Skip to content

AI Policy

If you are enrolled in any of my courses, please avoid using generative AI tools like large language models. The rationale is simple:

  • Generative AI tools do not have any semantic understanding of any scientific discipline. At their core, large language models take a bunch of words as inputs and predict what the next word is likely to be. They predict this based on a probability distribution for word sequences that they are trained on. These models can then be harnessed with a chat-like interface to enhance the user experience, but that is pretty much all there is to these models. Note, in particular, that because there is no first-principles understanding in such a next-word prediction model, there is no guarantee that these models will produce correct responses always. For aerospace engineering, the stakes are quite high both in terms of human and economic cost. Thus, there is very little margin for error. What you need to develop as an aerospace engineer is rigorous scientifically-backed engineering intuition that will allow you to push the frontiers of aerospace vehicle design without compromising on safety issues. Thus, using a tool that is guaranteed to hallucinate, even if to a much lesser extent with frontier models, is likely to severely impact your own learning and path to mastery. For this reason, I will not use any AI tools in this course, and I expect you to do the same.

  • You don’t learn anything when you use a generative AI tool. It is very easy to prompt a large language model and get back what looks like a very plausible text filled with equations, code, and numbers. With frontier models, these outputs are likely to be correct for simple enough problems. But what is almost always true is that by using such tools, you are preventing your brain from developing the skills you will need to be a successful aerospace engineer. If you rely on something that pretends to think for you, you will never learn how to think. Neural networks are very confident, even when they are wrong. Until you reach a sufficient mastery of the fundamental principles yourself, it is very hard to understand when the output of a large language model is legitimate and when it is a hallucination. In this course, I want you to focus on your own learning and not short-circuit this process by falling back on generative AI tools.

Frequently Asked Questions

  • Can I use AI for brain-storming ideas/ clarify basic concepts/ look for references/ or similar roles where I use it only to augment my learning?

Yes, that kind of usage where you treat the output of a generative AI model as a mere suggestion that you can then build on is acceptable. You should, however, exercise sufficient restraint in not running with the AI tool as an easy way out of the hard work of actually learning and becoming an expert yourself. An over-reliance on generative AI tools could hinder your own free thought and creativity.

  • Can I use AI to write code?

No, please don’t. Computer languages are more structured and less sophisticated than human languages. An interesting consequence of this is that large language models are particularly good at computer languages. For most undergraduate level courses, frontier AI models can generate working code for most problems you are likely to encounter in your assignments or exams. But using such a generated code indiscriminately can hinder your understanding of the underlying concepts. What you can, however, use AI tools for is getting help on syntax or understanding complex error messages.

  • Can I use AI for writing reports and/or proof-reading them?

No, please don’t. Writing is an essential part of scientific communication. Outsourcing this job to an AI could significantly impact your ability to digest, internalize, and communicate your ideas effectively. I would much rather prefer to see your own thoughts, even if they are imperfect, than a very fluent mimicry of expertise generated by a large language model.

  • If all my peers are using AI, will I be left out if I don’t use it?

Not at all. On the contrary, you will stand out among your peers because of the insights and experience you gain by truly working hard, failing, and learning. Your future employers are also more likely to reward your engineering insights and ability to navigate tough engineering challenges, which you can only acquire if you put in the time and effort yourself.