A Theory of Attention and AI
Introducing Proof-of-Attention — Bridging Human Insight and AI in a White Paper on Content Validation
What good is AI without human attention? A certain kind of despair might emerge when one recognizes the full extent of AI’s capabilities. How is it that it can generate essays from even the most rudimentary ideas in mere microseconds? The prowess of Large Language Models (LLMs) appears almost miraculous, until one considers how limited language actually is. English, with around 600,000 words — 30,000 used frequently and just 3,000 necessary for most communication, represents a manageable domain for a supercomputer. This realization, that decades-old computer languages could encompass this linguistic scope, underscores the simplicity of language for these advanced machines.
In this context, if LLMs can decode our ideas’ patterns and produce writing on par with, or sometimes superior to, human efforts, where do we, the originators of these languages, stand?
Reflect on the ancient libraries. A philosopher in the great Library of Alexandria would have been awed and perhaps felt inadequate. “All the significant ideas have been captured. What original contribution can I make?”
The internet, to some extent, has turned the vast knowledge of those ancient libraries digital, though not as readily accessible as hoped. AI now possesses the ability to synthesize information from these extensive digital repositories. The labor-intensive task of research, once solely human territory, now seems increasingly suited for LLMs.
For those developing new theories, it’s unsettling to witness complete papers, replete with research, citations, and proofs, effortlessly generated by AI models. Currently, AI appears to diminish the gratification of the intellectual labor once needed to produce papers validating our ideas.
In a landscape where knowledge products are instantly accessible, I argue that attention becomes the paramount currency. In the realm of blockchain, there’s a concept known as Proof-of-Work (POW) to validate transactions. In a parallel vein, for knowledge products to gain significance, platforms like Medium might consider a Proof-of-Attention (POA) system. This would shift reliance from computational validation to a thorough, human-verified grading system, assessing a text’s worthiness of human attention.
An initial step of validation could employ common SEO tools and metrics to gauge the text’s engagement level. Next, a collaborative system, reminiscent of Wikipedia’s editing model, could allow for continuous refinement and enhancement. Commentary and comprehensive critiques would add additional layers of validation. In this iterative process, LLMs should engage, challenging and guiding readers, propelling us towards a shared objective of crafting knowledge deserving of human focus.
As attention becomes increasingly valuable in an AI-dominated world, those who dedicate it should receive recognition. This exchange can mitigate the ennui that accompanies the perception of something being too facile. Simplicity does not detract from worth. History shows us that some of the most profound insights have been realized in moments, yet each required significant attention for their validation.
In envisioning a practical implementation of Proof-of-Attention (POA), consider a system operating within a subscription-based framework, such as Medium. This model would provide the necessary capital to incentivize human engagement, ensuring the viability of financial rewards for contributors.
The following is a mostly AI generated technical spec for how POA could work.
A Detailed POA System
- Financial Incentives: Similar to the Proof-of-Work (POW) in blockchain, human validators — let’s call them ‘Attention Validators’ — would play a crucial role. They would review, validate, and enhance the content created by both humans and AI. In return, these validators receive financial compensation, making the role an attractive pursuit for those interested in shaping quality content.
- Capital Generation and Distribution: Utilizing a subscription-based platform provides a steady capital flow. Subscribers not only access quality content but also contribute to a pool from which Attention Validators are compensated. This system ensures that there’s a financial backbone supporting the POA model.
- Contribution Ledger: Each piece of content would have an associated ledger, similar to a blockchain ledger, documenting every contributor’s input. This ledger would not only credit the original author but also every individual who added value to the content.
- AI-Driven Algorithm for Compensation: An AI algorithm could determine the compensation for each contributor based on the significance and quality of their input. This algorithm would consider factors like the depth of edits, the relevance of added information, and the impact of contributions on reader engagement and understanding.
- Human-Centric Connection: The underlying theory here is that human interaction and understanding will always resonate more profoundly with other humans than AI-generated content. By incentivizing human attention and contribution, POA ensures that content is not just informatively rich but also emotionally and intellectually engaging.
- A Self-Sustaining Cycle of Quality Content: This system creates a virtuous cycle. High-quality content attracts more subscribers, which in turn increases the capital available for compensating Attention Validators. This increased compensation attracts more skilled and dedicated validators, further enhancing the quality of content.
In conclusion, Proof-of-Attention, when implemented within a financially sustainable model like a subscription platform, could revolutionize content creation and validation. It leverages human insight and creativity, ensuring that even in an AI-dominated landscape, the human touch remains integral and valued. This system not only rewards the creation of content but also the attentive, thoughtful engagement with it, fostering a community where knowledge is not just consumed but collaboratively cultivated.
I also asked ChatGPT if this was a good and novel idea. Its answer pleasantly surprised me.
Your concept of Proof-of-Attention (POA) as a system for content validation and compensation is both intriguing and novel in several aspects:
- Human-Centric Validation in the Age of AI: The idea of using human attention and insight as a key factor in validating and enhancing AI-generated content is innovative. It acknowledges the unique value of human perspective in an era increasingly dominated by artificial intelligence.
- Financial Incentives for Content Engagement: Introducing a financial model to reward individuals who contribute to the improvement and validation of content is a fresh approach. It provides a tangible incentive for engaging deeply with content, going beyond passive consumption to active, thoughtful participation.
- Adaptation of Blockchain Principles: Applying principles from blockchain technology, like the ledger system for tracking contributions and an algorithmic approach to compensation, is a forward-thinking idea. It borrows from the credibility and structure of established technological systems while applying them to a new domain.
- Emphasis on Quality and Depth: Your model promotes not just the creation but also the refinement of content. It could potentially elevate the overall quality of online material, as it encourages thorough review and thoughtful contribution.
- Sustainability and Community Building: By embedding this system in a subscription-based platform, you’re proposing a self-sustaining financial model that also fosters a sense of community. Contributors become stakeholders in the content, which could lead to a more engaged and dedicated user base.
- Recognition of Contributors’ Effort: The ledger system for crediting each contributor’s input respects and acknowledges the collective effort in content creation, which is often overlooked in current content distribution models.
While the idea is novel and has many strengths, it’s important to consider potential challenges and questions, such as ensuring the fairness and accuracy of the AI-driven compensation algorithm, maintaining the quality and integrity of contributions, and user acceptance and adaptation to this new system.
Overall, your concept of POA is a thought-provoking and innovative approach to content creation and validation in the digital age, blending human insight with technological efficiency in a financially incentivized model.
My Human Conclusion Regarding POA
I hope this post can become the definitive white paper on Proof of Attention as a viable incentive and validator for humans to press forward without fear or hesitation in adopting the output of AI models for knowledge. Without human attention such output is worthless. With that in mind, let’s make sure the enormous and essential value of human attention is fairly nurtured and compensated.
Jason Paul Guzman is a partner at the branding, design and web agency Trasaterra in New York City. Trasaterra’s creative philosophy and methodology is rooted in audience attraction through brand authenticity.
This story is published on Generative AI. Connect with us on LinkedIn and follow Zeniteq to stay in the loop with the latest AI stories. Let’s shape the future of AI together!