Hello, Computer (squirrelsquadron.substack.com)

OpenAI’s ChatGPT has rolled out an “Advanced Voice Mode” that blends natural-sounding, interruptible conversation with multimodal inputs (the author’s wife had the model describe what it “saw” through her phone camera), turning the long-promised Star-Trek-like spoken computer into a practical reality. Unlike older voice assistants (Siri/Alexa), which require rigid commands, this system supports flowing dialogue, intonation and interruptions—features that make voice interaction genuinely conversational and immediately useful, especially for accessibility (the newsletter cites a blind art historian benefiting from camera-aware narration). The broader implication for AI/ML and product teams is seismic: even imperfect, conversational multimodal interfaces lower the barrier between users and functionality, accelerating adoption the way Napster did for music. Technically, this foregrounds robust ASR, latency-tolerant turn-taking, multimodal fusion of vision and language, and privacy-aware camera processing. For designers and engineers, the mandate is clear—prepare for voice-first UX expectations, prioritize inclusive multimodal models, and plan infrastructure for real-time, interruptible dialogue. The change won’t wait for perfection; small cracks in the old keyboard-and-mouse paradigm will widen fast.