The other day I saw a post here arguing that AI like Gemini doesn’t really promote critical thinking in classrooms because of the surface level answers it gives.
I understand the person’s point of view…
But now ChatGPT is reportedly testing a new feature called “Study Together” that flips that model. Instead of giving you the answer, it asks you the questions. Then you think and provide answers yourself.
I think this is brilliant. Students will have to shift from relying on AI for answers to studying themselves and knowing the answers to certain stuff. And then they can get to think independently.
From what I read, OpenAI hasn’t officially announced when or if the Study Together feature will be available to all users.
But if true, do you think this is the kind of educational AI we actually need?
From what I have seen it still seems to be an option rather than a requirement to use. It also looks more like a step by step tutorial which does work for basic teaching and breaking down problems, which may help in subjects like maths or programming which is clearly defined, but will be less useful in deeper meaning topics like literature which isn’t. I haven’t gotten the chance to test it as it is still in development, however it may be useful as a study tool. I think this is however a possible answer to how it can be used, but until its released we cannot test its effectiveness.
It should be rigorously tested before deployment by third parties to see how effective it is at teaching, because openAI does have an incentive(even as a non profit) to get it’s LLM’s a larger market due to competition from other AI’s(most notably, Google’s Gemimi and High-Flyer’s Deepseek). Cutting corners is possible unless independently verified.
It’s true that breaking down code or math problems step-by-step is way easier for any AI than helping a student unpack themes in a literature. But I also think you should know that the upload file feature is there for a reason. One of these reasons is to make AI useful in topics that require depth like literature.
And even if there is a third-party testing because of OpenAI’s race with Gemini, DeepSeek, and Anthropic, I would still have questions on whether it actually helps students think for themselves or just dress up the existing AI behavior in a new way.
Either way, we will never know for now until we see the feature. So let’s see how it plays out.
Essentially, this is the frontier-model research labs’ version of the popular AI wrapper apps for education minus the hacks, and kinda polishes it with less in-between steps to give you better-than-the-usual-Socratic-thinking and other good methods, or rather from “answer machine” to Socratic study buddy, asking you questions instead of shortcutting it.
If it sticks (and don’t see why not), it’s the kind of approach that AI education needs, polished from a big name, that teaches how to think, not just what to say.
I believe studying by chatting with an AI assistant is not for everyone. Most people I know like to have a focused and logically structured curriculum when they’re studying a topic, so just asking, say, ChatGPT questions and getting answers is not “learning” and should not really promote critical thinking; it’s simply just another tool for searching for answers. It makes life a little bit easier, so that instead of googling something and looking through different websites, you get a fairly accurate answer more quickly.
However, I like the idea of the “Study Together” feature, because it’s certainly a step forward. That said, whether it will be helpful for learning rather than pushing us towards more AI-dependency will only be determined with time.