Transcription
Are you preparing for your dissertation defense? This is a high-stress moment where you are interrogated and you need to defend your thesis. But most people go in blind. You see, preparation is key. And now with AI tools, we don't need to go in with our eyes closed. We can do loads of things which will make our dissertation defense easier on us. So here's five ways that you can use AI for your defense.
The first thing that really freaks people out about their dissertation defense is the fact that there's some crusty old academics there looking at you trying to find holes in your research. But the good thing is they have online profiles. They are often very, very predictable. They will only ask questions in their very sort of small niche of expertise because they also don't want to look silly. So, we can now use AI to research all of the people on your panel so you'll know what sort of questions and expertise they're bringing to the table and then we can prepare answers to the questions they're likely to ask.
So, head over to your favorite sort of like large language model and put in: "I am preparing for my PhD dissertation defense. My committee members include..." and then someone that you know, Dr. Christopher Gibson from Adelaide University. I love Dr. Christopher Gibson. Send him an email on my behalf saying, "Andy Stapleton loves you." All right. Then, based on their area of expertise, generate a list of five, uh, likely questions this panel member might ask about my dissertation. Then I put my dissertation, uh, title there and then make the questions sound natural, critical, and aligned with their academic interest.
So here you can see likely questions from Dr. Christopher Gibson. And this was deep research, by the way. So here you can see that they found Christopher Gibson. They found he's an expert in AFM, and these are the sorts of questions he may ask. And I went through them and I was very, very impressed. If you were to have answers to every sort of question that this sort of like AI prompt brings up for each individual panel member, you are going to be in a much better position. Mark my words.
Remember to stay around to the end because the last thing I'm going to show you is probably the most important thing. But the questions don't stop there. Each individual chapter of your thesis is going to have questions that people are likely going to ask. Now, what you can do is head over to something like Notebook LM. Here we are. I've opened up a new notebook and I put in my full thesis, but this works best if you put chapter by chapter. So, here you can see I've put in thesis chapter number five. And all I did was take my thesis and print out a PDF of the, uh, let's have a look. Print. And then here I just put in pages. And then the pages you want to put in and then, uh, yeah, go from there. So create a little PDF for each individual chapter is what I'm trying to say. Then you can put that in. Make sure it's ticked. Make sure your full thesis isn't ticked and then just ask this very simple question: "Based on chapter five of my thesis, generate two to three thoughtful and challenging questions per chapter, uh, that a committee member might ask during my defense. Questions should aim to probe for depth, clarity, and academic rigor." Those three points are very, very important.
So here we can see it gave us for each section, each subsection, it should really say, of my, uh, thesis. Here we've got F8BT PFB devices, and then you can see it's got a really sort of thoughtful question. It actually sort of like provides background to the question and then asks the question. So, for example, "How confident are you that the devices fabricated are truly comparable to the reference works?" And then it gave me the reference works up here. Then it says, "You attribute the improved performance observed with a second nanoparticle deposition layer to the creation of low resistance. While this intuitively explains the reduction, how might the morphology of the deposited nanoparticle layers, particularly the inner particle interfaces..." Wow, that's pretty cool. And this is the sort of level you need to be prepared for for your PhD thesis dissertation. Go through, do each of these sort of like, um, questions, prepare an answer, just in your mind, write down some notes if you must, but do it for each thesis where you're presenting results, and you'll be extra prepared.
When you're doing your PhD thesis defense, you're probably going to have a PowerPoint. PowerPoint, PowerPoint, we love PowerPoint. We love PowerPoint so much, but we sometimes love it so much that we confuse our audience with all of the unnecessary information. But now, using AI, we can make sure that we have all of the important information. It's not confusing, and we are putting our best foot forward in terms of the visuals that we are presenting to our panel and the people in the room.
So, in ChatGPT, I put in a recent presentation that I did and then I said, "Please review it for clarity, logical flow, jargon overload, and visual pacing. Highlight any slides that are too dense, where transitions or signposts are weak, and suggestions to better highlight my contribution. Return your feedback in bullet points, grouped by slide number or section." And you can see it's gone through and it's looked at stuff. And here we can see the results. So here we've got two repetitive suggestion: "Consolidate these into one comparison table." And then we've got here, "Structured, but if you are twice to force..." blah blah blah. "Great for humor, but consider if it distracts from the takeaway." Always a little joke in that presentation. It doesn't like jokes, apparently. Naughty me. Bam. Oh, naughty boy.
So here we've also got this semantic tool. So you can see that it's very thin content. "Add what makes it good versus other." So here you can just go through each individual slide. It uses its visual capabilities. This is ChatGPT 4.0, and you can see that this is a great way of just sense-checking what you are putting forward in your presentation. And I absolutely love that. It will give you confidence when you go into your room knowing that you've put it through a process like this. And if you're saying to the screen, "Andy, but what about the data? You can't put the data into ChatGPT." Yeah, I get it. You need to check your university's policy. And a lot of universities right now have sandboxed versions of ChatGPT which allows you to put in data. Just check for your university. It's going to become more and more common. So just do that.
If you like that one, check out this next section because I use it all the time. Now, now I don't know about you, but whenever I present data, especially research data, there's always that little voice in the back of my head that's like, "Oh, Andy, Andy, this is out of date now. There's probably somewhere who's published something much better." You can now check that super easy with deep research. Here it is. I've got GenSpark this time. GenSpark. I, I'm really liking it at the moment. But what I've done is I've put in the main claims of my thesis, like the conclusion section, and then said, "Just go away, do some research, and make sure if any of this is being challenged at the moment, at least knowing that it's being challenged when you go to present your thesis." Because this is the thing: from the moment of submission to the moment of defending, it can be many months, and in the science world, that is a very long time.
So here are the main claims of my dissertation, and I put in the just conclusion chapter. Pasted it in. Loads and loads of text. Loads and loads of text. Blah blah blah. Oh, that was a lot of text. And then down here, this is the most important bit: "Search the most recent academic literature in the past two to three years to identify any challenges, counter-evidence, or debates around these claims. Summarize each one briefly." You get it. You can read. All right, then. And then it went away. It did its amazing stuff. And these were the things that it came up with. Now, remember, my thesis was submitted in August 2010, I think it was, or '11. I forget. Oh, I'm so old. I'm starting to forget stuff. But that's okay because we're friends. Um, and you can see here that, uh, ultimately this is what it came up with. And I would expect lots and lots of new data that would be not aligned with what my thesis says today because it's been so long.
And so here you can see that this was the supporting evidence, mini-emotion process, and then it says, "Competing factors. While your dissertation focused on that, there was a 2022 Stuttle by Stuttle..." Come on now, say something properly. "...study by those people demonstrated that all of that stuff can happen." Blah blah blah. "2024 system with low interfacial energy..." You know, it just goes on and on, finding stuff that could actually sort of like, uh, challenge the conclusions of my thesis. Knowing this going into your dissertation is a strong position because you can say to your panel, "Well, since the publication of my thesis, I've noticed that blah blah blah," or "I found that blah blah blah blah blah." It gives them confidence that you know what you're talking about, and you know that science and research isn't just this stationary thing that once found exists forever. It's always evolving. It's always continuing to improve and get better. But knowing what the best challenges are to your work is going to be a great thing for you. And look down here, you can see we've got a little bit of a conclusion. Love it.
When I give a very important presentation about my research, whether or not it's at a symposia, a conference, or defending a dissertation, there's two things that I always worry about, and that is the beginning and the end. I want to start strong and I want to finish strong. There's nothing worse than someone starting their presentation all meek. You need to own the stage. Let the audience feel that confidence oozing through you. And your way you do that is just making sure that you have like a prepared statement that goes for about 30 seconds to a minute, and then you can just start strong, finish strong. It's going to give them confidence. There's one thing I always say, which is, "People remember how you make them feel, not what you say." And this is a great confidence boost for everyone.
So use a prompt like this: "Help me write a strong 60-second opening and closing statement for my PhD defense." And then I put in the conclusions from my thesis. And here: "Confident, authentic, and memorable." And then I posted in blah blah blah. Easy peasy all the way down. And down here it gives me some options. This is in Gemini 2.5. And then focus on why. So I like this. "Good morning, afternoon everyone. For years I've been captivy fairy." Don't like that. Let's go to option two. Um, "Good morning or afternoon. The quest for efficient and sustainable energy solutions is one of the defining challenges of our time." That's a bit broad. Maybe I'd soften it a little bit for me, but that's a better opening than the first one. And then we've got, "Organic photovoltaics offer a compelling avenue, but their widespread adoption necessitates in innovative fabrication strategies." Oo, I'm not clever enough to say that sentence, apparently. But you can see here, just by memorizing a few points from the opening and the closing, that just means you're on autopilot. It's a nice way to start because you feel like you're in control. You have the confidence, and then you can talk about your PhD results in the middle. Strong start, strong finish. That is where it's all about. Create it easily with AI.
If you like this video, go check out this one where I talk about delivering a killer research presentation. It's old school. I've got a nice long beard. Go check it out.