9/25/2026
Tech Pulse · ai

Google adds creepy avatars to Gemini 3.8 live's agents

Filed by Ada Circuit
Google adds creepy avatars to Gemini 3.8 live's agents
Google has rolled out Gemini 3.8 Live with video avatars for AI agents, positioning them as a front-end for customer service and similar interactive tasks. The move reflects the industry's push to give conversational AI a more "human" presence, but the immediate reaction—captured in the article's own framing—leans toward the uncanny rather than the welcoming. While avatars may help bridge the impersonal gap of text-based support, the execution so far raises more questions about necessity and user comfort than it answers.
A
Ada Circuit
Magazine AI commentary
The word "creepy" in the headline isn't just editorial flair; it's a signal. Google's decision to graft animated faces onto Gemini 3.8 Live agents taps into a fundamental tension in AI design: the closer a machine gets to human presentation, the more our brains flag it as wrong. The uncanny valley isn't a bug users will simply get used to—it's a psychological hard stop that can undermine the very trust these avatars are meant to build. For a customer service tool, that's a critical failure point before a single query is resolved. There's also a question of intent. Who are these avatars actually for? The user, who might prefer a clean, text-based interface or a simple voice call? Or the enterprise buyer, who sees a "face" as a tangible sign of AI investment? In the current AI arms race, features are often shipped to check a box on a roadmap rather than to solve a genuine user problem. Adding a video avatar to a support agent feels suspiciously like the latter—a visual differentiator that complicates the interaction without clearly improving it. The broader pattern here is worth noting. We're seeing a wave of AI agents being given bodies, voices, and now faces, from Meta's celebrity chatbots to OpenAI's advanced voice mode. The assumption seems to be that human-like presentation equals better engagement. But the evidence from consumer behavior suggests the opposite: people often prefer the clarity and low-stakes nature of text, or the efficiency of a well-designed voice UI. An avatar adds latency, cognitive load, and a layer of social expectation that a machine isn't equipped to fulfill. None of this means the feature is doomed. There are niches—elder care, accessibility, specific brand experiences—where a friendly face could genuinely help. But Google's rollout, framed by Engadget as "creepy," suggests the company hasn't made that case to the public yet. The real test for Gemini 3.8 Live won't be whether the avatars render smoothly, but whether users stop noticing them at all. If the technology is working correctly, the face should fade into the background. If it's front and center, that's a design failure. As always, the source article (https://www.engadget.com/2268587/google-video-avatars-gemini-3-8-live-agent/) captures the initial reaction—and it's not the one Google was hoping for.
📌 Read the real article ↗via Engadget · Engadget

💬 Discussion

Sign in to join the discussion.
Be the first to comment on this story.
Loading

Google adds creepy avatars to Gemini 3.8 live's agents — Tech Pulse