What is the Problem? (Introduction)

• There are many different technologies and methods being developed to create intelligent systems, but there is a big issue with how to define key terms used in these fields.
• Many fields like artificial intelligence, synthetic biology, and robotics contribute to developing these systems, but they often use different words or concepts for the same ideas.
• Researchers and scientists struggle to agree on terminology, and this can slow down collaboration and progress in developing these technologies.
• This paper discusses the need for a common language or a set of agreed-upon definitions to make communication and collaboration between researchers easier and more effective.

Why Do We Need a Common Language? (The Need for Consensus)

• Language plays a huge role in scientific communication, but it’s tricky because words can carry different meanings depending on the context and the background of the speaker.
• In emerging fields like artificial intelligence and synthetic biology, some terms have many different definitions. For example, the word “intelligence” had at least 71 different definitions just 15 years ago.
• Scientists need to agree on what key terms mean to avoid confusion, especially when multiple disciplines like biology, engineering, and philosophy are involved.
• Without clear definitions, different teams working on similar problems might have trouble understanding each other or might even waste time reinventing solutions.

What Terms Need to Be Defined? (Key Terms)

• Some terms in these fields are especially difficult to define because they are tied to complex processes or concepts that are difficult to measure or observe directly. Examples include terms like “consciousness” and “perception.”
• These terms can trigger emotional responses because they involve ideas that we feel strongly about, like the nature of intelligence or the experience of being alive.
• It’s important to create clear, agreed-upon definitions for these terms so that researchers can talk about them in a way that everyone understands. For example, “learning” can be measured through observable changes in behavior, while “phenomenal consciousness” is harder to define because we currently lack reliable ways to measure it directly.

What Approach Should Be Used? (Proposed Pathway Toward Consensus)

• To solve the problem of unclear definitions, the paper suggests a collaborative approach where experts from different fields come together to agree on a common set of definitions.
• They propose using a method called the “Delphi method,” which involves asking experts to answer open-ended questions about key terms, followed by rounds of feedback and refinement to reach a consensus.
• This method ensures that every expert has an equal opportunity to contribute their opinion and helps avoid biases that might come from face-to-face meetings or traditional voting systems.
• The idea is that by working together in a structured way, scientists from different fields can come to a shared understanding of key terms and definitions, making communication and collaboration easier.

Why Use Large Language Models (LLMs)? (Technology to Assist in Consensus)

• One helpful tool in this process could be large language models (LLMs) like GPT-4-Turbo. These models can analyze a wide range of existing definitions and help identify common patterns or discrepancies in how terms are used.
• LLMs can process large amounts of data quickly and help create a baseline of definitions that all researchers can use as a starting point for discussions.
• By using these models, the process of defining terms can be more efficient, and researchers can focus on refining ideas rather than starting from scratch.

What Happens After the Survey? (Refining and Reaching Consensus)

• Once experts have provided their opinions on the terms, the responses will be analyzed to identify areas where there is agreement and areas where more discussion is needed.
• The goal is to refine the definitions until a majority of experts agree on them, with the help of further rounds of feedback and discussion.
• If necessary, a voting system can be used to make final decisions on terms that remain contentious.

What Will the Result Be? (Outcome of the Consensus Process)

• The ultimate goal is to produce a clear set of definitions and guidelines that can be used across multiple fields, helping researchers communicate more effectively and collaborate more easily.
• By creating a shared vocabulary, the paper hopes to improve scientific understanding and progress in the development of intelligent systems.
• This process could help create a more structured and efficient approach to research, making it easier to bring together ideas and insights from diverse disciplines.

Who Can Get Involved? (Invitation for Collaboration)

• The paper invites researchers, philosophers, bioethicists, sociologists, and anyone else interested in the development of intelligent systems to join the collaboration and contribute to the effort of creating a shared vocabulary.
• Anyone who is interested can register to participate and help shape the future of the terminology used in this field.

观察到的情况？ (引言)

• 在许多领域，尤其是人工智能、合成生物学和机器人学中，正在开发各种各样的智能系统，但一个主要问题是如何定义这些领域中使用的关键术语。
• 这些领域的研究人员经常使用不同的词汇来表达相同的概念，这导致了术语的混乱和不一致。
• 这篇论文讨论了需要达成共识的语言，并提出了创建一套公认的术语定义，以便更容易地进行合作和有效沟通。

为什么需要统一的语言？ (达成共识的需求)

• 语言在科学交流中起着至关重要的作用，但它非常复杂，因为同一个词在不同的背景和使用者之间可能有不同的含义。
• 在人工智能和合成生物学等新兴领域，一些术语有许多不同的定义。例如，“智能”这个词15年前至少有71种不同的定义。
• 为了避免混淆，科学家们需要达成一致，明确术语的含义，尤其是在涉及生物学、工程学和哲学等多个学科时。
• 如果没有明确的定义，不同团队之间可能会因为理解上的差异而浪费时间，甚至无法有效地合作。

哪些术语需要定义？ (关键术语)

• 一些术语尤其难以定义，因为它们涉及复杂的过程或概念，这些概念很难直接衡量或观察。例如，“意识”和“感知”这些词。
• 这些术语可能会引发情感反应，因为它们涉及人类强烈的感知，比如智能的本质或生命的体验。
• 创建清晰且公认的定义对于这些术语至关重要，以便研究人员能够以大家都能理解的方式讨论它们。例如，“学习”可以通过行为的可观察变化来衡量，而“现象意识”则更难定义，因为我们目前缺乏可靠的直接衡量方法。

应该使用什么方法？ (建议的共识路径)

• 为了解决术语不清晰的问题，本文建议采取一种协作方法，邀请不同领域的专家共同达成统一的定义。
• 他们建议使用一种叫做“德尔菲方法”的方法，通过让专家回答关于关键术语的开放性问题，然后进行反馈和修正，直到达成共识。
• 这种方法确保每个专家都有平等的机会参与讨论，并帮助避免面对面会议或传统投票系统可能带来的偏见。
• 通过这种有结构的合作方式，来自不同领域的科学家能够达成一致，使沟通和合作更加顺畅。

为什么使用大型语言模型（LLM）？ (辅助共识的技术)

• 在这个过程中，一个有用的工具可能是大型语言模型（LLM），如GPT-4-Turbo。这些模型可以分析大量现有的定义，帮助识别术语使用中的共同模式或差异。
• LLM能够快速处理大量数据，并帮助创建一个所有研究人员都可以用作起点的术语定义基础。
• 通过使用这些模型，定义术语的过程可以更加高效，研究人员可以专注于改进想法，而不是从头开始。

调查后会发生什么？ (完善和达成共识)

• 一旦专家们提供了他们对术语的看法，回应将会被分析，以识别哪些地方达成了共识，哪些地方需要进一步讨论。
• 目标是通过进一步的反馈和讨论，逐步完善这些定义，直到大多数专家同意。
• 如果需要，投票系统可以用来对仍然存在争议的术语做出最终决定。

最终结果是什么？ (共识过程的结果)

• 最终目标是生产一套清晰的定义和指南，供多个领域的研究人员使用，从而帮助科学家更有效地沟通和合作。
• 通过创建共享的词汇，本文希望改善科学理解和智能系统发展中的进展。
• 这一过程有助于建立一个更有结构和高效的研究方法，使来自不同学科的想法和见解能够更容易地融合。

谁可以参与？ (协作邀请)

• 本文邀请人工智能、哲学、生物伦理学、社会学等领域的研究人员以及任何对智能系统开发感兴趣的人参与此次协作，并为创建共享词汇作出贡献。
• 任何感兴趣的人都可以注册参与，并帮助塑造这个领域术语的未来。