What is the situated approach in AI? — Klu
What is the situated approach (AI)?
The situated approach in AI refers to the development of agents that are designed to operate effectively within their environment. This approach emphasizes the importance of creating AI systems "from the bottom-up," focusing on basic perceptual and motor skills necessary for an agent to function and survive in its environment. It de-emphasizes abstract reasoning and problem-solving skills that are not directly tied to interaction with the environment.
Key characteristics of the situated approach include:
- Bottom-up design — It starts with elementary behaviors that can be combined to create more complex behaviors.
- Behavior-based — It does not rely on symbolic descriptions of the environment but on a model of the entity's interactions with its surroundings.
The situated approach was proposed as an alternative to traditional AI methods that were popular before the mid-1980s, which often involved disembodied intelligences that interacted with the world primarily through abstract representations. The situated approach, also known as Nouvelle AI, seeks to build embodied intelligences that are situated in the real world, engaging directly with tangible physical objects.
This approach is rooted in early insights from Alan Turing, who suggested that machines equipped with sense organs could learn directly from the real world, and the philosophy of Hubert Dreyfus, who argued that intelligent behavior cannot be fully captured by symbolic descriptions alone.
In the context of socially situated AI, the approach also involves learning from human interaction, which allows AI agents to improve their performance through social engagement.
The situated approach is considered a viable alternative to traditional computationalist approaches and is seen as essential for developing 'true' intelligence, both in natural and artificial systems. It acknowledges that intelligent behavior derives from the environment and the agent's interaction with it.
What are some examples of AI systems that use the situated approach?
The situated approach in AI emphasizes the importance of the environment and context in which AI systems operate. It is a bottom-up approach that relies on elementary behaviors, which can be combined to implement more complex behaviors. It does not rely on a symbolic description of the environment, but rather on a model of the interactions of the entities with their environment.
Here are some examples of AI systems that use the situated approach:
Robotics — AI in robotics often uses the situated approach. For instance, robots use AI for real-time awareness and decision-making, enabling them to act much quicker than human capabilities allow. This includes tasks like navigation, detection, and determining reactions accordingly. The situated approach has been influential in the development of robotic systems, as they need to operate in a wide range of environments.
Healthcare — AI in healthcare often uses the situated approach to tailor personalized care tracks for managing medical conditions. For example, Augmedix offers a suite of AI-enabled medical documentation tools for hospitals, health systems, individual physicians, and group practices. These products use natural language processing and automated speech recognition to increase productivity and improve patient satisfaction.
Transportation — AI in transportation uses the situated approach to improve safety and efficiency. For instance, self-driving vehicles use AI to interpret massive amounts of data gathered by the vehicle, enabling real-time awareness and decision-making.
Socially Situated AI — This is a reinforcement-learning framework that enables agents to uncover useful social interactions. It's critical for applications where effective human interactions are necessary to improve an AI agent's ability, including human-computer interaction, interactive robotics, personalized conversational agents, and accessible technology.
Nouvelle AI — This approach attempts to build embodied intelligences situated in the real world. It was first proposed by Rodney Brooks in the early 1990s and has been influential in the development of robotic systems.
These examples illustrate how the situated approach to AI is used in various fields to create systems that can interact effectively with their environment.
Supervised Learning
Supervised learning is a machine learning paradigm where a model is trained on a labeled dataset. The model learns to predict the output from the input data during training. Once trained, the model can make predictions on unseen data. Supervised learning is widely used in applications such as image classification, speech recognition, and market forecasting.
What is game theory?
Game theory in the context of artificial intelligence (AI) is a mathematical framework used to model and analyze the strategic interactions between different agents, where an agent can be any entity capable of making decisions, such as a computer program or a robot. In AI, game theory is particularly relevant for multi-agent systems, where multiple AI agents interact with each other, each seeking to maximize their own utility or payoff.