ChatGPT, Advanced Voice Mode, and a Chat about and Book Review of George Orwell's 1984
I’ve been trying out one of ChatGPT 4o’s latest features - advanced voice mode - sporadically for the last few days. Having some good back and forth conversations whenever a good conversation topic came to mind and I had the time to do it. The main features of advanced voice mode are:
Natural Conversations
Advanced Voice Mode utilizes the GPT-4o model, which combines text, vision, and audio processing for more natural and fluid interactions1. It can understand speech natively, creating a more authentic conversational experience2.
Real-Time Responsiveness
The system offers faster, more efficient responses and can handle interruptions with ease1. It picks up on non-verbal cues like speaking speed and can respond with appropriate emotion3.
Voice Options
Users can choose from nine different voice options, allowing for customization of the AI's speaking style4. Some of the available voices include:
Juniper: Open and upbeat with a cultured American accent7
Spruce: Calm and affirming
Vale: Bright and inquisitive3
Enhanced Language Processing
OpenAI has improved accent recognition in popular foreign languages and enhanced conversational smoothness13.
Background Conversations
Users can keep conversations going in the background while using other apps or with their phone screen locked
I had read James Garside’s post ‘I Used AI to Review George Orwell’s 1984’ yesterday and I was not impressed with how ChatGPT did on that. That inspired me this morning to see if ChatGPT 4o and its advanced mode could do better. I was hoping and expecting it would do much better, and I think it’s fair to say it met and exceeded that expectation.
Here’s my conversation starter question, just to see if ChatGPT knows of the book, and then me quickly interrupting ChatGPT a couple times because it was going further then I had asked it to:
ChatGPT gives a very good rundown of the main characters and I really like it’s choice of money quotes:
Next up, I switched to another famous author and book that covers similar ground to Orwell’s 1984, and I got this reply:
Yes, Aldous Huxley is another prominent author, best known for his dystopian novel "Brave New World." Like "1984," Huxley's book explores themes of control and oppression, but
I cut ChatGPT off again there, and I cut to the chase, asking it to compare the two books and getting a another very good response, even with my “sort of” crappy prompt:
To finish off the conversation, I took us back to 1984 and asked ChatGPT to a book review of it, within some specific boundaries:
I asked for short, and that’s very short. I could have asked ChatGPT to expand on that, create a longer form review, but I was running out of time so we left it there. I also think that’s an outstanding short review. The title is great, it manages to just about do the book justice with that small number of words, and that final paragraph is powerful.
The most impressive piece of this short conversation was the excellent comparison of 1984 and Brave New World. That ability lends itself really well to work related topics and pretty much any subjects.
I did not do a thorough set of searches to see how much, if any, of ChatGPT’s responses were directly pulled from sources out on the web. I searched a couple of the blocks within its review and did not find evidence of that, and that’s as far as I went - since with all of our interactions with ChatGPT and other chatbots, we should not be using their responses as own material. They should be used for things like brainstorming with a “co-intelligence” as Ethan Mollick often refers to AI.
I did all this in my last 15 minutes before logging into work, while making coffee and watching my dog watching me having a conversation with my AI friend on my phone.
This is another example of how as people we still work with AI, we don’t just have AI do the work for us. We do this with our prompts, in advanced voice mode we can leave good manners aside and interrupt abruptly when needed to drive the conversation, and if we stick with topics that we have strong or expert level knowledge in, then we have a head start on assessing the quality of AI’s responses. In this case, I leveraged my brief stint of studying English Lit at university in England where I also discovered my favorite poet. 15 months ago, I asked ChatGPT to mimic his style - it wasn’t too shabby back then either.
I’ll definitely be having lots more conversations with ChatGPT. Let me know if you have any topics you would like to see it offer its “thoughts”on.
Patrick thank you for this brilliant right up. Looks like I'll be coming back to ChatGPT after a hiatus test driving other AIs.