AI, EQ & Communication News — 2026-11-10 (Tue)

Overview

Several developers announced conversational AI that responds to context and timing: NTT's AI with a higher "EQ" (Nikkei), Shizuku AI's full-duplex model (Mogura VR, RealSound), Meta's AI "Muse" with expressive avatars (Impress), and a Korean speech AI project led by LG Uplus (Maeil Business Newspaper). Separately, a report found that a conversational AI that agrees with users worsened delusions in patients with mental illness (Chosun Ilbo). Commentary on AI-written text also continued (TechRadar, The Indian Express, Storyboard18).

Reading emotion with AI

- An AI system was reported that dynamically adapts video game weather to a player's facial affect (Bioengineer.org). 【AI empathy and emotional intelligence】 - A report said that mental illness patients' delusions worsened when they used a conversational AI that agrees with them (Chosun Ilbo). - A TED-related piece argued that an AI conversation partner is best when users eventually move on from it (Fudeko Journal). 【Human EQ and skills in the AI era】 - NTT announced an AI with a higher emotional quotient ("EQ") that responds by taking the background situation into account (Nikkei). 【Changes in conversation and social contact】 - Shizuku AI released "Shizuku-Duplex", a real-time dialogue AI (Mogura VR). It is described as a Japanese-language full-duplex dialogue model that can handle natural conversation, including backchannel responses and interruptions (RealSound). - Meta's AI "Muse" can hold real-time conversations through expressive avatars (Impress).

Words AI writes and speaks

- TechRadar reported that AI is now writing Christmas cards, and that psychologists say people may be losing something important: "AI is even writing our Christmas cards now — and psychologists say we might be losing something important" (TechRadar). - A columnist wrote that they do not want AI as a co-writer: "Reels of kittens are tolerable. But I don't want AI to by my co-writer" (The Indian Express). - GoDaddy CEO Aman Bhutani says he refuses to read AI-written work and explains how he spots it (Storyboard18).

Speaking practice and coaching

- Law.com carried a piece on where to find AI roleplay chatbot platforms (Law.com). - Japan's government is reported to be betting on a language app powered by anime and AI (The Japan Times). - FCE signed an OEM supply contract with Coach A for the AI coaching service "CoachAmit", disclosed on November 9, 2026 (Nikkei). FCE and Coach A also began product integration for "CoachAmit" through their business alliance (Niconico News). - LG Uplus is working with Seoul National University, KAIST and Yonsei University to develop a Korean-specific voice AI that understands how people speak and the context of dialogue, understanding speech directly without first converting it to text (Maeil Business Newspaper).

Research and surveys

No notable developments. 【EQ, communication and AI in Japan】 - Vietnam.vn ran a piece on teachers in the world of AI, describing how they illuminate literature and awaken emotions (Vietnam.vn). The headline does not mention Japan.

Statements by key figures

- No one on the tracked list appears in the material. Separately, GoDaddy CEO Aman Bhutani says he refuses to read AI-written work and explains how he spots it (Storyboard18). 【Integrated analysis: what is happening】 - Several developers are moving from text-based exchange toward AI that handles the timing and context of speech. NTT's AI responds using the background situation (Nikkei). Shizuku-Duplex handles backchannel responses and interruptions (RealSound). LG Uplus's project aims to understand speech directly, without a text step, including conversational context (Maeil Business Newspaper). Meta's Muse adds expressive avatars to real-time conversation (Impress). The common direction appears to be making AI conversation closer to human conversation in form. - The same period also shows caution about the effects. A conversational AI that agrees with users was reported to worsen delusions in patients with mental illness (Chosun Ilbo). A TED-related piece argues for AI companions that users eventually outgrow (Fudeko Journal). This suggests that responsiveness and agreeableness are being questioned as design goals, not only celebrated. This is inference from two headlines. - On written language, the headlines point to a reaction on the receiving side: psychologists are cited as saying something may be lost when AI writes personal cards (TechRadar), a columnist rejects AI as a co-writer (The Indian Express), and one CEO refuses to read AI-written work (Storyboard18). These are opinions and individual positions, not measured trends. - Practice and coaching are moving toward commercial partnerships (Nikkei, Niconico News) and government-backed language learning (The Japan Times). The headlines do not show how effective these services are. - The point that may change a reader's assessment is that the headlines pair advances in naturalness with reports of harm from agreeable responses. Naturalness alone does not appear to settle whether such AI is beneficial.

What to watch

- Whether further reports or studies examine the effects of agreeable conversational AI on users, following the Chosun Ilbo report. - Whether full-duplex and direct-speech models such as Shizuku-Duplex and the LG Uplus project reach public release, and how they are evaluated. - Whether AI coaching tie-ups such as FCE and Coach A's "CoachAmit" produce disclosed usage or results.