Voice-to-voice AI models promise more natural, low-latency conversations by processing audio directly. However, they are currently impractical for many high-stakes enterprise applications due to a hallucination rate that can be eight times higher than text-based systems.

While Genspark's calling agent can successfully complete a task and provide a transcript, its noticeable audio delays and awkward handling of interruptions highlight a key weakness. Current voice AI struggles with the subtle, real-time cadence of human conversation, which remains a barrier to broader adoption.

As frontier AI models reach a plateau of perceived intelligence, the key differentiator is shifting to user experience. Low-latency, reliable performance is becoming more critical than marginal gains on benchmarks, making speed the next major competitive vector for AI products like ChatGPT.

OpenAI found that significant upgrades to model intelligence, particularly for complex reasoning, did not improve user engagement. Users overwhelmingly prefer faster, simpler answers over more accurate but time-consuming responses, a disconnect that benefited competitors like Google.

Models that generate "chain-of-thought" text before providing an answer are powerful but slow and computationally expensive. For tuned business workflows, the latency from waiting for these extra reasoning tokens is a major, often overlooked, drawback that impacts user experience and increases costs.

The next wave of AI assistants focuses on "interaction" or "bi-directional" models that can process information and respond in real-time, allowing users to interrupt them naturally. Startups like Thinking Machines Lab are competing directly with giants like OpenAI to create a more fluid, human-like conversational experience, moving beyond today's turn-based models.

Companies like OpenAI and Anthropic are intentionally shrinking their flagship models (e.g., GPT-4.0 is smaller than GPT-4). The biggest constraint isn't creating more powerful models, but serving them at a speed users will tolerate. Slow models kill adoption, regardless of their intelligence.

The magic of ChatGPT's voice mode in a car is that it feels like another person in the conversation. Conversely, Meta's AI glasses failed when translating a menu because they acted like a screen reader, ignoring the human context of how people actually read menus. Context is everything for voice.

The key challenge for voice AI is mastering conversational flow—knowing when to speak and when to stay silent—rather than simply improving latency or voice realism. Understanding social cues is the next frontier.