I grew up hearing about “sapience” and “sentience” on TV shows like Star Trek (TNG).
I had a general idea of what they meant – usually an intelligent species on a planet somewhere was in trouble, and Jean-Luc and co had to swoop in to save the day.
On the show, maybe the scriptwriters knew the difference, maybe they didn’t.
Now, though, it’s becoming more important to understand the nuanced differences between these terms.
Let’s talk about the differences between sentience, sapience, self-awareness, consciousness, intelligence, autonomy, agency, cognition, and AGI.
Understanding this is crucial as we interact with more advanced AI, and as business leaders, we need to grasp the nuances of what AI experts are developing.
Table of Contents
Sentience
Sentience is the ability to feel or perceive things. A sentient being can experience pain, pleasure, happiness, and other sensations. Many animals, like dogs and cats, are sentient because they show signs of feeling pain or joy. Current language models, like ChatGPT, are not sentient. They process information and generate responses based on patterns in data but don’t have feelings or perceptions.
➡️ Example: In AI, understanding sentience is important for designing robots that interact with humans. For example, service robots in hospitals need to recognise human emotions to respond appropriately, even though they don’t feel these emotions themselves.
Sapience
Sapience refers to wisdom or the ability to think and reason. It’s about understanding complex concepts and making decisions based on that understanding. Humans are considered sapient because we can reason, solve complex problems, and think about abstract ideas. Potential future AGI aims to reach sapience, meaning it would be able to understand and reason like a human, though current language models are not there yet.
➡️ Example: Advanced AI systems used in financial markets for trading can be seen as striving towards sapience. They need to analyse vast amounts of data, understand market trends, and make decisions that maximise profits.
Self-awareness
Self-awareness is the ability to recognise oneself as an individual, separate from the environment and other individuals. Self-aware beings can reflect on their own existence and actions. Humans and possibly some higher animals, like great apes, dolphins, and elephants, show signs of self-awareness. Current language models do not have self-awareness. They can talk about the concept but don’t have an actual understanding of their own existence.
➡️ Example: Self-awareness in AI could revolutionise personal assistants. An AI that recognises its past interactions and adapts based on those could provide more personalised and context-aware assistance.
Consciousness
Consciousness is a broader term that includes sentience, sapience, and self-awareness. It’s the state of being aware of and able to think about one’s own existence, thoughts, and surroundings. Humans are conscious beings, and some animals might be considered conscious to a lesser extent. Future AGI might aim to achieve consciousness, but current language models do not possess consciousness.
➡️ Example: An AI with a form of artificial consciousness could potentially manage complex systems like smart cities, where it needs to understand and adapt to the needs of millions of residents in real-time.
Intelligence
Intelligence is the ability to learn, understand, and apply knowledge. It also involves problem-solving, recognising patterns, and adapting to new situations. Humans are highly intelligent, and animals show varying levels of intelligence. Current language models display a form of artificial intelligence (AI) by processing data and generating responses but lack true understanding or learning in the human sense.
➡️ Example: AI-powered chatbots used in customer service demonstrate intelligence by understanding queries and providing relevant responses, streamlining support and enhancing user experience.
Autonomy
Autonomy is the ability to make decisions and act independently. Many animals exhibit autonomy to some degree as they navigate their environments and make choices. Future AGI would aim to have high autonomy, making decisions without human intervention. Current AI systems can have limited autonomy, like robots performing specific tasks, but still rely heavily on human programming and guidance.
➡️ Example: Autonomous vehicles rely on AI to navigate and make driving decisions independently, reducing the need for human drivers and potentially improving road safety.
Agency
Agency is the capacity to act independently and make choices. It overlaps with autonomy but emphasises the active role in making decisions. Humans have agency, and certain animals exhibit it too. Current AI, including language models, lacks true agency as they operate within predefined parameters and don’t have independent decision-making power.
➡️ Example: AI in robotics used for disaster response may act with a degree of agency by making critical decisions in hazardous environments without waiting for human input, potentially saving lives.
The Difference Between Autonomy and Agency in AI
Autonomy and agency are confusingly-similar, but are separate. Here’s a breakdown of what they mean and when each is most useful:
Autonomy: Working Independently
- Refers to the ability to operate without external control
- Example: Self-driving cars that don’t need human input
- Best for tasks needing consistency and efficiency, like factory automation
Agency: Making Decisions
- Involves the capacity to make choices and take intentional actions
- Example: A robot that can assess situations and decide how to respond
- Ideal for complex scenarios requiring adaptability, such as emergency response
In essence, autonomy is about independence, while agency involves active decision-making. The choice between them depends on the specific needs of a situation.
Autonomous systems excel at repetitive tasks where rules are clear. Systems with agency shine in dynamic environments that require critical thinking and flexibility.
Understanding these differences is crucial for designing AI systems that best meet the needs of various industries and applications.
Cognition
Cognition is the mental process of acquiring knowledge and understanding through thought, experience, and the senses. It includes processes like thinking, knowing, memory, judgement, and problem-solving. Humans have advanced cognitive abilities, while animals have varying levels. Future AGI would aim to replicate human-like cognition. Current language models simulate some aspects of cognition by processing and generating text based on data.
➡️ Example: AI systems used in medical diagnosis apply cognitive processes to analyse symptoms, history, and medical data to provide accurate diagnoses and treatment recommendations.
Back to our friend Data
Data from Star Trek is a fascinating character to examine within this framework of terms.
As an android, Data exhibits characteristics that align with sapience and intelligence. He can reason, solve complex problems, and understand abstract concepts, which demonstrates his sapience.
Data is also self-aware, recognising himself as an individual distinct from others and capable of reflecting on his own existence and actions.
While he strives for sentience, often exploring human emotions and sensations, he is not truly sentient as he does not naturally experience feelings like pain or pleasure. Data’s high level of autonomy and agency allows him to make independent decisions and act on them, which is essential for his role on the Enterprise.
Data’s advanced cognition enables him to process vast amounts of information and make judgments based on that data.
Though he does not possess full consciousness or personhood as humans do, his journey in the series often revolves around seeking these deeper aspects of existence.
Terms / Examples
| Term | Definition | Example in AI |
|---|---|---|
| Sentience | Ability to feel or perceive sensations | None |
| Sapience | Wisdom or the ability to think and reason | Advanced AI systems in financial trading |
| Self-awareness | Recognising oneself as an individual | Potential future personal AI assistants |
| Consciousness | State of being aware and able to think about one’s existence | Potential future AGI managing smart cities |
| Intelligence | Ability to learn, understand, and apply knowledge | AI-powered customer service chatbots |
| Autonomy | Ability to make decisions and act independently | Autonomous vehicles |
| Agency | Capacity to act independently and make choices | AI in disaster response robots |
| Cognition | Mental process of acquiring knowledge and understanding through thought | AI medical diagnosis systems |
| AGI | AI that understands, learns, and applies knowledge across various tasks | Not yet achieved |
Conclusion
In summary, terms like sentience, sapience, self-awareness, consciousness, intelligence, autonomy, agency, cognition, and AGI are crucial in discussing the capabilities and qualities of animals, current AI, and future advanced AI systems. Each term highlights different aspects of mental processes, decision-making, and awareness. While animals and humans display various combinations of these traits, current AI, including language models like ChatGPT, are limited in their capabilities but are stepping stones towards more advanced and potentially conscious systems in the future.
