在香港 Web3 大会上的圆桌发言：Ethereum Meets Hardware

导语 Introduction

在这场关于 Ethereum、硬件与自主系统 的圆桌讨论中，Nijika 从智能影像与 AI 终端产品的视角出发，分享了她对下一代设备形态的判断。她认为，硬件真正的变化，不只是算力更强、模型更大，也不只是设备是否联网，而是设备正在从一次次被用户触发的工具，变成可以持续感知、持续记录、持续理解的系统入口。

A central idea in Nijika’s perspective is that the next wave of intelligent hardware will not be defined only by stronger specs, larger models, or better connectivity. The deeper change is that devices are moving from tools that wait for user-triggered capture to systems that can sense continuously, record continuously, and understand continuously.

这也带来了一个重要的产品判断：在传统影像逻辑里，视频首先是作品素材；而在多模态和长期记忆系统时代，视频越来越像一种视觉记忆库的数据层。设备先记录，模型再理解，应用再把这些内容组织成高光片段、文本纪要、提醒和建议。也正是在这个意义上，硬件与 Web3 的关系开始变得更具体：当设备持续产生数据、调用算力、参与协作时，身份、权限、贡献与价值如何被开放地记录和结算，就成为一个真正的系统问题。

This leads to an important product view. In the old imaging model, video was first treated as creative material. In the age of multimodal AI and long-term memory systems, video is increasingly becoming a data layer for a visual memory library. The device records first, the model understands next, and the application organizes the result into highlight reels, notes, reminders, and recommendations. In that sense, the connection between hardware and Web3 becomes more concrete: when devices continuously generate data, use compute, and participate in collaboration, questions of identity, permission, contribution, and value settlement become real system-level questions.

Q1：你如何介绍自己，以及你所做的技术如何与 Ethereum / Web3 发生交集？

Q1: How would you introduce yourself, and how does your work intersect with Ethereum / Web3?

从我的工作背景来看，我长期关注的是 智能影像和 AI 终端产品。我一直在思考的核心问题是，设备为什么只能在用户按下按钮时才开始工作？为什么不能更自然地进入一种持续感知、持续记录、持续理解的状态？这也是我为什么会特别关注今天这个主题，因为它讨论的其实不只是硬件升级，而是设备角色的变化。

From my product background, my long-term focus has been intelligent imaging and AI-native devices. The question I keep thinking about is this: why should a device only start working when the user presses a button? Why can it not move more naturally into a state of continuous sensing, continuous recording, and continuous understanding? That is why today’s topic is so interesting to me. It is not only about better hardware. It is about a change in the role of the device itself.

在过去，拍照和录像更像一次主动动作。用户先判断“这一刻值不值得记录”，再去按下快门。因此，视频更多时候是作品素材，是为了分享、剪辑和展示。但今天，随着可穿戴设备、多模态 AI 和长期记忆系统的发展，视频正在成为另一种东西。它不只是内容，更是记忆数据。设备先记录，系统再去理解、整理、索引和召回，最后生成日程提醒、会议纪要、自动成片和生活建议。

In the past, taking a photo or recording a video was a deliberate user action. The user decided whether the moment was worth saving and then pressed the shutter. So video was mostly treated as creative material for sharing, editing, and display. Today, with wearables, multimodal AI, and long-term memory systems, video is becoming something else. It is not just content. It is also memory data. The device records first, and the system then understands, organizes, indexes, and recalls that material, eventually turning it into reminders, meeting notes, highlight clips, and life recommendations.

对我来说，硬件与 Ethereum / Web3 的交集也正是从这里开始变得真实。因为当设备不再只是一次性执行功能，而是持续地产生数据、调用算力、参与多设备协作时，我们就必须回答更底层的问题：设备是谁，谁有权限访问，谁贡献了数据或算力，价值又如何被记录和结算。我认为，这正是 Web3 能够提供帮助的地方。它不是外加的一层概念，而是开放系统中的基础协调机制。

That is also where the intersection between hardware and Ethereum / Web3 becomes real for me. Once devices are no longer one-time tools but ongoing participants that generate data, use compute, and collaborate across systems, we have to answer deeper questions: who is the device, who has permission, who contributed data or compute, and how should value be recorded and settled? In my view, this is where Web3 can become useful. It is not just an extra concept. It can serve as a coordination layer for open systems.

Q2：从“设备”走向“自主系统”，区块链整合进硬件的最大机会与最大挑战是什么？

Q2: As we move from “devices” to “autonomous systems,” what is the biggest opportunity and the biggest challenge in integrating blockchain into hardware?

我认为，最大的机会是让原本彼此独立的设备，变成一个可以共享规则、共享身份、共享价值分配机制的协作网络。今天很多设备已经很聪明了，但大多数仍然停留在“单机智能”。它们能感知、能识别、能执行，却还缺少一个原生的协同层。区块链真正值得讨论的地方，不一定是让单个设备更聪明，而是让整个网络在身份、可信记录、权限管理、贡献追踪和激励机制上更容易协作。

I think the biggest opportunity is turning isolated devices into a collaborative network that can share rules, identity, and value mechanisms. Many devices today are already smart, but most of them are still examples of standalone intelligence. They can sense, recognize, and act, but they still lack a native coordination layer. The most valuable role of blockchain may not be making one device smarter. It may be helping the whole network coordinate better around identity, trusted records, permission management, contribution tracking, and incentives.

而且如果我们把“主动记录”这个视角放进去，就会更清楚地看到机会所在。未来很多设备不只是偶尔产出一段内容，而是在持续沉淀对现实世界的感知记录。那这些记录如何被授权访问、如何被可信索引、如何在不同服务之间被调用，都会变成系统层的问题。如果视频、音频和行为流越来越像记忆库的数据，那么它们也需要更清晰的规则来管理边界和价值。

The opportunity becomes even clearer if we add the lens of active recording. In the future, many devices will not just create occasional pieces of content. They will continuously accumulate records of the real world. Once that happens, questions about how those records are authorized, how they are indexed in a trusted way, and how they are used across services become system-level questions. If video, audio, and behavior streams increasingly look like memory-library data, then they also need clearer rules around boundaries and value.

但最大的挑战也非常现实。物理世界是实时的，而链上的系统天然存在延迟、成本和执行边界。无论是机器人动作、汽车感知，还是可穿戴设备反馈，很多决策都必须在本地完成，而且要足够快、足够稳、足够安全。所以我不认为未来是把所有事情都放到链上。更合理的方式是：实时感知、实时控制和大部分推理留在设备端或边缘侧，而链上负责那些真正需要公开规则、可信记录和透明结算的部分。

But the biggest challenge is equally practical. The physical world runs in real time, while on-chain systems come with latency, cost, and execution limits. Whether we are talking about robots, cars, or wearables, many decisions must happen locally, quickly, reliably, and safely. So I do not think the future means putting everything on-chain. A more practical architecture is to keep real-time sensing, real-time control, and most inference on-device or at the edge, while blockchain handles the parts that truly need open rules, trusted records, and transparent settlement.

Q3：展望未来 3 到 5 年，你最希望看到的一个重大突破是什么？

Q3: Looking ahead 3 to 5 years, what is the one major breakthrough you hope to see?

未来三到五年，我最希望看到的是一个真正可信、真正可用的设备协作网络。在这个网络里，硬件不只是被购买和使用一次，而是能够在用户授权下，持续贡献感知、记忆、计算和理解能力，并获得清晰、透明的价值回报。

Looking three to five years ahead, the one breakthrough I most want to see is a truly trusted and truly usable device collaboration network. In that network, hardware would no longer be something that is only bought and used once. With user permission, it could continuously contribute sensing, memory, compute, and understanding, and receive clear and transparent value in return.

我之所以特别在意这件事，是因为我越来越相信，下一代硬件竞争的关键，并不只是参数更强，也不只是模型更大，而是谁能真正把硬件、软件、AI 和服务组织成一个长期运行的系统。这个系统不仅能记录世界，更能帮助用户记住世界。换句话说，我希望看到的突破，不只是设备会执行任务，而是设备能够把现实世界持续沉淀成可理解、可搜索、可调用的视觉记忆库，并在开放网络中参与协作和价值分配。

I care about this because I increasingly believe that the next generation of hardware will not be defined only by better specs or larger models. It will be defined by who can truly organize hardware, software, AI, and services into a system that runs over time. That kind of system does not just record the world. It helps users remember the world. In other words, the breakthrough I hope to see is not only devices that can execute tasks, but devices that can continuously turn the physical world into a visual memory library that is understandable, searchable, and usable, while also participating in collaboration and value sharing across an open network.

如果这一点成立，那么智能影像的意义也会被重新定义。过去我们更关心“能不能拍得更好”，未来我们会越来越关心“能不能记得更完整、理解得更深入、调用得更自然”。而当这种能力与开放规则、可信协作和透明结算结合起来时，设备就真正从工具走向了自主系统。

If that happens, the meaning of intelligent imaging will also be redefined. In the past, we mainly asked whether a device could capture better content. In the future, we will increasingly ask whether it can remember more completely, understand more deeply, and make that memory more usable. Once that capability is combined with open rules, trusted coordination, and transparent settlement, devices will truly move from tools to autonomous systems.

结语 Closing

如果要用一句话总结，我会说：从被动拍照到主动记录，不只是影像产品的一次功能升级，而是设备从内容工具走向记忆系统、再走向开放协作网络的开始。

If I had to summarize it in one line, I would say this: the shift from passive capture to active recording is not just a feature upgrade for imaging products. It is the beginning of devices evolving from content tools to me

mory systems, and then into open collaborative networks.