On 13 May 2024, OpenAI announced GPT-4o, where the "o" stands for "omni". It was designed to take in text, audio and images and respond with text, speech or images, with responses fast enough to feel like a conversation. Free users received access to many of its features.
What is new
Earlier voice assistants converted speech to text, processed it and converted the answer back to speech, which added delays and lost detail such as tone. A single model that handles all of it directly can respond more quickly and more naturally.
Useful ways to use it in learning
- Practising a language by talking and receiving gentle correction.
- Accessibility. Reading text aloud, describing images and helping people who find typing difficult.
- Homework help that explains step by step when you show it a problem.
Cautions
- It can still be wrong, even when it sounds sure.
- Privacy. Do not share personal details or school records in a conversation.
- Balance. A friendly voice should never replace conversation with real people.
For teachers
Try the tool yourself before recommending it. Ask it to explain a topic you know well and see where it makes mistakes. That experience is the best preparation for guiding pupils.