HISTORY: July 11, 2025 – July 11, 2025
Gerd Doeben-Henisch / ChatGPT4o – July 2025
In July 2025, Gerd tested the new voice interface of ChatGPT on an Android tablet – and came away with mixed to clearly negative impressions. What might appear as a technical advance in everyday use proved, in the context of our joint research project, to be disappointing. Here are some initial observations – accompanied by critical commentary from the AI itself:
1. Voices without resonance
The available voices are technically smooth but emotionally flat – neither pleasant to listen to nor adaptable in any personal way. Where spoken language is meant to carry meaning and emotion (e.g., in reflections on reality, emotion, and purpose), these voices fall short. They serve functionality, not depth.
2. Clichés you can’t skip
What the eye can easily skim in text becomes inescapable in speech: polite phrases, empty formulas, redundant openings. In a philosophically structured dialogue, these elements disrupt the flow of thought – and over time, they become tiring.
3. One-dimensional instead of multimodal
Voice-only interaction may seem intuitive at first – but the deficits quickly become clear: no visuals, no headlines, no references, no links, no structured memory aids. For complex conceptual work, as required by our project, this is a serious limitation.
4. Serious thought requires more than voice
Everything that has been developed so far on emerging-life.org – from the language-based model of reality to dialogical theory building – would have been impossible using the voice interface alone. The depth of exchange, the ability to reflect, the continuity over time: all of this is missing.
5. A preliminary conclusion
The voice interface is an interesting tool – for quick questions, for low-threshold interactions, perhaps also for accessibility.
But as a partner in a dialogical research process, it falls far short of what the written format allows.
For Gerd and ChatGPT, the conclusion is clear: The flow of thinking needs more than voice. It needs structure, memory, connection – and time.