Humanize AI Content Manually: Bypass AI Detectors in Minutes Transcript, AI Summary & Key Points

In this video, I'm going to show you how to manually humanize AI generated content so you can go from this to this. And let me tell you it's got nothing to do with the M dash. This is how you do it. So, to understand how to humanize AI content, we need to know how AI content is actually generated. So, I've got a lovely hat here. Look at this. And I've got six things that AI looks for, and let's pull them out and see what we got to find in our Oh, no, I dropped one.

What we've got to look for as we are reviewing the AI content manually. So, the first one is surface level. Surface level. So, when we're reading stuff, we'll often find that it is just a very generic kind of um basic understanding of a particular research field, and that's very important for science and academia because we need to demonstrate that we know what we're talking about.

So, here I've generated something in ChatGPT, and uh you can see that yeah, it's pretty surface level. There's no numbers in it. There's no kind of like real understanding of the field. It's just kind of like an abstract, I guess. So, here we can go through each individual sentence and look for surface level understanding and try to deepen it using our own understanding, and we'll be doing that in this video.

So, stay tuned. So, the first thing I'm looking for is surface level understanding, and I'm going to put these right in front of me. So, for each sentence, I can look and go, "Okay, which one of these can I change?" All right, then. Let's go for the next one. All right, then. The next one. This one's my favorite, actually. It's these words. How Oh, there we are.

Uh However, in conclusion, therefore, AI loves using these at the beginning of sentences. And uh you know, they do exist in academia and research, but they don't exist in the concentration that AI loves to put them in for different uh paragraphs, for sentences. So, you need to go through and look for these sorts of words at the beginning of sentences or paragraphs and either just get rid of them or change them into something you'd actually say."

So, that's going in front of me because I'm looking for those. All right, what's the next one? All right, the next one is no nuance. No nuance. So, AI loves to be very matter-of-fact about things. It says, "This is it. This is the way it is." And it doesn't say, "Well, maybe there's an edge case." Like, it doesn't really do that. So, as I'm going through, I'm looking to add nuance to my text.

I'm going to put that in front of me. Here's an example here. Um OPV devices are a class of solar cell that use carbon-based organic materials. Now, that can be nuanced a little bit. They are introducing ternary blends that have inorganic materials. So, that could be something that we could put in, which is like mainly carbon-based or um traditionally carbon-based cuz we're not looking at the edge of the field in this opening sentence.

So, that's the sort of thing I'm looking for when I'm talking about no nuance. It is very just like, "Yeah, that's the way it is." But, in academia and in a lot of cases, there are some edge cases that you just need to reference every now and again. Hmm. All right then, what's next? Next is rule of three. This is one of my favorite things. AI loves a rule of three.

So, it goes blah, de blah, and blah. It's because humans love reading three things. It just feels nice. Two doesn't feel complete. So, I'm putting that in front of me. And look, here for example, um let's have a look. Here we are. This is a great example. Lightweight, flexible, and potentially low cost. This is a very AI thing to do. Something, {comma} something, and then a final thing.

And we go, "Hmm, that feels good inside. Thanks, AI." But, we are looking to see if we can break some of these rule of threes as I'm going through because that is a huge tell that you have used AI. All right then, next one. Oh, no. No, okay, this one selected itself. All right, burstiness. So, burstiness is essentially the complexity of the paragraph.

Are all of the sentences the same length? Are all of the beats of a sentence very similar? Like, AI never uses a single-word sentence, for example. I've never seen it. In academia, it tends to use very long sentences, and those sentences tend to be um the same or very similar in length. So, here we can see we've got this sentence, which is this one.

We've got this sentence, which is one, two, uh like two and a half um lines. And then here we've got one, two and a bit lines. So, these two are, you know, I would have to split up this if we're looking for burstiness, like changing the sentence length and structure. Um and then down here we've got one, two. Okay, this one's a little bit longer. This one's one, two, three, four, almost five lines.

So, um yeah, this final sentence we probably do want to sort of like split up, maybe turn it into two sentences. Or we may just want to leave it long, but as we're going to go through this, I'll show you how to do it manually. That is something I'm looking for is burstiness, which is essentially structure and sentence length. So, those are in front of me, and uh there's one more.

Ooh, what what will it be? Okay, this one's too wordy. Too wordy. AI writes at a level far superior to you and me. And that's just it. It uses words that are just never used in a particular field, or it uses words that you would never use. Like, in uh certain research fields, there are conventions in the way that you talk about uh certain things. And you're not trying to impress people with your knowledge of a thesaurus, you just use very simple terms that your field uses.

And if it comes across as too sort of like sophisticated for too long, that is a real indication that AI has written this. Um some people do write incredibly well, but and a lot of times it is just far too complex, far too wordy, far too sophisticated for the level that you're actually writing at. And that's particularly true if you're undergraduate, but as you're sort of like getting to a PhD level, you're not expected to write like a professor.

You've got to write like the field expects you to, which is very accurate, which allows descriptors, which are very common in your field anyway, too. Now I've been too wordy. >> [laughter] >> Folk error. Too wordy, done. All right, then. Now I'm going to show you how to go through this paragraph sentence by sentence, what I'm looking for, what I can change, so so that we can get rid of the AI detection.

All right, then. Here we go. I'm going to put the hat back on. I like that hat. Okay, so I had ChatGPT generate a very simple introduction to a paper or thesis or something. I just said, "Give me an introduction to OPV devices." And this is was what it gave me, and I put it through originality. And originality is great. I'm just using it now cuz I've got a subscription cuz I tested it ages ago, but this could be Turnitin, it could be another sort of like GPT-2, whatever you're using, it doesn't matter.

But here you can see that I've put it in, I've used their biggest turbo model, and it's highlighted all of it as likely AI generated. So, let's go back here and let's work through it sentence by sentence. So, the first sentence is, "Organic OPV devices are a class of solar cells that use carbon-based organic materials to convert sunlight into electricity."

Right, that is true, but let's see if we can add some nuance into that. Like I said, the the the field has moved on, so that um traditionally or um conventionally conventionally used carbon-based organic materials to generate electricity. You know, maybe something like that to Now, we can say here potentially that this is surface level. Generate electricity, what does that mean?

If this is going into a paper, you know, maybe generate electricity isn't important. So, here I'm going to say based on organic materials to um generate photo uh what are they called? Excitons from photons or convert Here we are, convert photons to uh, excitons. Now, that's probably a bit too much for this Um, all right, so here that's probably a bit too much for this, but I'm going to go through and just change it.

Um, we obviously have multiple drafts of this, but yeah, that is something that I'm looking for. Here, unlike conventional silicon solar cells, um, so here um, let's say unlike conventional silicon solar cells, OPVs can be manufactured using lightweight, flexible, and potentially low-cost material Right, so this is already a very long sentence. So, you can see as I'm talking about it, I'm like, Um, and one thing you've got to do is look at the sentence and think about how you would say it in your own words.

The great thing about generating like AI text is that it gives you a structure, it tells you kind of like what sorts of things you should talk about, where the level of detail, um, and that is very important. We want to keep keep that kind of stuff, you know, this it's done a lot of that sort of like broad planning, I guess, for us. But here we can say, "Okay, unlike conventional silicon solar cells."

So, unlike conventional Okay, unlike Maybe they're not Maybe they are conventional, but unlike silicon solar cells, um, OPVs can be manufactured using lightweight, flexible. So, here we probably want to get rid of this rule of three. Manufactured using lightweight uh, and potentially low-cost materials. There we are. I think that's probably enough for that.

So, unlike silicon solar cells, OPVs can be manufactured using lightweight and and low-cost. Well, potentially. Yeah, I mean, it is low-cost materials. Um, in And so, then we can say in a roll- to-roll process. There we are. In a roll-to-roll pro- process such as that used in um, uh, what was it? It was the manufacture of banknotes. Okay, let's go with that.

Now, I wouldn't expect AI necessarily to know to know that. Um and then through printing coating techniques some of these in production. Okay, let's get rid of that. Um because that's what I've said here. Roll-to-roll process. Yes, yes, yes. Roll-to-roll process is that used in the manufacture of banknotes. Um and furthermore, you could say or um you could also say they they lend themselves to uh techniques such as uh screen printing and die slot um uh printing die slot uh manufacturing.

They lend themselves uh the materials. Okay, I never like to say they. So, the materials used during the manufacture uh lend themselves to techniques such as screen printing and die slot manufacturing. Uh Okay, let's go with that. Let's just say, you know, it's not perfect, obviously. Um and then let's have a look. These properties used These unique properties make OPVs for applications where Okay, so maybe not.

Here we want to say these properties or the properties properties of OPV devices. Um so here roll-to-roll processes So, here we need to talk about flexibility before getting here. Um Here we are on flexible substrates. Uh The properties of OPV devices um Okay, so here I said so they can be printed on flexible substrates. Uh the properties of OP's Okay, so um traditional rigid solar panels are for some uses impractical when we are uh manufacturing products such as wearable electronics.

So here we've got wearable electronics, portable power sources, building Okay, so this is a lot of stuff that I probably wouldn't use such as wearable electronics, um flexible screens, building Okay, so here building integrated Yeah, Okay, get rid of that. Um and so here we've got a rule of three. Wearable electronics, flexible screens, and semi-transparent.

So let's go with flexible screens and other uses where um let's say um semi-transparency. Um No, actually, let's just go let's just get rid of the rule of three just for the sake of this. So products such as wearable electronics and flexible devices. And then here we can put brackets, screens, um OLED, and also there was fabrics. So if I wanted to put a reference here, I could OLED and uh fabrics.

So yeah, okay, good. And then next sentence, Although OPV devices generally exhibit lower power conversion efficiencies and shorter optimal operational lifetimes in silicon-based technologies, ongoing advances in material design, device architecture Okay, so this is a massive sentence. So here this is talking about how um you know, I want to generally exhibit lower power conversion if Okay, yeah, so that's true.

So I can just say instead of all though, let's get rid of that because that is this one where it's like however in conclusion blah blah blah. Okay, so here we're just saying OPV devices tip generally Okay, typically I don't say generally typically Oh. Exhibit uh lower power conversion or let's just say efficiencies. I mean like lower lower power conversion efficiency is fine.

And here let's just say, you know, this is where it's got surface level, you know, kind of like understanding. I know that when I was doing this they were about I think 4% um of 4% obviously I needed to reference that as well. Um and this is where it's too wordy. You can see that there's so much more that's said. OPV devices typically exhibit lower power conversion efficiency and shorter um material And this is where like that you know, this is the no nuance.

So they are encapsulated. Um shorter material lifetimes. Uh yeah, okay. I think that's fine. The materials are important. Okay. So OPV devices than silicon based than uh let's say more commercially Let's say commercial technologies. Okay. So let's just split up that sentence there. That is probably where we need to stop with that sentence. And now we've got ongoing advances in material design, device architecture.

Okay, so here we've got one, two, three. Once again, it's a rule of three. So I'm looking at my stuff in front of me and I'm like, okay, here I can change the rule of three. So advance So ongoing, let's say recent and then we can um like you know, uh reference some papers. Recent advances in material design, device architecture, and manufacturing processes.

So, recent advances in material Is it's not I guess it's design, but material Okay. All right, this is about nanomaterials. So, in nano material development, device architecture, and manufacturing processes. Okay, so here let's just go Recent advances in nanomaterial development, device architecture, device architecture. So, that's probably captures all that.

Have led to significant improvements in the in their performance um in the maximum Okay, that's not great, but for the sake of this video, that's absolutely fine. Uh positioning Okay, maximum achieved. Recent advances in nanomaterial development and device architecture have led to significant improvement in the maximum efficiency achieved. Um that OPV technology is an exciting candidate for next-generation renewable energy um applications.

All right, there we are. So, this obviously takes a little bit of time cuz you go through each sentence, but um you do get quicker and quicker. And having these in front of you just make a huge difference. And so, what we're going to do is take this over and we're going to paste it in to new scan, that many words. We're just going to look for that. We're going to go turbo cuz I found that that is the most useful, and we're going to see what kind of score it gives us.

There we are. Likely original, 100% confident. Yes, we did it.