specific goals, making them suitable for complex decision-making tasks.
Examples include — (a) Advanced chess AI engines that have the goal of winning the game, planning moves that maximize the probability of success and considering a long-term strategy and (b) Route optimization systems for logistics that set goals for efficient delivery and plan optimal routes by setting clear priorities
Utility-Based Agents
Utility-based agents employ search algorithms to tackle intricate tasks .
They use utility functions to assign a weighted score to each potential state, facilitating optimal decision-making in scenarios with conflicting goals or uncertainty.
Examples are —
(a) Autonomous driving systems that optimize safety, efficiency and comfort while evaluating trade-offs such as speed, fuel efficiency and passenger comfort, and
(b) Healthcare diagnosis assistants that analyse patient medical records, label patient data and optimize treatment strategy recommendations in cooperation with doctors.
Deloitte predicts that half of companies that use generative AI will have launched agentic AI pilots or proofs of concept that will be capable of acting as smart assistants, performing complex tasks with minimal human supervision.
Example of an AI agent system: Autonomous vehicle AI agent system
A human user gets into an autonomous vehicle (AV). The AV is comprised of an AI agent system that includes agents for perception, path planning, localization for finding its specific place on the road and control to steer and brake. The perception and localization agents are dedicated to continuously mapping the environment through sensors, the global positioning system (GPS) and cameras. The planning agent calculates the optimal trajectory by factoring in real-time traffic, weather and road conditions. The control agent handles the vehicle’s core mechanics, such as braking, accelerating and steering.
The AI agent infotainment system serves as the interface with the passenger and handles elements such as processing voice commands and adjusting routes, climate, entertainment or other in-car settings based on user preferences. All agents work together in a coordinated and centralized manner to ensure the vehicle reaches its destination safely and efficiently, prioritizing both passenger comfort and safety.
The Future or Agentic AI
The future of Agentic AI is all about multi-agent systems; Multi-agent systems (MAS) consist of multiple independent AI agents as well as AI agent systems that collaborate, compete or negotiate to achieve collective tasks and goals. These agents can be autonomous entities, such as software programs or robots. This allows agents to perform tasks in parallel, communicate with one another and adapt to changes in complex environments.
The architecture of a MAS is determined by the desired outcomes and the goals of each participating agent or system. There are several architectural types –
(1) Network architecture: In this setup, all agents or systems can communicate with one another to reach a consensus that aligns with the MAS’s objectives. For example, when autonomous vehicles (AVs) park in a tight space, they communicate to avoid collision. In this case, the MAS objective to prevent accidents aligns with each AV’s goal of safe navigation, allowing them to coordinate effectively and reach consensus.
(2) Supervised architecture: A “supervisor” agent coordinates interactions among other agents in this model. It is useful when agents’ goals diverge, and consensus may be unattainable. The supervisor can mediate and prioritize the MAS’s objectives while considering each agent’s unique goals, thereby finding a compromise. An example could be when a buyer and seller agent cannot reach an agreement on a transaction, which an AI agent supervisor then mediates.
While current efforts largely focus on developing AI agents within closed environments or specific software ecosystems, the future is likely to see multiple agents collaborating in different domains and applications. In MAS, different types of agent could work together to tackle increasingly complex tasks that require multistep processes, integrating expertise from various fields to achieve more sophisticated outcomes.
Risks
Technical Risks
These include risks from malfunctions due to AI agent failures and malicious use and security vulnerabilities.
Socio-economic Risks
Increased reliance on AI agents for social interactions, such as virtual assistants, AI agent companions, therapists, and so on could contribute to social isolation and possibly affect mental well-being over time
Employment Risks
The use of AI agents is likely to transform a variety of jobs by automating many tasks, increasing productivity, and altering the skills required in the workforce, thus causing partial job displacement.
Ethical Risks
The autonomous nature of AI agents raises ethical questions about their decision-making capabilities in critical situations. Many AI models operate as “black boxes”, making decisions based on complex and opaque processes, thereby making it difficult for users to understand or interpret how decisions are made.
To address these risks, there is a need for a multidisciplinary approach that includes diverse stakeholders, from scientists and researchers to psychologists, developers, system and service integrators, operators, maintainers, users, and regulators, all of whom are needed to establish appropriate risk management frameworks and governance protocols for the deployment of more sophisticated AI agent systems.
Look ahead
The ongoing development of AI agents is linked to increased autonomy, improved learning capabilities, enhanced decision-making abilities, and multi-agent collaboration. As the architecture and emerging use cases for AI agents continue to proliferate, the shift towards multi-agent systems is likely to continue. Increased autonomy plays an important part in the evolution of AI agents and creates opportunities for new applications while also presenting unique risks to society.
In the backdrop of a surge in the development of Agentic AI capabilities and proliferation of use cases, which will impact the global economy and alter the roles of humans in new ways, stakeholders from technical, civil society, business, and governance-facing communities need to collaborate, research, discuss and build consensus on novel governance mechanisms.
What must businesses do to adopt Agentic AI?
Agentic AI will require collaboration proficiency, either system working with people in different departments or with other AI models.
While business process automation is not new, it can be a complex undertaking to include rules and codes within automated process descriptions. Forrester Research analyst Brian Hopkins said more sophisticated neural networks are enabling technology like agentic AI to break down complicated tasks, use different tools to accomplish those tasks and communicate with other AI models.
For enterprises to automate processes using agentic AI, a team beyond just AI experts is required. It will involve company leaders who understand different facets of business operations as a whole and change management.
References:
Why Agentic AI is the Next Big Thing in the Business Industry
What are the risks and benefits of ‘AI agents’? | World Economic Forum
