5 Things I Think PhD Students Should be Using AI For
Hint: None of them are reading, writing, or thinking.
I regularly trash-talk AI on the internets these days, primarily because of the continued dangerous and undisclosed ads that Anara and its team of ‘influencers’ are flooding TikTok with. Reading, writing, and thinking form the basis of the PhD process, so I have to wonder why anyone who wants to outsource any of that to AI is even doing a PhD in the first place.
But my goal isn’t to eliminate AI from every part of PhD life. There are plenty of things I’d have used it for had it been around when I was a grad student.
Here’s my list:
1. Vent about your boss, other professors, fellow students, and labmates
Seriously. Especially if you’re relatively new. Someone is always going to piss you off. There will also be no shortage of people on campus willing to listen to you bitch about them… and that is the problem. If I could tell every new grad student just one thing, it would be this: Academia is the adult version of high school, and you need to get to know who you can trust through building relationships and making (a lot of) observations before you share things that should only be shared with people you can trust. Your new ‘friend’ who’s been there 4 years already and wants to hear all your complaints might be the lifelong collaborator you still share conference hotel rooms with 40 years later… s/he might also be feeding everything you say right back to the people you’re talking about. That supportive faculty member might want to help you succeed… s/he might be passing notes to the graduate education team indicating that you’re not a good fit for the program. I could go on, but you get the point.
There are many people who love to hear gossip because of what they can do with it, not because of what they can do for you. Some of them might even use you as an (anonymized) example of what not to do in a Substack post 10 years later. You never know.
The primary goal of grad school might be to become a researcher, but besides that, the most important aspect is relationship- and reputation-building. You don’t want to be seen as weak, a problem, difficult to work with, uncooperative, sensitive, or annoying. Whether it is fair or not, venting about people to the wrong people (and there are many) could get you labeled any or all of those things, and it’s hard to lose a reputation like that once you have it.
ChatGPT won’t call your professor and tell them you thought their exam question was ass. It won’t tell your labmates you called the postdoc a bully. What it will do is help you work through how to handle your feelings associated with whoever is pissing you off this week and strategies that will help you get over it (including assessing your own role in the situation, which I know you don’t want to hear, but it has to be said). It will let you vent and give you a little encouragement and hopefully make your day a little better—which is usually all we’re looking for when we vent to a person anyway.
2. Practice difficult conversations
Committee meetings can be daunting if there are people on your committee for whom you often have to use item #1 on this list. The Q&A session after a presentation almost always invites the not-a-question-but-a-comment types that can derail the entire discussion. Conference poster sessions attract the occasional tenured professor nobody can seem to get rid of who’s more interested in what you’re wearing and making sure you’ve had enough wine than what your Western blotting data show (iykyk). These are tricky situations to navigate if you haven’t had much experience with them, and they’re also not things that you want to be gaining substantial experience in.
ChatGPT is relatively good at role-playing conversations. Ask it to act like your nemesis of the moment and work through your responses with it to see what works, find out what doesn’t, and play with different approaches to figure out what comes naturally to you. Having an idea of how you’ll handle things in real time before they happen takes a lot of the anxiety out of the situation when it does happen. You’ll be more prepared and confident and feel that you have control over the situation. It’s harder for people who don’t have the best intentions to gain power over you when you already have an idea of how you’ll respond because you won’t be immediately thrown off by their actions.
3. Organize your chaos
Paper deadlines, 300 data collection time points, seminars you need to attend, writing, keeping your lab notebook up to date… often all in one week. There’s a lot to keep track of, and it’s easy to let things fall through the cracks. AI is a great organizational tool that can save you a lot of time. Give it a list of things you need to do, when they need to be done, how long they’ll take (including any buffer time), etc., and you’ll get a structured plan for getting it all done. There’s no point in making planning another thing to add to your to-do list when AI can do this for you—using it this way isn’t having it do any of the work for you, so you’re not losing out on learning opportunities by having it help with this task.
If you record your meetings (make sure you have permission first!), AI notetakers can give you a meeting summary and to-do list afterward so you can stay present and focused during the meeting without getting caught up in taking notes. If you’re a person who processes and plans through talking out loud, you can talk to your notetaker by yourself and get a summary of what you’ve said so that it won’t all vanish the moment you stop talking (why do our brains do this?!).
4. Work through feedback
Your committee chair said you're unfocused. Your professor told you you're too descriptive in your essays. Reviewer 2 said your manuscript lacks clarity. OK... but what do those things actually mean? Sometimes feedback is vague, sometimes it will contradict everything you've been told before, and sometimes it's just plain demoralizing. Using AI can help you unpack the feedback you receive and interpret it in a way that will help you be objective and avoid spiraling. (Feedback often feels like a personal attack. Remember that it almost always isn't.) Ask ChatGPT to interpret feedback objectively, provide some outlines of what that feedback might really mean, and give you some actionable suggestions that you can use.
This is especially helpful if you're dealing with poor communicators or if you’re struggling with impostor syndrome (most of us do, at some point). This isn't using AI to tell you how to write and think but to help you process feedback in a way that won't block your ability to think clearly.
5. Unpack impostor syndrome, self-doubt, and that feeling that you want to give up
Doing a PhD comes with a lot of emotion, and we're not always naturally good at handling this. There's seemingly endless demand on your time as you're expected to write and do your own experiments while supervising undergrads, teaching, attending seminars and professional development workshops, getting funding... And on top of that, there's the disappointment of failed experiments and maybe a passive-aggressive email or two to respond to. Managing our emotions amid the everything isn't usually considered work, but it's also not a talent that people just have. It's a skill that needs to be developed and worked on. You might be expected to be calm, rational, and productive while juggling 400 invisible expectations, but that doesn't mean you can just be calm, rational, and productive. You have to work on it. And AI can give you a private space to acknowledge that emotional labor. It won’t solve your burnout, but it can help you name it, map it, and figure out what you need to feel a little more confident in yourself again. You can tell it how you feel when, for example, that conference abstract you spent a week on gets rejected, and it'll tell you that it's normal to feel crushed by such rejections. Then, you can ask it to help you brainstorm how to work through that feeling.
I'm not suggesting that you use it as a therapist, but it is a nonjudgmental tool that is always available for a frustrated emotional rant. It's quite good at objectively unpacking that information and giving you practical steps to work through the feelings and turn them into something you can do something with. If you're alone in the lab at 2 AM and your experiment has just failed for the 3rd time, a 10-minute rant with ChatGPT might help you go home feeling ready for the next day instead of like you want to get in the car and just keep driving until you're far away from your pipettes in another state wondering if it's even worth turning around (I know someone who quit a job doing exactly that).
Those are the five things I think AI is good for during a PhD. What would you add?