Thank you for reading this topic. I have a few questions to ask whether ChatGPT can achieve the below functions,
As a software developer for an education company, is there a way to send a list of requests to ChatGPT and ChatGPT replies back with the output? The requests include building questions using inputted keywords, photos and audio?
Does ChatGPT support audio response? For example, the user asks a question by voice and ChatGPT responds by voice?
For math questions, can ChatGPT read a question and study the student’s inputted answer in a photo with solution and correct it?
ChatGPT doesn’t do audio or images. It only understands textual data. While it does demonstrate some visual understanding once the image is translated into ascii art, the capability isn’t very usable.
On its own, chatGPT doesn’t support audio response or input. However some speech to text model (either built-in for OS or dedicated like Whisper) could be used to send inputs and the received output can then be converted to speech with TTS service(s) (provided by OS or as SaaS)
ChatGPT demonstrates problem solving but doesn’t reliably solve mathematical problems. At times it makes mistakes so confidently that unless it is explicitly told in chat that there’s a mistake, it won’t correct on its own. However if it is provided the correct answer behind the scenes, along with explanations, it may be able to explain deviations provided by the user.
Again it doesn’t have vision capabilities, so for the visual to even be interpreted, it will need to be processed by an OCR or computer-vision service and then the result of that service be sent to chatGPT.
Also, currently chatGPT isn’t generally available on the API. There’s a waitlist for the beta.