Stop Prompting Like a Beginner – Use This 2025 AI Strategy for Academic Results in Minutes

How does AI actually see your prompt? Academics waste hours and hours of time prompting the wrong way because they don't understand how AI interprets what you put in. Now, here's the thing is AI doesn't think like you.

It sees patterns. It sees sort of like connections. It doesn't immediately think, oh, if you say write a thesis, it doesn't think of this.

It thinks of the patterns is created in his database around thesis. sort of like needs to draw on different things to understand really what you mean. You know what a thesis is.

You've written one. You've thought one. You've touched one.

You've seen one. All of these things AI doesn't have. So, what we've got to do is lay the breadcrumbs to make sure that it picks up the right bits of information from your prompt and gives you a really valuable response.

You have to learn how to speak AI's language. And once you learn this, you will never go back to prompting the same way again in an academic context. I promise you.

Let's check it out. The first problem is people aren't specific enough. They say write a paper, write a thesis, write an abstract, but this is the sort of level you need to go into to make sure that it's picking up on the right sort of like clues from your prompt.

So here we've got summarize this paper. It's not exact enough. It's not enough context or information.

So then we need to go to something like this. Summarize this neuroscience paper. So now it's pulling in what it knows about neuroscience, about the the words it should look for, the words it should use, the language, the types of uh content that is around the neuroscience field.

It knows it. So you just need to sort of like lay the breadcrumbs so it can extract it from that part of its knowledge base. And then it says in 200 words highlighting the main hypothesis, methods and conclusions for a graduate level um audience.

That is a much more precise and accurate prompt. And if you start sort of like writing prompts like this, you can't go too wrong. But they get far more impressive and interesting.

Check out this next rule. Another really important part of prompting is giving it a target audience. So you could just ask, explain quantum entanglement.

But to who? You need to give the AI clues as to what sorts of language it should use so it can go away into its databases. Extract the patterns of information that it knows about a certain audience.

Then present that uh response in an appropriate way. So this is a much better one. Explain quantum entanglement to a high school student using analogies and simple language.

So here it knows to a high school student analogies and simple language. is going to get you a far better response and a non frustrating response if you prompt like this. This next trick is really important in academia because it really sort of like dials in the response you're going to get.

The third thing that people forget to do is to give it an output. How do you want the information? Do you want it as a table?

Do you want it as a figure? Do you want it as an image? Do you want it as a bulleted list?

All of these things need to be provided to make sure that you get the appropriate response. Check out this. So you could just say write about the impacts of climate change.

The term write just implies it's going to give you some paragraphs. It doesn't know how long, how much. In fact, it's going to be on the shorter side.

So it doesn't use up so much processing power. But if you give it something much more tangible that you want and more valuable for you, that is a much better prompt like this. Create a table comparing the effects of climate change on agriculture in three countries using peer-reviewed data only.

You can see it's an exact um prompt and also it gives me this create a table. Much better. I want a table.

That's just how I want the first bit of information to be presented to me. Super easy. Give it a format.

The thing about academia is there's so many stakeholders. Oh, look at me using corporate language. You've got students, you've got professors, you've got peer-reviewed editors, you've got all of the different sort of like people that are involved in academia and you need to speak to them or have um an output that's appropriate for every single type of person in academia.

And that is a lot of people. So, you need to give it a role. If you're writing a peer-reviewed paper, for example, you may want to use a prompt like this.

You can say, "You are an academic journal editor. Rewrite this abstract to meet the publishing standards. " and improve clarity.

So you are telling it to act like a journal editor. That is going to be valuable for you because if you are doing this before submission to a journal, it's going to give you the appropriate response and the appropriate sort of changes you need to make for your abstract because otherwise it doesn't know where it's going. Giving it a roll will make sure that it's a much better collaborator for you.

Let's face it, AI can always be a little bit wild. you're never quite sure exactly what it will give back to you, especially if it's a using like a new model or it's updated its sort of like knowledge base. All of these things means that you can end up with some responses that are close to what you wanted, but for some reason have included other bits of data that just aren't useful to you.

So, you need to give it constraints. Check out this prompt here. We've got write a 300word summary of this article focusing on only the methodology.

And then this is the important thing. Avoid background information or future directions. So if you want something very specific, you can actually say to it avoid this, don't include this.

I only want this. Giving constraints is very, very important. It knows about constraints.

Apparently, it struggles with images though when it's like don't include a white elephant. There's always a white elephant in the image. But for text stuff, giving it constraints is really going to work for you because if you're finding that it answers like a little bit crazy, say you don't want that.

Easy peasy. One thing I found when it's trying to do really specific academic tasks is if you just bombard it with all of the information at once, it doesn't know exactly where to start. So, what I like to do is write down the steps that I would go through manually and then feed those into the AI to make sure that it goes through a stepwise process.

Never be scared of breaking down the steps of the prompt if you really sort of just not getting the highlevel response that you need. Check out this. So you've got this.

Summarize the article. Step one. Now revise it for an academic conference abstract 250 words.

Third step simplify the language slightly for a non-speist in the audience. Now that is a step-wise approach. If you trying to do all of this in one go, you'd probably get very different responses.

Giving it a step-by-step process just means that you are leading it through the appropriate uh pathway to get the results that you want. Great. Stay around to the end of the video because at the end is one of my favorite prompt engineering tricks you should know about.

After you've been in academia for a while, you get a sense of the right style and tone to use. Chat GPT and other AI models can write like that, but they need to be prompted. It's not something they just know if you just say like write an article.

You can be much more specific. And I highly recommend you do that. Check out this here.

It says, "Rewrite this paragraph in an academic tone using active voice and hedging statements like it appears or the data suggests. " Those are the ways that we write about our research. If we just sort of like give AI a simple prompt, it will write about it in a fundamentally different way.

So, you need to make sure that you give it the tone that you want. Do you want it academic? Do you want it more creative?

And in fact, stay around to the end of this video because I'm going to show you a little trick that I absolutely love and use all the time because it's really going to help dial in that kind of tone for your writing. Now, I remember when I was first using AI and I couldn't not say please and thank you. The problem is is this doesn't help the AI prompt at all.

Saying please and thank you actually means that it's going to sort of like try to change the tone maybe in a more sort of like uh friendly manner than you want for an academic setting. So ditch the fluff. No more please and thank you.

It feels rude but remember you're an academic so you can be rude. Whatever. Yeah.

See you. No thanks. Huh?

Bye. That one by the way was just like no fluff. Just look at your prompt and like cut out the fluff.

That's what I really wanted to say. All right. This is one of my favorite ways of using any AI and that's giving it examples.

It works so well if you give it examples. For example, for example, examples. If you're writing an abstract, give it three abstracts from papers that are similar to the one you are writing.

If you're writing a literature review, give it the literature review. If you give it stuff, it will work within the kind of constraints, the tone, the style that you give it. So, for example, one of my favorite favorite prompts that I use just all the time is this.

Read this and say read when done. And then I put in a load of stuff. I may put in a table.

I may put in sort of paragraphs of the sort of style of writing I want to reproduce. It's not cheating, but it is sort of like shortcutting all of the prompts we've just talked about where it's like, write like this. This is the style.

Emulate this. And once it says read, I'll say something like using what you just read. Blah blah blah blah blah.

super easy. Match the tone, match the style. Um, I want to create something similar for myself.

All of that really works well. So, just give it examples. Easy peasy.

So, if you're not sure where to start with all of this prompting stuff, you can actually get AI to prompt for you. Check this out. Head over to Anthropic and just type in console and you'll end up here.

And it's this one we're interested in. It's generate a prompt. You click here and then you say, "Okay, describe your task.

" Clearly, you need to buy credits. You describe your task and then uh you will generate a prompt that is perfect for you. It is like um chat GPT playground.

They also have a prompt generator but you can also just go to chat GPT and say I want to do something like this um generate a prompt that would be suitable for an AI tool to achieve this task. Super easy. So you can actually get AI to write the prompts for you.

This next little technique is something I use all the time and I think you should too. Check it out. Super bonus tip.

All right, then. This is something I really like doing. If I'm finding the responses I'm getting for AI just like aren't quite hitting the mark, I'll tune the parameter called temperature.

And you do that just by putting temperature and then a number. Check this out. So, if we head over to chat GPT, at the end of a prompt, you can put temperature colon 0.

It goes from 0 to one. Zero to one. Zero means it will not be very random or creative.

It will be like a student who's following all of the rules. But if you want to inject a little bit of randomness, a little bit of creativity into your response if you're writing for like an audience outside of academia, for example, you can turn up the temperature. You can go to 0.

5, you can go to 0. 2, you can do whatever. Or if you just want it to be super sort of like random and super creative, you can turn the temperature up to one.

So just by adding any of this temperature colon then a number between zero and one, you can really dial in the parameters to make sure you're getting the best response possible. If you like this video, go check out this one where I talk about using chat GPT's deep research for research save hours. Go check it out.