Usefulness grounds truth - Inverted Passion

🌈 Abstract

The article discusses the debate around whether large language models (LLMs) are intelligent, and argues that the question of "intelligence" is not as useful as asking what LLMs can actually do. It suggests that instead of focusing on abstract definitions, we should evaluate the usefulness and real-world impact of LLMs and other technologies.

🙋 Q&A

[01] The Debate Around LLM Intelligence

1. What are the key points made about the debate on whether LLMs are intelligent?

• The article states that debates on whether LLMs are intelligent often devolve into debates on what LLMs can or cannot do, rather than focusing on the core question.
• It argues that the answer to whether something is "intelligent" depends on how one defines intelligence, and that this question quickly becomes circular.
• The article suggests that instead of debating abstract concepts of intelligence, we should focus on the real-world usefulness and impact of LLMs and other technologies.

2. How does the article frame the question of truth and reality?

• The article argues that there is no "Platonic realm" of absolute truth that is disconnected from the world we inhabit.
• It suggests that instead of asking what is "true", we should ask what is "useful" - i.e. what makes a real difference in the world.
• The article draws on the pragmatist philosophy of William James, which emphasizes focusing on the practical, real-world impact of ideas and theories.

3. What examples does the article provide to illustrate the idea of truth as usefulness?

• The article uses the example of scientific theories, where the "truth" of a theory is judged by its ability to make accurate predictions that are borne out by experiments and observations.
• It also gives the example of the question "do you love me?", arguing that the meaning of love is defined by how it manifests in actions, not just abstract declarations.

[02] Reframing the Question of LLM Intelligence

1. How does the article suggest reframing the question of whether LLMs are intelligent?

• Instead of asking whether LLMs are "intelligent", the article proposes reframing the question in terms of what specific capabilities LLMs have, such as summarizing articles, driving cars safely, or writing scientific papers.
• It argues that the term "intelligence" is not very useful, and that we should focus on the practical, real-world usefulness and impact of LLMs and other technologies.

2. What is the article's overall conclusion about the debate on LLM intelligence?

• The article concludes that we should "forget about truth" and instead focus on what is "useful" - i.e. what makes a real difference in the world.
• It suggests that the question of whether LLMs are "intelligent" is less important than understanding their specific capabilities and how they can be applied to solve real-world problems.