Open AI released a new AI model last week and it is amazing.
It is called ChatGPT and you can chat with it here https://chat.openai.com/chat
In order to experience the revolution of OpenAI I created this app that is using the OpenAI API davinci-003 and davinci-002. The responses differ between versions. Thunkable
To test it on your own, copy the project and enter your own SECRET API KEY.
What are your thoughts?
Have you tried it?
What questions did you make?
ChatGPT is fascinating! It can also produce surprisingly useful code. It doesn’t handle Thunkable questions all that well in part because it’s a block language and in part because its data source is outdated by a couple years. But there’s great potential there!
Thank you for providing the project link… I was just going to see if I could figure out how to use the API in Thunkable.
Starting with your project, I’ve added a few things. I’m trying to get voice recognition, text-to-speech, and image generation working [now working!]
Project Link:
[no longer available]
Features:
• toggle text vs. speech input
• toggle text-to-speech response
• toggle image response (click thumbnail for larger view)
• tap & hold to reset inputs and outputs
Limitations:
• (Many! )
• Text response length is truncated and sometimes not scrollable
• Image generation is based on response text, not on prompt text. To force it to use prompt text, type or say Respond only with “[prompt]”
• Image generation uses DALL-E rather than more impressive generators (DALL-E-2, Midjourney, etc.)
• Does not auto-listen as Siri et al would
You’re welcome. I added the ability to stop the text-to-speech by clicking the stop button. I’ve updated the link in the post above. The app is just a sandbox at this point but I hope to improve it’s UI soon.
Update: cleaned up the UI, added a switch for text vs. speech input, added checkboxes for speech output and image output. See link above.
I’m sorry, I’ve had to delete projects to make room in my account so I no longer have it. Hopefully the link in the first post still works. If you get stuck on any aspect of it, I may be able to help since I’ve spent a lot of time with the ChatGPT API at this point.