Biography of AI pioneers

In this IP, I will compare my answers with AI (ChatGPT3.5)’s responses. I will be answering the questions below and comparing them with ChatGPT:

• Alan Mathison Turing, John McCarthy, Herbert Simon, Marvin Minsky and Timnit Gebru:
  • Who were these people?
  • How did each contribute to the development of artificial intelligence?
  • How did/does each think “intelligence” could be identified? 

For ChatGPT, I will use the prompt below to answer the questions:

Who is (name of the person) , How does each contribute to the development of artificial intelligence and How did each think “intelligence” could be identified? Try to answer within 50 words.

The ChatGPT version I have used for this prompt is version 3.5.

Alan Marthison Turing

My response:

Alan Mathison Turing is a British mathematician who first invented the artificial intelligence standard in the tech industry (Alan Turing – Education, Movie & Quotes, 2020). Alan thinks intelligence can be identified by how well the machine can play the “imitation game” (Turing, 1950).

ChatGPT:

Alan Mathison Turing, a British mathematician and computer scientist, made significant contributions to AI. He developed the concept of the Turing machine, a theoretical model of computation, and proposed the Turing Test as a way to identify intelligence in machines, suggesting that if a computer could mimic human conversation, it would demonstrate intelligence(ChatGPT, 2024).

John McCarthy

My response:

John McCarthy is a professor at Stanford University(McCarthy, n.d.) who invented LISP, which is a program used for artificial intelligence (ThinkingAllowedTV, 2011). John defines intelligence as the ability “to achieve goals in the world”(McCarthy, 2007).

ChatGPT:

John McCarthy was an American computer scientist who made significant contributions to artificial intelligence (AI). He coined the term “artificial intelligence” and developed the programming language Lisp, which played a crucial role in AI research. McCarthy believed that intelligence could be identified through problem-solving and reasoning abilities in machines, leading to AI’s development (ChatGPT, 2024).

Herbert Simon

My response:

Herbert Simon is a Nobel Prize winner renowned for his research into the decision-making process, specifically in the area of economics (Donovan,Paul, n.d.). He extended the boundaries of artificial intelligence by researching the human thought process. (Computer Pioneers – Herbert A. Simon, n.d.) Herbert defines intelligence as an ability to make decision making.

ChatGPT:

Herbert Simon was an American psychologist and computer scientist. He contributed to AI by developing the Logic Theorist, one of the first AI programs. Simon believed that intelligence could be identified by observing problem-solving and decision-making processes in machines. His work laid the foundation for AI research in cognitive science and problem-solving (ChatGPT, 2024).

Marvin Minsky

My response:

Marvin Minsky is a mathematician and a computer scientist who co-founded MIT’s AI lab with John McCarthy. He defines intelligence as the skill to process commonsense reasoning (“AI Pioneer Marvin Minsky Dies Aged 88,” 2016).

ChatGPT:

Marvin Minsky was an American cognitive scientist and computer scientist. He co-founded the Massachusetts Institute of Technology’s AI Lab and made significant contributions to AI. Minsky believed that intelligence could be identified by simulating human cognitive processes, and he worked on neural networks and robotics to achieve this goal, advancing AI’s understanding of perception and problem-solving.(ChatGPT, 2024)

Timnit Gebru

My response:

Timnit Gebru is a leader in AI research and ethics. She revealed that facial recognition AI technology is less accurate at identifying people of color and women. Although Timnit does not mention how intelligence can be defined, she mentioned the importance of reducing bias in AI (Hao, Karen, n.d.).

ChatGPT:

Timnit Gebru is an Ethiopian-American computer scientist known for her work in AI ethics and fairness. She co-authored influential research on bias in AI systems. Gebru emphasizes the importance of addressing social and ethical concerns in AI to ensure responsible development, and her work has sparked discussions on AI’s impact on society (ChatGPT, 2024).

Comparing AI with Humans

Similar to section above, I will be comparing the questions below with ChatGPT:

• How do “machine (programming) languages” differ from human (natural) ones?
• How does “machine (artificial) intelligence” differ from the human version?
• How does “machine learning” differ from human learning?

Programming Language vs Human Language

My response:

One of the major differences between programming languages and human languages is that programming languages have a set of rules and definitions that are already defined. Therefore, unlike human languages, programming does not change the rules depending on the context. (Ana, 2018). For example the word “chicken” can mean animal or coward depending on the context that is used. In programming languages, languages such as “if” or “else” will always have the same meaning regardless of the context.

ChatGPT:
Prompt used: How do “machine (programming) languages” differ from human (natural) ones. Answer within 100 words

Machine programming languages, like Python or C++, are designed for computers to understand and execute instructions efficiently. They are precise, structured, and consist of a limited set of commands and syntax. In contrast, human natural languages, such as English or Spanish, are complex, ambiguous, and evolved for communication between people. They have nuances, idioms, and cultural context that computers struggle to grasp. While humans use natural languages for diverse communication purposes, machine languages are specialized for programming tasks, with strict rules to ensure accurate and unambiguous instructions for computers, making them fundamentally different from the rich and expressive nature of human languages (ChatGPT,2024).

How does AI differ from the human version?

My response:

Compared to humans, AI performs well when they do specific tasks, and is unable to perform tasks that are outside of the training data. (Chollet, 2019, p.3) Humans, on the other hand, are better at tasks that require generalization, a skill that involves dealing with tasks not connected to prior knowledge.

ChatGPT:

Prompt used: How does AI differ from the human version? Answer within 100 words

AI differs from human intelligence in key areas: AI excels in specific, data-driven tasks but lacks the broad generalization capabilities of humans. It relies heavily on large datasets for learning, while humans can learn from fewer examples and apply abstract thinking. AI struggles with context, common sense, and emotional understanding, aspects where humans naturally excel. Additionally, humans possess creativity, moral reasoning, and adaptability, skills that AI currently cannot replicate. Essentially, while AI is powerful within its scope, it lacks the versatile, intuitive, and ethical dimensions of human intelligence (ChatGPT, 2024).

How does “machine learning” differ from human learning?

My response:

Compared to humans, machine learning relies heavily on training data (Heilweil, 2020). Due to this nature, if the training data is biased, the accuracy of AI data detection will also be biased. For example, Buolamwini discovered that the dataset of faces collected for face recognition AI contained only 5% women of color, which potentially caused inaccuracy of AI facial detection for well-known women of color celebrities such as Serena Williams (Buolamwini, 2019).

ChatGPT:

Prompt used: How Does machine learning differ from human learning? Answer within 100 words

Machine learning differs from human learning in its approach and capabilities. It requires vast amounts of data to identify patterns and make decisions, often focusing on specific tasks. Machine learning lacks the ability to understand context or apply common sense, and it cannot easily transfer knowledge between different tasks. In contrast, humans can learn from limited data, understand context, and apply knowledge broadly and flexibly across various domains. Human learning is also driven by curiosity, emotions, and social interactions, elements absent in machine learning. While machine learning is efficient for data-driven tasks, it lacks the adaptability and depth of human learning. (ChatGPT, 2024).

How my response differed from AI

Comparing my answer with that of ChatGPT 3.5, I find that the answers from ChatGPT are generally accurate. For example, the answer that ChatGPT provided for Herbert is more detailed than the answer that I have provided.

However, what ChatGPT lacks is the ability to cite sources. When I asked ChatGPT, “Can you provide a citation on where you found this info?”, ChatGPT replied, “As an AI developed by OpenAI, I don’t have the ability to provide direct citations from external sources or access to my training data” (ChatGPT, 2024). ChatGPT claimed that its provided information is part of general knowledge and recommended looking into academic databases for verification.I find this part of ChatGPT is problematic because this lack of citation makes it more difficult to verify the facts.

I also noticed that in formulating answers, I used analogies or specific examples to aid understanding. For example, in my response about programming and human language, I provided examples of how the meaning of the word “chicken” can change depending on the context.

In terms of distinguishing between AI and human writing, I would say it is almost impossible. The only noticeable difference is the lack of citations in AI responses, but in terms of sentence structure and vocabulary, the writing style is very similar to that of a human.

---

References

This work by Rie Namba is licensed under CC BY 4.0