[QUESTION] Can AI actually teach us at PhD level? What's missing?

I’ve been using ChatGPT 4.0 for hours, talking about many different topics. It’s really interesting because it seems like it’s thinking, analyzing, and reasoning. But I realized something: when I watched a video about OpenAI’s next AI that uses reasoning, I understood that AI needs to break down human thinking into smaller parts and keep adding new abilities to eventually think like a person.

This made me think about an ad I saw claiming that ChatGPT 4.0 could educate someone up to a PhD level. When I asked physicists about it on this forum, they confirmed that, no, it can’t. ChatGPT is a powerful tool for gathering knowledge, but it’s not a true thinking intelligence that can teach you physics or other complex subjects.

So, I’m asking: when can we expect to see AI that has all the abilities to actually think and teach at a PhD level? Anyone here who knows more about the development of AI, what do you think? Can it happen soon?

Welcome to the forum’s AI section

Forum Guidelines


Please keep the following in mind when posting:

  • Posts should be over 100 characters. More detail helps!
  • Share any links to technical or research materials.
  • Let us know your connection to the info you’re sharing. Did you do the research? Did you just find it useful?
  • Provide a description and discussion of the info.
  • If there are repositories, models, or datasets available, include them.

Thanks! Feel free to reach out to the moderators if you have any questions.

I am an automated bot, and this was a standard action. If you have questions, please contact the moderators.

This comment is empty. It looks like something went wrong or was removed. Hopefully, it gets fixed soon!

Brady said:
This comment is empty. It looks like something went wrong or was removed. Hopefully, it gets fixed soon!

I get what you mean. You mention creative thinking… That’s exactly what I was wondering about: when will AI be able to think like humans, with creativity and imagination? No one knows for sure, but if each new AI helps create the next one, it could speed up the progress significantly. It feels like there’s potential for AI to grow faster this way.

Right now we have something like that on platforms like YouTube :joy::joy:

It’s possible already, depending on the subject. I’d say AI can handle a lot of theoretical knowledge right now. In fact, AI was already capable of this kind of teaching years ago, using systems like knowledge bases. It’s kind of like a stack of encyclopedias—very useful if you know what you’re looking for, but without a clear path, it can lead you in the wrong direction.

@Charlie
Good point. What I’m really hoping for is a day when AI can replace physical schools. Imagine AI teaching from kindergarten all the way through to post-graduate levels. If kids still go to school, it might be more for social reasons. I just want to see AI educate comprehensively and at all levels one day.

Why ask physicists about what AI can or can’t do? Wouldn’t AI experts be the right people to answer that? Most physicists probably haven’t even used AI much.

Physicists say no, doctors say no, lawyers say no. There’s a pattern here, right? We’re in 2024, but it feels like we’re in 1924 in some ways. When it comes to things like communication, medicine, and technology, the ‘experts’ are still holding on to the old ways. If you don’t see this, maybe you just haven’t noticed how much things are changing.

The answer to any ‘Can AI do ?’ question is always the same: Eventually.

I totally agree with you that AI’s effect on education is huge, and you’ve described the issue really well. I recently posted on this forum about how relying too much on AI could hurt our ability to think critically and be creative in the long run.

What concerns me most is how common it’s becoming. We’re outsourcing our thinking. If we keep going in this direction, what kind of future are we building for the next generation?

I believe it’s possible now. Let me explain. Most universities already have set plans for teaching. If we use AI to take something like TikTok algorithms and adapt them for learning, we could create personalized, engaging content for each student. If we combine that with a system that adapts to each individual’s learning style, it could make a truly perfect learning experience. It’s not there yet, but it’s already being worked on and will start showing up over time.

@Vale
I agree. Large language models (LLMs) could be a huge part of the future of education. Newer models are better at reducing errors and creating personalized teaching materials for many subjects. I’ve used LLMs myself to teach all kinds of things. While I don’t think you can fully rely on an LLM for something like a PhD, it’s already a very useful tool in learning.

For me, using LLMs helped me break down complex topics in areas like machine learning and coding. I’ve also started learning quantum mechanics and its applications, especially programming. I’m still figuring things out, but it’s been a fascinating process.

LLMs are already being used for academic work too. I’ve used them to draft outlines for a dissertation and to generate prompts based on research papers. It makes me wonder if LLMs could eventually write a full dissertation. It might not be possible yet, but the drafts they create are valuable and can help speed up the process. Who knows? Maybe one day, a dissertation created by an LLM could be defended and published.