Key Takeaways
- Artificial intelligence tools showed mixed results when tested as public speaking coaches.
- Analyzing audio directly proved difficult for AI like ChatGPT, even with a paid subscription, often failing to process files or provide voice-specific feedback.
- Submitting a written transcript to AI yielded more useful, specific critiques on speaking habits, such as filler words and tone.
- While AI can offer some insights and help with rehearsal, it’s not seen as a substitute for dedicated training or real-world speaking practice.
Many people share a fear of public speaking, often rooted in past negative experiences. This apprehension can make even the thought of presenting a daunting task, triggering a fight-or-flight response.
Hoping for straightforward, unbiased feedback, one writer decided to explore whether artificial intelligence could serve as an effective public speaking coach. The initial choice was ChatGPT, a familiar tool, though any chatbot could potentially be used.
The experiment began with the free version of ChatGPT. The plan was to upload past call recordings, instructing the AI to focus only on the speaker’s voice and emulate a well-known communications coach for feedback on clarity, pace, and confidence.
However, the free version struggled. It claimed it could analyze a significant portion of a long audio file but then hit data limits. Even when a shorter, two-minute clip was provided, the AI couldn’t extract or analyze the audio as hoped, requiring manual editing of the file which was cumbersome.
To see if more power would help, an upgrade to ChatGPT Plus, a paid subscription, was made. Unfortunately, issues persisted. The AI still had trouble with audio files, at one point stating transcription tools were unavailable, which was frustrating after paying for the service.
The approach shifted to uploading a transcript instead. This method finally produced some results. The AI reviewed a portion of the text, offering feedback that resonated as true.
ChatGPT identified strengths like being personal and approachable. It also pointed out weaknesses, such as using too many filler words, a tendency to ramble, and over-apologizing. The AI suggested using more declarative language and even offered to create a voice training routine.
Curious about other options, the writer also tried Yoodli AI, a tool designed for speech coaching. According to the account on CNET, this attempt had a hiccup; the AI critiqued the interviewee in the audio, not the interviewer. Yoodli did offer interesting features like role-playing scenarios for practice.
Ultimately, while AI confirmed some existing assumptions about speaking habits and provided some useful advice from transcripts, it wasn’t a transformative solution. The tools could be handy for rehearsing and getting some immediate feedback, but they don’t appear ready to replace professional training or the invaluable experience gained from real-world public speaking opportunities.
