Beyond Attention: Attention is NOT all you need 超越关注：DIKWP框架的人工意识评估

• September 2023

• DOI: 

• 10.13140/RG.2.2.10379.05920

• Yucong Duan

Beyond Attention: Attention is NOT all you need

超越关注：DIKWP框架的人工意识评估

段玉聪（Yucong Duan）

DIKWP-AC人工意识实验室

AGI-AIGC-GPT评测DIKWP（全球）实验室

DIKWP research group, 海南大学

duanyucong@hotmail.com

Abstract:

The landscape of artificial intelligence (AI) is rapidly evolving, with AI systems becoming increasingly integrated into various aspects of human life. However, one of the fundamental challenges in AI development is achieving a level of consciousness that aligns with human cognition. While human consciousness is known for its cognitive relativity, where each individual's awareness and experiences are unique, AI systems require a different approach—one that seeks absolute standards. In this Nature paper, we delve into the concept of achieving absolute standards in AI consciousness and explore the essential role of the DIKWP language as a necessary component in constructing artificial consciousness.

摘要： 人工智能（AI）的快速发展已经彻底改变了我们生活的方方面面。然而，伴随着AI技术的广泛应用，我们也面临着一系列挑战，包括透明度、可解释性和伦理责任等方面。本文提出了DIKWP（数据、信息、知识、智慧、目的）人工意识评估标准，这一框架超越了以往对“关注”的简单理解，强调了数据、信息、知识、智慧和目的在实现AI系统人工意识方面的关键作用。我们将深入探讨DIKWP框架的各个要素，以及它如何帮助解决当前AI领域所面临的关键问题，推动AI技术朝着更为负责任和智能化的方向发展。

引言

人工智能技术的快速发展已经改变了我们的日常生活、工作和社会互动方式。从智能助手到自动驾驶汽车，AI系统已经成为现代社会的不可或缺的一部分。然而，AI的快速发展也带来了一系列严峻的问题，包括如何确保AI系统的决策是可靠和负责任的，如何解释AI系统的行为，以及如何保障AI系统与人类之间的交流是有效和透明的。为了回应这些挑战，我们需要建立更为全面和可信的AI评估标准，以确保AI系统的可控性和可信度。

DIKWP：重新定义AI的认知

《Attention is all you need》？ NO

DIKWP 团队：《Attention is NOT all you need》

For D : attenion is statistics

Data::= Attention     Yes

For I : attention’s attention is

Information::= (Attention)2     NO

For K :attention’s attention’s attention is

Knowledge::= (Attention)3     NO

For W: Wisdom::=Information originates in human civiliation and culture.

For P : Purpose is description of human requirement and expection.

Purpose ::=<Input(DIKWP ), Output(DIKWP )>

Purpose driven: is the execution of DIKWP

The combination of Description and Execution of  DIKWP  ensures :

the conformance of the Human - Machine cooperation

重新定义关注：作为统计的关注

传统上，AI领域一直在强调“关注”（attention）的重要性，这一概念首次在《关注是你所需要的一切》中得到广泛应用。然而，DIKWP团队提出了一个截然不同的观点，将关注视为数据处理的统计组成部分。在DIKWP框架中，数据被认为是AI认知的起点，而关注仅仅是数据处理的一种统计方式。这一重新定义有助于我们更全面地理解AI的认知过程，不再局限于仅仅关注。

信息不仅仅是关注的平方

在DIKWP框架中，信息被表示为（关注）2，这意味着信息不仅仅包括关注的平方。这一观点强调了信息的多层次性，信息的获取需要对数据进行更深层次的抽象和处理，以符合人类的意图和需求。信息在DIKWP框架中被看作是AI系统对世界的更深层次理解。

知识是多层次的理解

知识作为DIKWP框架的一个重要要素，超越了（关注）3。DIKWP团队认为，知识的获取是实现AI意识的关键一步，它源自于AI系统对抽象和综合信息的能力。知识的多层次性使AI系统能够更全面地理解和解释世界，进一步推动了AI系统向人工意识的发展。

源自人类文明的智慧

智慧在DIKWP框架中源自于人类文明和文化的积累。与仅仅依赖关注驱动算法不同，DIKWP框架强调了AI系统需要汲取人类历史的集体智慧，以做出伦理和负责任的决策。这一观点有助于确保AI系统的决策不仅是智能的，还是道德和负责任的。

目的驱动的AI

在DIKWP框架中，目的是对人类需求和期望的全面描述。目的驱动的AI系统在执行任务时清晰了解其目标，与人类意图和目标保持一致。这确保了AI系统与人类用户之间的和谐合作，使得AI系统更具有实用性和可行性。

描述与执行的交汇

DIKWP框架强调了描述（目的）与DIKWP组件的执行之间的协同作用。这种协同作用确保了AI系统的行为是可解释的，并与用户期望一致。DIKWP团队认为，仅仅依赖关注是不足够的，我们需要包括对数据、信息、知识、智慧和目的的全面描述，以实现AI系统的可控性和可信度。

结论

总之，DIKWP（数据、信息、知识、智慧、目的）人工意识评估标准重新定义了AI的认知范围。它超越了仅仅关注的范畴，强调了数据、信息、知识、智慧和目的在实现AI系统人工意识方面的关键作用。通过包括这些要素，AI系统可以更好地模拟人类认知，解决未来伦理、可解释性和可靠性方面的挑战。

未来展望

随着AI技术的不断发展，DIKWP框架为开发具有意识的AI系统提供了一个前景广阔的途径。未来的研究和实践需要进一步完善和验证这一框架，以进一步缩小AI与人类认知之间的差距。DIKWP框架代表了一个全新的AI发展方向，一个能够更好地理解和满足人类需求的方向，塑造出一个负责任和智能的机器未来。

本文探讨了DIKWP框架的重要性，它不仅重新定义了AI认知的要素，还提供了一个更为全面和可信的评估标准，有望推动AI技术朝着更为负责任和智能化的方向发展。它强调了数据、信息、知识、智慧和目的在实现AI人工意识方面的关键作用，为AI与人类之间的协作铺平了道路，开创了一个新的机器时代。

Beyond Attention: Attention is NOT all you need

Yucong Duan

DIKWP research group, Hainan University

Email: duanyucong@hotmail.com

Abstract: The rapid advancement of artificial intelligence (AI) has transformed every aspect of our lives. However, as AI technology becomes ubiquitous, it brings forth a series of challenges, including transparency, interpretability, and ethical responsibility. This paper introduces the DIKWP (Data, Information, Knowledge, Wisdom, Purpose) framework for assessing artificial consciousness, which goes beyond the conventional understanding of "attention" and emphasizes the crucial role of data, information, knowledge, wisdom, and purpose in achieving AI system artificial consciousness. We delve into the elements of the DIKWP framework and how it addresses key issues in the current AI landscape, propelling AI technology towards greater responsibility and intelligence.

Introduction

The rapid development of artificial intelligence (AI) has profoundly changed our daily lives, work, and social interactions. From intelligent assistants to self-driving cars, AI systems have become an integral part of modern society. However, the rapid advancement of AI has also brought about a series of formidable challenges, such as ensuring the reliability and responsibility of AI system decisions, explaining the behavior of AI systems, and guaranteeing effective and transparent communication between AI systems and humans. To address these challenges, we need to establish more comprehensive and trustworthy AI assessment standards to ensure the controllability and trustworthiness of AI systems.

DIKWP: Redefining AI Cognition

Redefining Attention as Statistical Focus

Traditionally, the AI field has emphasized the importance of "attention," a concept widely applied for the first time in "Attention is All You Need." However, the DIKWP team proposes a radically different perspective, considering attention as a statistical component of data processing. In the DIKWP framework, data is regarded as the starting point of AI cognition, while attention is merely one statistical way of processing data. This redefinition helps us to comprehend AI's cognitive processes more comprehensively, moving beyond the concept of mere attention.

Information Is More Than Just the Square of Attention

Within the DIKWP framework, information is represented as (attention)², signifying that information encompasses more than just the square of attention. This viewpoint underscores the multi-layered nature of information, as its acquisition demands deeper abstraction and processing of data to align with human intent and needs. Information, within the DIKWP framework, is seen as the AI system's deeper understanding of the world.

Knowledge as Multilayered Comprehension

Knowledge, as a critical element in the DIKWP framework, goes beyond (attention)³. The DIKWP team posits that acquiring knowledge is a crucial step towards achieving AI consciousness. Knowledge's multilayered nature enables AI systems to gain a more comprehensive understanding and explanation of the world, propelling AI systems towards artificial consciousness.

Wisdom Rooted in Human Civilization

Wisdom, within the DIKWP framework, is rooted in the accumulated knowledge of human civilization and culture. Unlike relying solely on attention-driven algorithms, the DIKWP framework underscores that AI systems need to draw from the collective wisdom of human history to make ethical and responsible decisions. This viewpoint ensures that AI system decisions are not only intelligent but also ethical and responsible.

Purpose-Driven AI

In the DIKWP framework, purpose refers to the comprehensive description of human needs and expectations. Purpose-driven AI systems have a clear understanding of their goals when performing tasks, aligning with human intent and objectives. This ensures harmonious cooperation between AI systems and human users, making AI systems more practical and feasible.

Intersection of Description and Execution

The DIKWP framework emphasizes the synergy between description (purpose) and the execution of DIKWP components. This synergy ensures that the behavior of AI systems is interpretable and aligns with user expectations. The DIKWP team argues that relying solely on attention is insufficient; we need to include a comprehensive description of data, information, knowledge, wisdom, and purpose to achieve AI system controllability and trustworthiness.

Conclusion

In summary, the DIKWP (Data, Information, Knowledge, Wisdom, Purpose) framework redefines the scope of AI cognition. It goes beyond the concept of mere attention, highlighting the crucial roles that data, information, knowledge, wisdom, and purpose play in achieving AI system artificial consciousness. By encompassing these elements, AI systems can better emulate human cognition, addressing future challenges in ethics, interpretability, and reliability in the AI landscape.

Future Outlook

As AI technology continues to evolve, the DIKWP framework provides a promising path for developing conscious AI systems. Future research and practice should further refine and validate this framework to bridge the gap between AI and human cognition. The DIKWP framework represents a new direction for AI development, one that enables a more comprehensive understanding of and alignment with human needs, shaping a responsible and intelligent future of machines.

Appendix:

《Attention is all you need》？ NO

DIKWP 团队：《Attention is NOT all you need》

For D : attenion is statistics

Data::= Attention     Yes

For I : attention’s attention is

Information::= (Attention)2     NO

For K :attention’s attention’s attention is

Knowledge::= (Attention)3     NO

For W: Wisdom::=Information originates in human civiliation and culture.

For P : Purpose is description of human requirement and expection.

Purpose ::=<Input(DIKWP ), Output(DIKWP )>

Purpose driven: is the execution of DIKWP

The combination of Description and Execution of  DIKWP  ensures :

the conformance of the Human - Machine cooperation