Tag
Linus Ekenstam shares a Silicon Valley veteran David Lieb's impressed reaction to 'subvocal' silent input technology, arguing that silent typing/mumbling is just the beginning and that direct brain activity reading could be next.
论文提出 ProACT，一个性能驱动的自适应 VR 消防训练系统，通过闭环调节火灾蔓延动力学与物体可燃性来动态调整训练难度，实验表明其在客观训练效果和主观体验上均优于基线方法。
Subvocal，一家YC 2026秋季批次的三人初创公司，推出可通过面部肌肉微动作读取无声言语的眼镜，实现超过200词/分钟的静默输入，并已融资130万美元，产品预计2027年一季度发货，预订价199美元。
Subvocal 推出了一款可通过面部雷达感知无声口型、以 200+ WPM 将默读词语转为文字的智能眼镜，实现无键盘、无语音的输入方式，并已获 YC 与 Afore Capital 的 130 万美元融资。
Subvocal 宣布获得 YC 与 Afore Capital 130 万美元融资，其智能眼镜通过非接触式传感器追踪内部言语肌肉，将无声的默读解码为文本，速度超过每分钟 200 词，并开启预售。
Aina Interface 宣布结束15个月的隐身开发，将于11月5日正式发布其人机界面产品，号称将带来类似鼠标/GUI和触摸屏/iOS级别的计算变革。
The author questions why consumer AI products still rely on a plain text box and an unstructured pile of chats, arguing for basic UX features like topic-based chat sorting, automatic splitting of wandering conversations, and easily accessible memory controls.
推文探讨计算时代输入方式的演进（键盘、鼠标、触屏、语音），并提出AI智能体未来可能直接从大脑获取意图输入的设想。
一篇 Hugging Face 与 Data & Society 合作的立场论文，指出当前 AI agent 设计中的人工监督（HITL）不仅因工作量与复杂度难以有效执行，还会导致监督者技能退化，呼吁在设计与组织层面同等重视监督者的认知需求。
A Nature Machine Intelligence piece argues that AI systems used by newsrooms to emotionally optimize content require proper behavioural evaluation to assess their impact on users.
The Lightbulb Computer 是一个将投影仪与计算机视觉结合、采用灯泡外形的计算设备构想与原型系列，旨在以环境投影取代智能眼镜和手机屏幕，实现在真实空间中的空间信息显示与交互。
论文提出 Spatial Canvas 接口与海报生成模型 Compo，让用户通过语义、身份、文本和像素四种绑定方式在空间中直接组合生成意图，取代传统的文本提示方式，并支持直接推理与 agentic 模式，实验显示其组合可控性优于通用图像生成与专用海报生成系统。
九州大学的研究团队在SIGGRAPH Emerging Technologies '26上发表了一种无需佩戴头显的VR座椅系统，利用微镜阵列和空中成像光学技术实现立体视觉，并通过陀螺仪追踪座椅旋转实现视角联动。
LiteEMG-FM is an efficient hybrid CNN-Transformer foundation model pretrained on 16 EMG datasets for robust, zero-calibration gesture recognition. The paper demonstrates real-time deployability on resource-constrained edge devices via a hierarchical wake-up gate and evaluates latency, power, and memory trade-offs across offloading paradigms.
A multi-phase academic-eBay study benchmarks current post-deployment user feedback mechanisms for generative AI, identifies common usability issues such as low discoverability and unclear terminology, and proposes best-practice recommendations along with a prototype feedback-collection tool designed to be both user-friendly and actionable for product teams.
MIT CSAIL, Google, and Northeastern researchers unveiled InstructMesh, an AI-driven design tool that pairs Microsoft's TRELLIS 3D generator with GPT-4 so users can generate, prompt-edit, and repair functional 3D-printable models before fabrication.
A position paper introducing 'Model as a Library' (MaaL), a software architecture that packages small, community-enrolled speech models as versioned on-device dependencies for offline, hallucination-free data collection in low-resource African languages. It proposes keyword spotting to turn closed-vocabulary digital forms into voice-based, on-device data collection for low-literacy populations, and transpiling existing form tools into MaaL schemas.
A CHI 2027 paper analyzing 19,930 ChatGPT conversations from 158 young adults and clinician reviews of five distress examples, finding that ChatGPT often responds with overly dramatic, solution-first reactions during acute distress, and proposing three-stage design guidelines (ask about safety, de-escalate, explore concerns without agreeing).
This paper investigates the equivalence between intuitive and familiar design in human-computer interaction.
提出了 ExpActivator，一个免训练框架，通过在冻结的 GUI 骨干网络的潜在空间中匹配当前屏幕与历史状态，仅激活与当前情境相关的经验，而非盲目追加任务相关历史；在四个 GUI 骨干网络上将轨迹内步骤成功率平均提升 28%，同时仅使用约五分之一的历史 token。