Wollo University

College of Informatics

Department of Software Engineering

Assignment on Software Agent

By: Kaleb Tesfaye

ID: 1443/13

Date: January/2025
Kombolcha, Ethiopia
Reactive AI Agents

Intelligent agents in AI that make decisions just based on the status of the
environment now without considering past performance or internal states are known
as reactive agents. These agents just respond to the present circumstance by following
a predetermined set of condition-action rules or a mapping function; they are devoid
of any kind of memory or thinking ability.

According to the stimulus-response paradigm, reactive agents take in their

surroundings and act instantly in reaction to any new information. Their behavior is
dictated by immediate sensory input and hard-coded rules or functions; they lack an
explicit description of the environment or objectives.

Examples: Simple robots, thermostats, and reaction agents from video games are a
few examples of reactive agent devices that react to particular circumstances without
the need for long-term planning or memory.

Advantages

▪ Speed: Reactive agents are fast because they bypass complex reasoning
and respond directly to stimuli in real time.
▪ Simplicity: Their design and implementation are straightforward, relying
on rule-based or behaviour-based models.
▪ Robustness: They are reliable in dynamic and unpredictable
environments, handling changes without delays.
▪ Low Computational Cost: They consume fewer resources since they
don't require maintaining a detailed internal model or running complex
calculations.
▪ Scalability: Easily adaptable for scenarios involving multiple agents, as
their decentralized nature avoids bottlenecks.

Disadvantages

Lack of Planning: Reactive agents cannot foresee future states, limiting their
ability to handle multi-step tasks or long-term goals.
 Limited Flexibility: Their rule-based behaviour restricts adaptability to new or
unexpected tasks beyond pre-programmed rules.
Short-Term Focus: Decisions are made based solely on current stimuli, often
neglecting broader context or strategic outcomes.
Behavioural Conflicts: In complex systems, conflicting reactive behaviours can
arise without a mechanism for prioritization or resolution.
No Learning Capability: Reactive agents typically don't improve over time,
lacking mechanisms for adaptation or self-optimization.

Deliberative Agents

Conversely, deliberate agents are sentient entities with an internal state that can
reason about their choices in response to information, objectives, and beliefs. Based
on their observations and behaviours, these agents update their internal model of the
environment.

Deliberative agents possess the capacity to strategize, weigh options, and make
choices depending on their understanding and the expected outcomes of their choices.
They possess the ability to think through their objectives, come up with plans of
action to reach those objectives and choose the optimal course of action based on their
internal decision-making process.

When deciding what to do, deliberate agents often use strategies from disciplines like
decision theory, reasoning, and planning. They may investigate many options and
choose the best course of action depending on their objectives and the condition of the
environment at the time by using search algorithms, heuristics, and knowledge-based
systems.

Advantages

Strategic Decision-Making: They excel at handling complex, long-term tasks

requiring careful planning and foresight.

Logical and Consistent: Based on symbolic reasoning, they behave predictably

and make decisions aligned with predefined logic.
 Adaptability in Complex Scenarios: Using an internal world model allows them
to simulate and adapt to different scenarios effectively.

Goal-Oriented: They focus on achieving specific objectives, ensuring actions are

purposeful and well-directed.

Collaboration-Ready: Can work with other agents by sharing plans and

reasoning through protocols.

❖ Disadvantages

▪ Slow Response Time: Their computational reasoning process can delay real-time
reactions, making them less effective in dynamic environments.

▪ Resource-Intensive: Require significant computational power and memory to

maintain internal models and perform reasoning tasks.

▪ Limited Robustness: They can struggle to adapt to sudden changes or unexpected

situations due to their reliance on static planning.

▪ Complex Development: Designing and maintaining deliberative agents is

challenging due to the need for precise modelling and programming.

▪ Overhead in Dynamic Systems: Their reliance on comprehensive world models

makes them less efficient in rapidly changing or highly unpredictable environments.

❖ Hybrid Agents

Hybrid agents combine the strengths of both reactive and deliberative agents. They
utilize a layered architecture where reactive components handle immediate responses
to stimuli, while deliberative components manage long-term planning and decision-
making. The two components work in tandem, often with a mediator layer
coordinating their interaction to balance real-time reactivity with strategic reasoning.

Examples
Systems like RAP (Firby), Interrap (Müller), AIS (Hayes-Roth), and TouringMachine
(Ferguson) embody hybrid architectures, excelling in environments requiring both
immediate and planned actions.
❖ Advantages

Balance of Reactivity and Planning: Can handle both immediate actions and
long-term strategic goals effectively.

Flexibility: Adapts to dynamic environments while still achieving complex

objectives through planning.

Robust Performance: Combines the robustness of reactive systems in handling

unforeseen situations with the strategic depth of deliberative systems.

Scalable for Complex Tasks: Capable of addressing tasks requiring varying

levels of response complexity.

Error Handling: Reactive components can act as a fallback mechanism if

deliberative components fail or are slow to respond.

❖ Disadvantages

High Complexity: The architecture is more intricate, requiring careful

coordination between reactive and deliberative layers.

Resource Intensive: Requires computational resources to manage both reactive

responses and detailed planning processes.

Integration Challenges: Synchronizing the two components can lead to conflicts,

especially in fast-changing environments.

Increased Development Effort: Designing and implementing hybrid systems is

time-consuming and requires expertise in both reactive and deliberative
methodologies.

Potential Latency: Depending on the task, switching between layers can

introduce delays if not optimized effectively.