Exploring ChatGPT's New Features: GPTs, Voice, and More

Table of Contents

• Accessing GPT-4 and GPT-3.5 Models
• Creating Custom GPTs
• Testing Pre-Built GPTs
• Using Voice Capabilities
• Additional Settings and Features
• Conclusion and Next Steps

Accessing GPT-4 and GPT-3.5 Models

The upgraded ChatGPT interface provides access to GPT-4, the latest natural language model from Anthropic. To access it, you need to subscribe to ChatGPT Plus for $20 per month. This gives you access to the new Explore tab with pre-built AI assistants, as well as GPT-4 and GPT-3.5 models directly in chat.

There is no longer a model selector. Instead, you can choose between GPT-4 and GPT-3.5 at the top right of the chat screen. GPT-4 has additional capabilities built-in like DALL-E image generation, data analysis, and code interpretation without needing to specify.

The new interface shows loading indicators and progress bars so you can see when a response is being generated. There are also easily accessible buttons for copying, liking, disliking, and regenerating responses.

Upgrading to ChatGPT Plus

To upgrade, click on the Upgrade button in the top right and select Upgrade to Plus. This unlocks additional capabilities for $20/month. The upgraded interface no longer has the model selector. Instead, there is a switch between GPT-4 and GPT-3.5 at the top right. GPT-4 has DALL-E, data analysis, and code interpretation built-in.

New Interface and Capabilities

The new interface shows loading indicators so you can see when a response is being generated. It also has easily accessible buttons for copying, liking, disliking, and regenerating responses. In the Explore tab, there are pre-built AI assistants called GPTs for specific tasks like analyzing data, generating images, and more. These showcase the capabilities of GPT-4.

Creating Custom GPTs

The GPT Builder tool allows anyone to create a custom assistant powered by GPT-4. You give it a name, description, instructions, and more. Then the builder helps you customize capabilities.

As an example, we created Soothing MD, an AI doctor focused on reducing health anxiety. We trained it on medical resources and optimized its responses to be reassuring.

GPT Builder Tool

The GPT Builder provides an easy way to create a custom AI assistant step-by-step. You provide key information like the name, description, instructions, and conversation starters. In the playground, you can test it out and refine the capabilities. Additional knowledge resources can be uploaded to improve responses.

Configuring a Soothing Doctor GPT

We created a custom AI doctor for reducing health anxiety called Soothing MD. We trained it on medical resources like the DSM-5 criteria. The instructions emphasize providing reassurance and avoiding triggers. Test questions help refine the responses to be optimally soothing.

Testing Pre-Built GPTs

The Explore tab contains pre-made GPTs focused on specific use cases like explaining games, analyzing data, generating images, and more. These showcase capabilities of GPT-4.

For example, the DALL-E integration creates images from text prompts. The data analysis GPT parses datasets and answers questions.

Game Time, Data Analysis, and More

Game Time GPT explains board game rules. Data analysis GPT loads datasets and answers questions with automatically generated graphs. These demonstrate how custom instructions plus GPT-4 can create focused AI assistants.

Creative Image Generation

DALL-E integration allows text-to-image generation without needing to specify a particular model. Cosmic Dream Visionary uses DALL-E 2 to create AI-generated psychedelic artwork from text prompts.

Using Voice Capabilities

The mobile app provides a voice interface for natural conversations with ChatGPT. Tap the microphone to start talking.

You can interrupt, pause, continue, and end conversations. Useful for steering longer dialogue.

Conversational Interface

The voice conversation capability enables natural back-and-forth interactions. You can interrupt, pause and continue as needed. This allows you to speak at your own pace for improved control and responsiveness.

Steering Conversations

Being able to interrupt or pause/continue during long responses helps steer the conversation. You can correct ChatGPT or add additional context before it sends an incomplete response.

Additional Settings and Features

Additional personalization settings allow you to link your own verified domain to showcase custom GPTs.

Language and voice selection provide more options for localized and accessible interactions.

Conclusion and Next Steps

ChatGPT is rapidly improving capabilities with the launch of GPT-4 integration and advanced creative tools.

While still an early stage, the voice interface shows promise for natural interactions. We can expect continual enhancements over 2023 and beyond.

FAQ

Q: Do I need to pay to access GPT-4 features?
A: Yes, GPT-4 and associated capabilities like DALL-E image generation are only available to ChatGPT Plus subscribers, which costs $20/month.

Q: Can I build my own custom GPT assistant?
A: Yes, ChatGPT provides an easy-to-use GPT builder tool that allows you to create tailored GPTs for specific use cases with customized data, instructions, and more.

Q: Does the mobile ChatGPT app have voice capabilities?
A: Yes, ChatGPT Plus subscribers can have natural conversations by voice with ChatGPT on iOS and Android devices.