# The Evolution of AI Agents and Their Potential for Augmenting Human Agency In this transcript, Maya Akim, an AI content creator and agent builder, discusses various aspects of AI, focusing on generative AI, large language models (LLMs), and AI agents. She shares her insights and experiences, highlighting the potential and challenges of these technologies. This is an excellent analysis of the field, and well worthwhile watching. {% embed url="" %} Maya Akim, an AI content creator and agent builder {% endembed %} ### Introduction The rapid advancement of artificial intelligence (AI) technology has led to the development of increasingly sophisticated **AI agents** that are capable of autonomously performing tasks and making decisions. This video explores the philosophical underpinnings of agency and action, tracing the historical development of AI agents, and discussing how AI can be leveraged to augment human agency in various domains. ### The Philosophy of Action and Agency The concept of agency has been a subject of philosophical inquiry for centuries. Aristotle, in his work "Nicomachean Ethics," asserted that ***humans deliberate not about ends, but about means***. This implies that ***the road between one's current state and desired goal consists of a series of actions.*** In the 13th century, Ramon Llull developed logical operations using mechanical wheels, foreshadowing early computing. Later, Blaise Pascal's mechanical calculator and Ada Lovelace's groundbreaking algorithm further advanced the notion of machines performing tasks. The 20th century saw significant developments in the philosophy of action and agency. Alan Turing's seminal paper, "Computing Machinery and Intelligence," posed the question, "Can machines think?" and introduced the famous Turing test. The 1956 Dartmouth Conference coined the term "artificial intelligence," with its attendees optimistically proposing a 10-month study to make machines capable of using language, forming abstractions, and solving problems ### The Evolution of AI Agents Early AI systems, such as MYCIN[^1] in the 1970s, relied on ***symbolic AI and knowledge-based systems***. These systems employed rule-based reasoning, where a ***set of if-then statements*** guided the AI's decision-making process. However, this approach proved limited due to the inherent uncertainty, ignorance, and complexity of real-world problems. The 1980s marked a paradigm shift in AI, with the rise of ***probabilistic and statistical methods***, deep learning, and reinforcement learning. This new approach enabled AI agents to learn from their environment, adapting their behaviour based on rewards and punishments. The development of OpenAI's gym environments in 2016 further accelerated the training of AI agents in simulated environments. Modern AI agents possess several key characteristics, including ***autonomy, memory, reactivity, proactivity, and social ability***. They can perceive their environment, make decisions based on their goals and knowledge, and interact with other agents and humans to accomplish tasks. ### Augmenting Human Agency with AI AI agents have the potential to significantly augment human agency by performing tasks at a scale and speed that humans cannot match. For example, an AI agent can analyse vast amounts of data from various sources, such as social media, news articles, and academic papers, and provide summarised insights in a matter of minutes – a task that would take a human hours or even days to complete. AI agents can also assist with decision-making by considering a wide range of possibilities and scenarios. In complex domains such as finance, healthcare, and strategic planning, AI agents can help humans make more informed decisions by analysing data, identifying patterns, and predicting outcomes. However, it is crucial to recognize that ***AI agents are not a replacement for human judgment and oversight.*** As AI technology continues to advance, it is essential to develop robust frameworks for human-AI collaboration, ensuring that AI agents are aligned with human values and goals. ### Conclusion The development of AI agents has its roots in centuries of philosophical inquiry into the nature of agency and action. As AI technology progresses, AI agents are becoming increasingly capable of autonomously performing tasks and augmenting human agency. By leveraging the unique capabilities of AI agents, humans can make more informed decisions, tackle complex problems, and achieve goals more efficiently. However, it is critical to ensure that the development and deployment of AI agents are guided by ethical principles and human oversight to maximise their benefits while mitigating potential risks. [^1]: MYCIN was an early [backward chaining](https://en.wikipedia.org/wiki/Backward_chaining) [expert system](https://en.wikipedia.org/wiki/Expert_system) that used [artificial intelligence](https://en.wikipedia.org/wiki/Artificial_intelligence) to identify bacteria causing severe infections, such as [bacteremia](https://en.wikipedia.org/wiki/Bacteremia) and [meningitis](https://en.wikipedia.org/wiki/Meningitis), and to recommend [antibiotics](https://en.wikipedia.org/wiki/Antibiotic), with the dosage adjusted for patient's body weight — the name derived from the antibiotics themselves, as many antibiotics have the suffix "-mycin". The Mycin system was also used for the diagnosis of blood clotting diseases. MYCIN was developed over five or six years in the early 1970s at [Stanford University](https://en.wikipedia.org/wiki/Stanford_University). It was written in [Lisp](https://en.wikipedia.org/wiki/Lisp_programming_language) as the doctoral dissertation of [Edward Shortliffe](https://en.wikipedia.org/wiki/Edward_Shortliffe) under the direction of Bruce G. Buchanan, [Stanley N. Cohen](https://en.wikipedia.org/wiki/Stanley_N._Cohen) and others.