OpenAI's product director said users should try the voice mode or combine images to enhance their experience with ChatGPT.
In this week's Hard Fork podcast, Nick Turley stated that the tips are not meant to "speed up like a rocket" when using ChatGPT. However, from the ChatGPT version using the large language model GPT-4o onward, features have been integrated to help users shorten their usage time in a more efficient direction, rather than just inputting text.
Voice mode experiment
The new chatbot feature supports 9 types of human-like voices, developed with the participation of professional voice actors, which can be changed in the settings.
"This is a completely different way to interact with ChatGPT," Turley said. "It has never existed in previous versions, providing an unprecedented experience compared to similar technologies before."
Currently, the free version of ChatGPT supports standard voice use. Meanwhile, the advanced mode with more natural conversations in real-time is available in the Plus and Team paid versions.
Ask ChatGPT to remember
If you don't like repeating information, users can ask ChatGPT to remember some things about themselves to improve OpenAI's AI responses in the future.
"You don't have to wait for it to infer about you," Turley said. "You can just tell it the things you want it to remember."
Turley added that users can "create memories" for ChatGPT by asking it to remember about work, family situations, hobbies, favorite foods... In this way, the AI's "memory" improves over time. If not needed, users can turn off the feature via a command, or even ask the AI to forget them.
Create a custom GPT
According to Turley, users can also consider creating a custom GPT version of their own. "People do this for many reasons," he said. "If you need a prompt to reuse multiple times, create a GPT for yourself."
On ChatGPT, users can adjust for this AI to focus on specific tasks, from learning languages to data analysis based on user instructions through the GPT Builder feature. Additionally, users can also configure GPT with additional settings such as advanced data analysis, adding the Dall-E Image Generation model.
"It's very easy, it only takes a few seconds," Turley said.
Upload files
Turley commented that the file upload feature on ChatGPT is little known but provides a "great experience". Currently available for Plus and Enterprise users, this feature allows users to upload any text files, documents, spreadsheets, and presentations.
ChatGPT then analyzes this text and provides a summary or feedback, extracts information or specific quotes, as well as compares and contrasts two documents. It is particularly useful if users need to go through a long article or data that needs immediate processing in a document of dozens of pages.
"Many people don't know this feature exists," he said. "It's really, really powerful."
Combine images
"I see that there are people who prefer text, and there are also people who prefer images. However, the overlap between the two factors is less common in the same user," Turley said.
An OpenAI representative said he hopes for more "interplay" because ChatGPT supports both. In addition to text, this chatbot can also understand and interpret images.
Currently, OpenAI's image generator Dall-E is also available in ChatGPT, which can create images from conversations. Users can combine them to enhance their experience or create from it, such as making a birthday card. "You might want to use both: a small image alongside a cute poem," he added.