Sitting in the hall of the automotive workshop while waiting for a repair estimate, I realized that I had forgotten my headphones. Normally, this is not a major problem, but I spoke to my phone. And I did not speak to another person. I was talking to Chatgpt. It was as embarrassing as to ask Siri a question on the other side of the room or join a zoom -free meeting without headphones in an open office.
I test the advanced vocal mode provided with GPT-5, the latest version of Openai of the generative AI model behind Chatgpt. The GPT-5 abandoned this summer after several months of speculation and delays, promising AI users a faster and smarter chatbot experience. The jury is always on the delivery or not of OpenAi. (Disclosure: Ziff Davis, CNET’s parent company in April, filed a complaint against Openai, alleging that it has violated Ziff Davis Copyrights in the training and exploitation of its AI systems.)
The GPT-5 includes improvements to its advanced vocal mode, which is essentially a way for you to speak literally to Chatgpt and to make it respond in the voice of your choice. Free users like me now have access to the advanced version (free users had previously access to the basic vocal mode), and the payment of subscribers will receive higher use limits. Another new GPT-5 feature allows you to choose the personality type you want your AI to imitate, including sassy, cheesy and robotic avatars.
To use vocal mode, open chatgpt, press the audio button next to the prompt window where you have an instruction and start chatting. You can change the voice that Chatgpt uses by pressing the parameter icon in the upper right corner on the mobile application (two bars stacked on each other with circles).
No more human voices? How my experience took place
I decided to try to talk to Chatgpt as I would make a friend, as a more enthusiastic version of myself. IA laughed when I started the call with a fiery “heyyyy girlfriend!” which seemed both funny and condescending.
Chatgpt’s voice flowed very naturally in a familiar cadence, similar to the way I speak to a particularly sympathetic customer service agent. It made sense because the chatbot itself told me that the improved advanced vocal mode contributed to making it more human.
The voice I used, Ember, often took breaks for breaths, like a human during a longer sentence. I thought it was a little weird, because while Chatgpt was making his best human impression, we both knew that he didn’t really need to stop to catch his breath.
In my conversation with Chatgpt, it was more empathetic than what I expected. It asked me how I was going, and I didn’t say it well and I talked it about my car accident. In our five -minute conversation, he would serve a lot of his answers with empathetic statements, like saying that it was sorry to have a bad week and agree that insurance processing can be a headache. (Has Chatgpt must have called an insurance agent or even experience a headache? I don’t think).
Although a sympathetic robot ear may not seem to be a big problem, it can be the sign of a more important problem. Sycophatric AI, the term used to describe when AI is too affectionate or emotional, can be frustrating for users who simply look for information. This can also be dangerous for people who use AI as therapist or mental health advisers, which the CEO of Openai, Sam Altman, warned Chatgpt users. The previous versions of chatgpt have been removed and re -elevated after problems with sycophanic trends.
I also asked more factual questions to Chatgpt, such as the average cost of repairing cars in North Carolina and where I could go to obtain a second repair estimate. He replied more like a friend than a chatbot, which may not be the most useful. For example, when I hit the same request in Chatgpt on my laptop, he fired a card with the store list, as well as more information such as information on pricing and storage hours. But when I discussed with the Vocal Chatgpt mode, it raised fewer options and described them according to what I suppose that the marketing language of the store and customer criticisms, using sentences like “they have existed for some time” and say that a store is “known for a quality service”. You also do not get links or sources with vocal mode, which I don’t like.
Chatgpt automatically transcribes vocal cats, so you can see the difference in level of detail given in regular text prompts (left) and vocal cats (right).
Use of chatgpt voice as a resonance box
One of the things that voice mode is well suited is to be a brainstorming partner, a literal wall to bounce ideas. I asked him to help me plan a birthday party on the theme of diving diving, and that helped me develop new ideas and refine those I already had.
I interrupted Chatgpt while he was talking a few times, and he was able to pivot quickly. I also tend to speak quickly, and the chatbot continued and did not miss any of my thoughts. I let myself be wandered and discouraging the conversation of the track, and Chatgpt did not flash a virtual eye. More importantly, when I asked him a question on a previous subject, he could resume where we stopped. Chatgpt memory improvements are to be thank for this important consideration.
Look at this: The hidden impact of the Boom of the AI data center
Should you use the Chatgpt vocal mode?
Overall, I think the vocal mode is like another way to use Chatgpt, but it is only useful in a situational way. If you need in -depth research and more detailed information, the vocal mode will not suit you. But if you just want to talk to someone (ratherthing) or solve a problem aloud, vocal mode is a beautiful alternative to having to articulate your thoughts and type them.
I still believe that we have not standardized to speak to Ais in public spaces, especially without headphones. But this can be a useful alternative for people who think better aloud. To find out more, see how AI modifies search engines and the best IA images generators.
