Transcripts
1. Intro: We are reaching a point where AI can write almost anything, emails, articles, wedding cows, and they're not bad. In fact, sometimes
they are really good. But that's exactly the problem because once AI can mimic us, we stop being so sure
who is speaking. We start second guessing
almost everything. Review honest? Was this speech heart felt? Was this essay from a student
or from their ChatGPT? When that line blurs, our relationship with content
and each other shift. And if we want to
stay around it, we have to learn to notice, not to judge, but
to stay more aware. For some people, spotting AI
content is just a fun game. Like, was this Substack
post written by a person or by their
ChatGPT? And that's fair. It's a little addictive once you start noticing the
patterns, to be honest. But for others, teachers,
marketers, editors, people in HR, talent managers, this isn't just a game
it's kind of a big deal. I know advanced
prompting techniques and the ins and outs of
prompting engineering. So I figure out I use that
knowledge for a good cause. So whether you are TI or
fully obsessed with it, I want to show you how to
tell the difference between AI generated content
and something made created
entirely without it. I will walk you
through a bunch of unusual signs and
little patterns I've picked up over the past
months and actually years of noticing these
patterns and working with AI. The kind of patterns that don't show up in AI detection tools. But once you spot them, you can't unsee them. I will spill the
tea. So let's go.
2. Class Project: Last project. I want this course to sharpen your attention and train your pattern
spotting radar, not so you can go around judging or criticizing
people for using AI, because honestly, we are
all just trying to make our lives a little easier
and life gets tough. But so you know, so you can recognize the signs and patterns,
and you are more aware. For your project,
I'd love you to share a screenshot
of a personal note, just the note where
you've written down a sentence you suspect
was written by AI. Drop it into the class project
section so I can see it. A few guidelines. Don't share screenshots
of websites, articles, or posts because we are
not here to shame anyone. Copy one sentence into your
own notes and upload it. If you'd like, you
can also describe the article 0R post
in your own words and tell me what triggered your suspicion without
quoting it directly. So tell me what made that
little voice in your head go. I think TAGPT helped
write this one. So, that's your class project. And now let's dive in. Let me walk you through all
the patterns I picked up from months or years
of working with AI.
3. Why Should We Care About Knowing What’s AI-Generated Content and What’s Not?: Why should we care about knowing what's AI generated
content and what's not? FGPT recently passed
the Turing test, which basically means the line between what's AI generated and what's purely human made is getting more difficult
to spot now. I've been deep in the space since the very
very beginning and I'm constantly tracking
patterns in what AI generates. Yes, even with the most
advanced prompting techniques. I use AI in my day
to day work too, we integrate AI into the tools, streamline processes
with it. All of that. But personally, after hours, I love reading substack essays and posts where I
know really know, they come straight from someone's heart and
someone's real experiences, not the result of a five line TGPTPmpt or the most advanced
prompting technique. But to be honest, every day, we all consume a lot of AI generated content
without realizing it. Yes, every day. You probably and I can skip the word probably
because I'm sure of it. You consume a lot of AI generated content
without realizing it. That tweet that went viral could have been
written by Claude. The job ad that felt
oddly polished, passively ChatGPT that
heartfelt post about burnout, unfortunately might be a
polished ChatGPT draft. It's not about pointing fingers
because people use tools. That's what they are for.
That's totally fine. But when we stop noticing, we stop questioning and
when we stop questioning, it becomes easier
for fake sincerity, for very fake sincerity to
pass as truth. It matters. Says something, not
just what is said, because context
always adds weight. I believe authorship still carries meaning because
not everything that sounds right or sounds
legit is right and we can't afford to be
passive readers anymore. I think and I hope you will find this course
useful in more ways than just one because it's not about spotting AI
generated content for fun. It's about becoming more
and more aware as a reader. You can tell when a piece of writing has that ChatGPT touch. Maybe that will be
helpful in your job. Like when someone
handed something that really should have
been written by them, not by ChatGPT maybe you are a writer yourself
and you just want to avoid raising any red flags that you work was AI generated. For me, it also got
personal lately because there is this one
substack writer I love. I really adore her. I even
pay for her newsletter. I'm one of just a few
paying subscribers because that newsletter
isn't big yet. I've been a loyal paying
reader for months. One day, I asked her a question about one of
her paid city guides and the reply she sent surprised me a bit because
it was totally AI, like unmistakably written
by ChatGPT be honest, it broke my heart a little. It makes you wonder if sometimes not knowing would
be easier, right? But no, I really do believe it's always
better to know because awareness helps us take
grounded and it helps us see the reality around us more clearly and we
are more observants. By the way, did you hear
that headline recently? Apparently, 10% of the top performing Substack newsletters are written by AI. There is also the
official survey conducted by Substack itself. Substack said that
in a survey of over 2000 subtckers 45%
said they are using AI. I'm showing you the
address to the report, so you can read it later to
dive deeper into the topic. The report clearly shows that many successful substackers use AI also for brainstorming
and research. In the report, you can also analyze interesting tendencies. What age are the
people who use AI for their Substack newsletters and what kind of content
they create with it. Which stinks a bit because soapstock is supposed to be the place for conscious
thoughtful writing, a return to real voices, a place for honest writing. And that's just one reason. It's worth learning how to tell human writing apart from
AI generated content, but there are many more. It messes with trust. Have you ever read something and feel like
it should move you, but it just doesn't. Like the article 0R the
post is trying really hard to sound smart or
smart deep or poetic, but something is just off. As you can clearly
see in this example, the text tries
really hard to sound emotional to awaken
emotions in us. But when you read it, you can quickly realize it, it just feels blend and it relies on generalization and cliche things we
all know too well, and it's only surface level. Even though the text
is quite short, there are already
so many red flocks. Unfortunately, that's
often AI because it mimics tone,
structure, even charm. But as you know, AI
can't feel anything. And when readers sense
that disconnect, they don't just feel
disappointed, I think. They often feel tricked because trust isn't built on
perfect sentences. Sometimes I don't
know about you, but I think it happens
very often for me. I catch myself wondering why a certain article 0R certain
post didn't move me, why it fell flat emotionally, even though it should
have hit home. And I used to blame
myself for becoming numb. I just don't appreciate
the written word anymore. But unfortunately, it's
not me, it's not us. It's just that the
writing didn't land because it wasn't
emotional, honest writing. It was generated.
And that subtle lack of emotional depth, that weird disconnect,
then it makes total sense. And when we know what's AI
generated and what's not, we can make more
intentional choices. We can give our time, attention, and even
money, for example, when it comes to substack newsletters to
creators who aren't churning out 100 posts a
night with a few prompts. We can support the ones who put their heart in creating a blog, newsletter and Instagram profile or who put their heart
into writing essays, for example, who have
something to say, academic integrity,
is TGT, the problem. I really feel for
teachers these days. You as a teacher assign
a reflective essay, and your student hands in
something that they've spent too minute generating
and they the clay, they think, that's it. I've done my homework. And you are thinking, Okay, when you are
reading the essay, either this 19-year-old suddenly developed a PhD
level knowledge of postmodernism they outsource
it to charge the PT. And you're not wrong for wondering I think students
aren't trying to be evil. They are just trying
to survive in a system that rewards
output over process. But still, it's a problem, not because of grades, but because we are
outsourcing the heart messy, important work or forming
our own thoughts. I'm really a little
bit afraid of our brains because when
we outsource everything, our brains become
lazy and sometimes we just forget how to do
the work ourselves. As you know, I'm really into technological development and I work with AI on daily basis, but we can't close our
eyes to the downside. If society uses AI
for everything and it replaces our thinking and
communication skills, could lead to the dumping down of humanity, unfortunately, the newest research always
proves it if we can't express our thoughts and ourselves without
the help of AI. We don't use our
cognitive functions to construct sentences. These brain functions
will atrophy, and the current state of the human brain
already isn't perfect. You know how short our
attention span is. SEO and search rankings. Let's talk about Google
and other search engines. The Internet is
drowning in content that all sounds almost the same. Here's what search engines
rules are starting to say. If you must produce
less content, just to climb the rankings, we are going to notice and you're going to sync
with your content. Whether that's true or
just a gentle threat, I don't know. But
I do know this. The Internet is loud and people crave meaningful
and unique articles, not this noise that
sounds all the same. AI alone, I think it is very often a noise because people are generating articles that sound very similar and they don't
use advanced prompting, so the articles are
very, very simple. Of course, they are simple, but they use this polished
pattern matching language. They are incredibly boring. But because there
is just so much of this blend SEO optimized
content out there, it's getting harder and harder to even find
something different, something that
really stands out, something that actually
feels worth your time. It's a little bit
like searching for a real voice and a sea of recycled praising and
polished nothingness. You scroll and scroll hoping to stumble across
something that just hits the deeper misinformation
and confident nonsense. This is the one that
really makes me sad when I think about it because AI
doesn't lie on purpose. It just doesn't know
the difference. It hallucinates facts
like confidently claiming the Eifel Tower is in
Barcelona from time to time or that vitamin
C cures loneliness. Here is the most scary part. It does it incredibly
convincingly with citations and food notes sometimes and
that smug academic tone. If you're not double checking, it's way easy to walk away believing something
totally false, just because it
sounded polished. Of course, serious
hallucinations don't happen constantly, but they are showing
up more often lately, especially with how over trained some of the newer models are. That's where things get really risky because people
are publishing millions of AI assisted articles that repeat the same unchecked info, simply because they don't have time or energy to fact check, which means a whole lot of misinformation is quietly
spreading across the web. It's about more than
just spotting AI. Because this isn't about
becoming a grammar detective. This course isn't about
becoming a grammar detective. It's not about tracking
every like a red flag, even though I will admit I unfortunately do
that sometimes, it's about asking
better questions. Are we still thinking
for ourselves? Are we still listening
for the sound of a real voice behind the screen and whose work
are we really reading? Are we okay letting
AI imitate us so well that we forget how
to recognize real futs? Because I think
real writing still has so much power over
the AI generated content. Though, yes, I work with
AI and I teach classes, how to use advanced
prompting techniques, but the real writing is
something I truly enjoy after hours and I'm not
ashamed to admit that. I remember reading
an Airbnb review once that almost made me cry, which is also a little bit shameful to
admit in the course. But it almost made me cry not because the
host of the place. I also stated that
place was amazing, or the place had surprisingly
soft towels, though it did. But because the way these guests described
their stay and I couldn't agree more
with them because they didn't just say it
was cozy and clean. They wrote about arriving late at night after a
long travel day and how the host had left them handwritten note with a peppermint tea bag
taped to the corner. This review was very specific. It was very vulnerable
and tender and it just felt like a little window
into someone else's brain. And real memories. And I remember thinking
after reading it. Yes. Finally, this was
definitely written by e person, not because it was perfect,
because it wasn't. It was messy, but it was
imperfect in all the right ways in all the right ways we
crave now in the young era. For example, messy punctuation. You know, I was just like
someone was thinking out loud. There was this warm the
phrasing that you just can't get from AI that
never felt that way. That's the humanness we are looking for now in the AI era. These days, I read
something with perfect rhythm,
clever metaphors, a few too many dashes and unfortunately, I
feel suspicious. It's like my brain plays
a game of spoil the AI, not because I hate AI
because you already know I absolutely don't
use it all the time. But I really think identifying
AI content matters, not because we want to go backwards because I
believe and I truly hope AI will change the world for better and I don't
want to feel scared. I know there are so
many, so many dangers, but I want to feel hopeful, even though I'm really aware of the good size
and the bad size. I really think it helps
a lot in medicine and it will save so
many lives that way. That's the positive side
I want to focus on. But back to identifying written
content generated by AI, I just don't want to lose what's human on the way forward. So we are not here to
bash a eye because I think we are frilled and we are lucky to witness it develop. But something in me, and I think you will agree, something in me still
wants to feel like a person is talking to
me when they write, not a father, not a eye. I really like to know how to
distinguish when it happens, and I really want to
show you how to do.
4. How AI-Written Text Stands Out: How AI written text stands out. Sometimes, I think many times I read something online
and I instantly get that weird how to describe
feeling like biting into a croisst that
looks golden and flaky, but somehow tastes really
bad a second after. You have big expectation and
it looks really perfect. But then you discover it just pretends to be
a perfect crosst. The content isn't bad exactly. It's just blanched too
clean. Too perfect. And when I talk with my
colleagues and with my clients, they say that that's what AI
writing often feels like. You can't always explain it. But somehow sometimes
you got to knows. How exactly do you tell? Let's break it down, but not like AI would with bullet points and perfect logic. But just like curious human
with a little t to spill, the sentences feel and
***** too perfect. Have you ever read
something that sounds fine? Like grammatically. Yes, yes, it checks out, but emotionally. It's as exciting as reading
shampoo instructions. Unfortunately, that's
often AI because it's Polish liner balanced and
predictable in its rhythm. It doesn't manger on
or contradict itself. Let me show you what I mean. A real person might write. I was going to
tell you about how I first spotted AI text, but honestly, there is so much to discuss I don't even
know where to start. Now compare that to NAI version. AI generated content
is typically easy to detect due to its structured
and balanced nature. I mean, that's
technically accurate, but it's also a little bit dead behind the eyes
and very polite. That's because AI loves structure and it loves
balanced sentences. Grammar that follows
the rules so closely, it feels like it's trying
to get extra credit for it. But what AI doesn't do AI
doesn't start a sentence, then abandon it halfway, then back around it because it remembers
something like this. I was going to tell you about
how to spot AI writing, but then my mind jumped back to this one Substack post
I read last week. It was about vulnerability, but halfway through,
I realized, wait. This person doesn't
actually feel anything. They are just pretending. AI doesn't do that. It doesn't jump drugs or have
mid sentence self doubt. It stays on script like a student trying to impress
a strict professor. When it comes to
real human writings, I really think it wonders, it pauses, and then it spirals. Weirdly that's exactly
what makes it so charming. I think it's funny, isn't it? The very things
we were once told not to do in school
starting a sentence with batch or dropping into the middle of a
thought are now signs that something was
written by a real person because that's how we talk
in real life, don't we? The most engaging writing, I believe it mimics
how we speak. For example, when I'm creating
scripts for my courses, I was listening and I was watching my
courses from some time ago when I wasn't feeling so confident in front of camera. I noticed such big
difference between how I was behaving then
and how I'm behaving now, and of course, it made me a
little bit proud of myself. But in the past, I scripted almost everything because I
wanted to feel confident, I wanted to feel more
sure, what I will talk, that I won't lose the sense
of what I want to say. But now I script just the most important
parts, the highlights. So I can the order of the things I
want to tell you about. But I just prefer to listen to people talking the
way they really talk. When I, for example, watch
other teachers courses, I really appreciate
if they talk to me the way they will
talk to a friend, not when they write this script, for example, AI means
artificial intelligence. No, I want to hear it
in your own words, you would describe
it to a friend. I think that's exactly the
same with good writing. We can imagine this author. Telling it all loud and it
wouldn't feel superficial. It wouldn't feel weird because
that's the way they talk. I think that's why this
kind of writing just flows better and that's why
it feels so human. To show you that in practice, to give you good
examples of this, I have prepared a few
side by side conferences of how people tend to write versus how AI would phrase that. I want you to notice
the difference in tone, rhythm, and vibe and tell
me if you notice it. And let me know in the
discussion section how you feel about that
difference and if it's very visible to
you or maybe not yet.
5. The “Ask-Then-Answer” Pattern: The ask and answer pattern. You've definitely seen this one. It goes like, are you
looking to build your brand, wondering how to stand
out in a crowded market? The answer lies in
your authentic self. I feel like it's the
digital equivalent of those self help books that keep asking rhetorical questions to
make you feel engaged, but never actually go
anywhere surprising. Because humans when
we ask questions, we don't always
answer them neatly. Sometimes we ask
questions because we are actually wondering
in that moment. Like do I even have
a personal brand? I mean, I run a
personal block and I wear the same white
shirts for days. Maybe that's a good start. Do you see the difference? One is a polished pitch. The other is a real person
thinking in real time. Another example, want to
improve your productivity, the answer lies in time block. And this is AI's
favorite way to sound useful without actually
offering anything personal. Now let's humanize it because
a real person might say, Look, I've tried every
productivity trick in the book. So helped. Some made me cry into
my Google sheet. But the only thing
that's ever really stuck blocking up to ours. Do
you see the difference? Do you see that one
is strength to teach, and the other one
strength to connect? Below are a few real
life comparisons side by side showing
you how human wrote it. Versus how AI typically leans into this ask
and answer pattern. I really want you to notice how the human version often
sounds more casual, more unfinished
while the AI version goes straight into tidy answers, tidy reasons structured logic, and unfortunately
a very predictable flow because that's
the thing with AI. It doesn't just ask questions. I also answers them
too neatly many times. Take a look. Okay.
6. The Storytelling Gap in AI Writing. No Personal Anecdotes (or Super Weird Fake Ones). : No personal anecdotes or
super weird fake ones. Okay, it isn't a surprise, but AI has no personal life. It doesn't miss its
hometown or get nervous on first dates or remember
the smell of its grandmas, apple pie in October. It doesn't know your
old work stories, the tiny detail of
every heartbreak, or that one joke your uncle tells at
every family gathering. Even if you pasted
all your journals and all your personal
diaries into a prompt, there would still be a gap. Because memory isn't just
data, isn't just raw data, unfortunately, AI doesn't
live through anything, so it doesn't have
anything to relate to. Even when it tries to tell
a story, after a while, it feels off because it just can't relate to
the real emotions. Sometimes it tries, and the
results are really funny, but something feels off. Like I once found myself facing a difficult situation that tested my patience
and resilience. As you can see, it
sounds very vague, very over polished and it sounds like a
high school essay. There is no real setting, no real moment or the human
messiness or any fic details. It could happen to anyone or
no one and that's the worst. There was a time
when I experienced failure and it taught
me an important lesson. I think it feels very generic
and even though yeah, we all have failures and we all try to earn
a lesson from them, but anyway feels emotionally distant and as I said before, very generic because
this sentence already skips the details and jumps right to the
moral like a fable. Back in 2017, I encountered a unique opportunity
that allowed me to grow both personally
and professionally. And this is where AI
loves mixing time stems with basswood grow
personally and professionally. And it unfortunately sounds
like linked in not life. I remember walking
into the room, hot pounding, uncertain of
what would happen next. This one tries to be cinematic but still
feels copy paste. AI often uses this vague suspense
structure to fake depth. But when you look closely
at it, there is no depth. During my college years, I faced various challenges that help shape who I am today. It also sounds really generic, no scene, no specificity,
just a timeline, and that's what AI
very often does and then it tells you a story
that is very generic. A fun fact. One rainy afternoon as I simmed my coffee and
reflected on life, that's a fragment of a
longer thing TGPT generated, and I have to tell you when
AI tries to sound poetic, and often reaches for cliches like rain,
coffee, and reflection. But there is no real thought or surprise beneath it because
a real person might say, the first time I pitched
my startup idea, I was sweating so hard, I almost short
circuit my laptop. I tried to look
confident while silently praying my voice wouldn't
crack when I said synergy. I hate that feeling. Do you see the difference. Stories rooted in
awkward, vulnerable, only specific truths sound
more natural because they are more human stories
that sound like they were pulled from an
airport business magazine. Probably they are AI
generated because humans are walking story
machines. We can't help it. We narrate everything
from the way we still copy on ourselves
before a job interview. That one stranger told
us we look like someone who reads poetry or someone
who reads Harokimurakab. AI doesn't have that. So even when it tries
to fake a story, it sounds very often like this. One time I was nervous
before a presentation, but I use breathing
techniques to stay calm. And that's cool, but
that could be anyone. That's very, very generic. Now imagine this version. I once walked into
a meeting with a piece of spinach stuck
between my front teeth and pitched an idea about mindfulness while feeling
anything but mindful. So yeah, breathing
helps, I guess, but flossing first
would help me more. And what's your verdict? I think we agree
that's much more human because it's weird, it's a little bit embarrassing
and it's actually lived. So for me personally, the big red flag that something
might be generated is writing without context because I can generate words endlessly, but the text just
doesn't go deeper. It doesn't include
advanced storytelling and relies on
generalizations and cliches. And when you read it, as
you go on, it's very, very surface level
and just locks natural variation as
the text continues.
7. AI’s Obsession with Transition Words: Transition word addiction. AI loves transitions
like moreover, in contrast to
summarize, therefore, which is a little
bit funny or even hilarious because when was
the last time you said, moreover, out loud in
a casual conversation? I think human
writing sounds more like and here is the weed part. So basically,
anyway, where was I? Because we talk like
we are trying to catch a friend up over a coffee, not submit a dissertation
to the Queen. And AI hasn't quite
caught up to that yet. Moreover, for them all in
conclusion, consequently, my brain shuts off the second, I see this in casual writing because we just don't
talk like this. If I texted my friend, food them all, your
outfit is cute. She'd think I was having a
fever because human writing sounds like also your
outfit today girl unreal. You are giving the main
character energy, for example. Even when we write formally, there is usually some
rhythm variation, a pause. Anyway, moment and
AI doesn't do that. Yet. It's like it has been trained on the most polite part of the world writing or the most polite
part of Wikipedia, even when we prompted not to use that kind of writing style. It's always more polite
than it should be and more polite than we are in
real life, than most of us.
8. The Confidence of AI. ChatGPT Doesn’t Hesitate: I doesn't hesitate. When I'm writing
something honest, I almost always pass halfway through a sentence
and rethink the word I used. Sometimes I even
talk to myself out loud like a little
weirdo, for example, is chaotic energy
a little bit too much in that sentence
or does it sound okay? AI doesn't do this. Here's a little secret, at least it's my
personal opinion. Good writing is often
filled with uncertainty, not in the message,
but in the voice. You can feel when someone
is thinking on the page. They kind quiet, very
internal tug of war. A guy skips that part. It always sounds
so sure of itself, even when it's wrong. We as humans, we hesitate. We say things like, I don't know if this
will make sense, but maybe it's just me, but, hear me out. The little wobbles
on the good stuff. That's what makes it feel like
someone is actually there, figuring it all out with you. And most of the time AI is super confident of
the things it generates. Like, it's the confidence
that is always a little bit frustrating and
it can get on your nerves. Also notice that at this point, AI isn't very good. I yes no responses. It always makes them longer
than the used to be. Like AI wants to be super proactive in army with this extra knowledge
and extra insights. But in real life, we didn't talk or write like this before
because real humans, we don't want to mention so many facts in every
answer possible. I completely agree
with Joshua here. You can easily recognize when someone use HGPT or other
large language model when answering because
of formatting and the fact that the answer
is like a mini essay. People weren't writing
like this before AI. Also, these responses
often follow a header subheader bullet
points formatting style.
9. Repetitive Phrasing : Repetitive phrasing. There are moments
when I'm reading a paragraph and suddenly I realize I'm stuck in a loop because the writing
keeps repeating itself, not in a dramatic
slow build way, but more like someone copy pasted the same sentence
three times and just change a word
or two or told GPT to rephrase it many times. Communication is
key relationships. Without good communication
relationships can struggle. That's why improving
communication is essential to
any relationship. This is being told the
same thing three ways, but never deeper
than surface level. It's also giving
high school essay with a word count to it. The future of work is changing. As work changes, the
workplace must adapt. Adapting to the future of
work is now a top priority. You read this and think, cool, but when will you say
anything more specific? Something a little bit more
original. Here's the thing. We humans, we usually
know this when we are starting to
repeat ourselves. We get bored reading
our own work aloud. We stop and think, Wait, didn't I just say that? And we will find a better way to say it or at least
break the rhythm a bit. I doesn't have
that inner editor. It doesn't feel that when it repeats the
same phrase again, so it doesn't fix it. So what to look for? The same phrase, appearing
twice in one paragraph. A sentence that is just
a reworded version of the one above it or the one
in the previous paragraph, no emotional or
narrative progression. When you read it, it's
like walking in a circle. When I spot it, I
immediately pause and think, this doesn't sound like
someone who had a thought. This sounds like something
feeling space and I hate this feeling and I hate thinking
that, to be honest.
10. Unnatural Emotional Tone: Natural emotional tone. This one is tricky
because at first glance, it might sound inspiring. You know, those lines
that feel like they belong on a tech
companies about us page. They sound so grand. Like those old Miss
Universe speeches packed with big fancy words, no one really uses
in everyday life or one you last saw in a
textbook back in college. Of lofty phrases and word
pairing that feel vge. It all sounds
polish and elegant. So. But hey, what was the
writer actually trying to say? In the depths of uncertainty, we find the courage to unlock innovations
through potential. Yeah, it feels
like it belongs in a movie trailer about
corporate brainstorming. But most people don't talk about product death this dramatically. I was deeply moved by the
transformative power of spreadsheets and the clarity they bring to daily operations. Okay, call down AI. You talk about Excel. It's super helpful, but not exactly so
spiritual awakening. Navigating the labyran
of modern existence, I discovered that productivity isn't just a goal,
it's a way of being. It sounds like a
philosophy major, having a quarter life
crisis on notion templates. Through the darkest
moments of my journey, I realized that optimizing my task list was the
key to personal growth. Oh, the ramatic buildup to do a slightly disappropriate
emotional weight. Don't you think?
Artificial intelligence doesn't just assist. It elevates, inspires and redefines what it
means to be human. Intense energy for a
sentence that might just mean AI helps us write faster. The future of creativity
is being revolutionized by artificial intelligence in ways never imagined, never
before imagined. Now you get the point. Sometimes AI sounds
like it's trying to hype me up for something
I didn't ask for. Other times, it swings in the opposite direction and
tries to fake vulnerability. But anyway, ends
up sounding off. It's not that humans always write with the perfect
emotional tone. Of course, we don't
and not all of us are talented copywriters or
anyhow skilled at writing. But we tend to sense
when something is too dramatic or too
flat and we adjust. For example, I might write. Honestly, I was excited
about AI first. Then I read a few pieces
and felt a little weird. The tone was trying really
hard to be emotional, but there was no
heartbeat underneath. Did you notice it to? The difference is that this one feels like someone is
processing the thought in real time and the sentences from AI often feel like someone is trying to win best use of adjectives in some weird
marketing competition. So if a post, even an article 0R any essay feels like it's
either overselling or oversharing without
really saying anything. Chances are, it
wasn't written by someone who has really
felt these words.
11. Look at Sentences’ Length: At sentences length. Do you know that feeling when we are walking on a treadmill, even though you are
technically moving, your brain starts
to fall asleep. I think that's the gym
inspired comparison, but that's how it feels
when I read a block of AI writing with exactly the same
sentence length and rhythm. It many times goes like this. AI is changing the way we live. It affects how we work. It influences the
way we connect. The future is coming fast. It's time to adapt. And as you can easily spot, all the sentences are
around the same length and all the ideas are
equally weighted. After a while of reading
a text like this, your brain just drifts. At least mind us. Most of the time, AI loves to build paragraphs like metronome. Some pace, some tone. Every sentence is
roughly the same length delivered in neat tidy
rows like bricks, which means pretty quickly
the writing starts to lose something and even
when the ideas are solid, something about it just
make your eyes glaze over, not because it's incorrect, but because it just feels flat, like the writing is on autopilot because humans
don't talk like this, neither when we speak
nor when we write. Ba when we are excited
about something, we suddenly got really
intense about something. And then trail off when we are not quite sure
how to end the fight. Here is a rhythm I like in
human writing and talking, short, long, short, for
example, like this. Oh, no, it didn't land. I think it's because we were trying too hard to say something impressive instead of saying something onous and spontaneous. Maybe we can try
again. Do you see it? The spacing, that
variation of length, that what wakes a
reader up again. AI doesn't do that, at least not without advanced
prompting techniques. Not unless you specifically tell it to with this
advanced prompt. So next time you are reading something and it feels stiff, check the sentence length. If every line is marching
to the same bit, there is a good chance
there's AI behind the wheel. Always look for these red flags. Look for uniform sentence length because as
you already know, AI tends to keep sentences
in the medium range, about 15, 25 words. And as you can see
in the example, there's a huge difference
between how we speak naturally, how we write, even though yes, there's an in my
human like variation. Don't mind me. But we humans use sentence length
to control pacing, short for tension,
long for reflection, and AI unfortunately keeps things very steady
and AI also uses the looping cadens
which I personally hate and how to recognize if there is something strange
about the rhythm, read the text loud. Does every sentence take
about the same breath to say? Analyze that. Notice that. The very short or very long sentences
always never appear. It's really so much easier to notice that when you
read the text loud.
12. AI and the Em Dash Habit. Does the Em Dash Give ChatGPT Away?: Dash. Let's talk about the
dash for a second or okay. Maybe for much longer has
dash deserves this time. I've used dashes for as
long as I've been writing. They are my little
verbal side glances, a way to pause, break a rhythm, add a twist of a tone. But now you might
be seeing headlines and reading press calling
them AI giveaways. Wait. Watch. Look, if you're not deep into the world
of punctuations, here's a quick catch up. M dash is that long
horizontal line used to break up a sentence. Usually we a coma or
parent hass could go. It's called N because
it's roughly the width of the letter M as
opposed to the N dash, which is closer in size
to the letter N. And yes, typographers name them and
writers just kept using. And now now AI loves them too. But why AI loves them? It's very, very simple
to explain that. M dashes show up a lot in AI generated content because they also show up
a lot in books, academic writing
and news articles. That's exactly the kind of material large language
models LLMs were trained on. Formal structure and
heavily edited contents and M dashes were them. I AI writes the sources
it learned from and those sources were
packed with dashes. Well, here we are. But here
is the very important twist. It's not the dash
itself that's robotic. It's how AI uses it. In human writing, the dash
often carries personality. It can punch up a
line, create a breath, or offer an ast that feels like it's happening
right inside your head. But in AI writing, it's often very mechanical. Just another transition,
just another placeholder. Just another way to
move from one clause to another without really
committing. The sentence. No, I won't stop using them in my personal writing because I always love them and
maybe you like me. I think there's nothing
wrong with using them now. But I think it is very, very suspicious to
see them in posts of creatures or people who have never used them
before, like wait. Now instantly, out of the blue. They fall in love
with using dashes. Why haven't you before? Because I think we
all know someone who never really had
a way with words. But lately, their
captions, emails, even birthday wishes suddenly
sound weirdly polished. Yeah, suspiciously polished. Dashes everywhere. Perfectly spaced bullet points. Sprinkle of italics for flair. Yeah. Chances on, they discovered
GPT or Cloud or Grock. I don't think we need
to bend the dash L it's some kind of
forbidden artifact. It's not cursed, it's
just overexposed. What bugs me is
this new paranoia. But if I include an MD, someone will instantly assume
I didn't write the thing myself that I pasted something in from
TGPT and hit publish. I completely feel and I
completely agree with the people defending MDAhes
just like I did a moment ago. Writers have always used MDAhes. But this wasn't a
very big group. But some people from
our daily lives, they never used MDhesF example, in Linked in post or
Instagram captions and suddenly they
are using them. To be honest, I never really
noticed them that much. Social media until AI
writing took over my feet and now they are everywhere on LinkedIn and Saptac essays, even in emails from
people who used to end every sentence with
laughing out loud. Suddenly, everyone is
writing like they've been worshipping personal
essays in Paris cafes. That's where the
doubt creeps in. Because even when you are
reading something brilliant, there is an MDS sitting
there mid sentence, you start squinting,
looking for other clues, you start questioning
the whole thing. Even if it passes
the vibe check, something has shifted and the
trust gets a little cloudy, maybe sometimes even
when it shouldn't. It scares me. Oh, yeah, MDs are now everywhere. Everywhere in AI writing, but they are still in
good human writing too. But if someone never
used them before and now they suddenly pop up in every post and every
fourth sentence, it's suspicious because the MDs has traditionally lived in
essays, personal reflections, and the writing of
people who have well, actually read a lot of books
and know how to write, but now they are everywhere. This is case from the future. And I need you to notice that there is no dash button
on most keyboards, which means if you want
to use dash in your text, when you write on your phone, it takes a few extra
seconds at least, and you know how people type on their phones, Busy
coworkers, especially. Their replies are usually
very quick full of typos, missing commas, sometimes
no punctuation at all. But then suddenly this
person start sending long structured paragraphs and they are really
perfectly structured. Every coma in the right place. M dashes, where you'd
never expected them. And you look at it like,
Wow, it's flawless. Really, MDs and comas and punctuation like they are
suddenly writing a novel. That's great for them, of course, but here's the thing. If ChaGPT disappeared tomorrow, they'd be right back to their
old style of responding. Short, rushed, full of mistakes, full of typos, and
definitely no dashes. Really? Really? If ChaGPT
disappeared tomorrow, they'd be right back
to sending Okay, send from my iPhone with free
typos and no punctuation. How do you feel when
you see an dash? Let me know in the
discussion section.
13. The Other Patterns That Caught My Eye: Other pattern that
caught my eye. Yes, the more of them, there is a certain writing that I see on
Substack Instagram. Well, everywhere now that looks very polished
at first glance, it flows, it transitions. It wraps everything up in
neat little paragraphs. But the more I read it, the more I could feel it, the unmistakable voice of a eye trying to sound emotional. It's the predictability, like it was mimicking the structure
of a human thought. And once I saw the patterns, I couldn't and see them. Now I want to share them all
with you more of them so you never look at the
generic Instagram post the way you did before. But let's dive deeper
into the next ones. The voice praises.
You know the ones. Here is the kicker. Here is
the truth. Spoiler alert. They show up everywhere. And yes, humans use them too. I have to tell you, I love to say spoiler alert in my courses, and I think I've been using this phrase more
often than I should. But unfortunately, AI also loves to use
it, but not like this. EI seems to sprinkle those
phrases in confetti, hoping they will make the
writing feel more casual. But to me, it reads like someone doing a bad impression
of a newsletter writer. Dramatic line break trick. This little structure shows up everywhere and AI tries
to sound smart and human. Have a look at these
examples to identify it. The see it now, short wood, question mark, then pause. Why does AI leans on
the structure so much? Because it's statistically
rewarded in its training data. Because as you may already know, AI is trained on a diet
of academic writing. Fd leader posts, trying to be profound sometimes in
free lines or less. Copywriting formulas that
split ideas for emphasis, block intros that
feel like they are pitching a tech to
nobody asked for. So it learns. Oh, wow, when humans want to sound deep. They break lines
dramatically, and yi. Mimic that cadence again
and again and again until every other paragraph
sounds like this. Turnout? It's not the badge
of honor you think it is. Influence is what happens
when you values get loud, discomfort, that is
growth in disguise. Now to be fair,
people absolutely might sometimes use this
structure naturally. I do it too. But when I do, I think there's a
different tone behind it. Usually more self
aware, less preachy. The AI version often reads
like it's trying to preach. The human version is
more a confession. For example, fear. Yeah, I know it well. Regret? Yes, I
carried it for years. Joy, I didn't
recognize it at first. Do you see how somehow it
feels softer and more honest? Log of text and
symmetrical paragraphs. I have to admit I often think that AIRI like someone who took one creative writing class and then immediately
preferred Excel. Everything is organized
into tiny little blocks. It goes like this. Fought
transition, next fought repeat. Here is what I Programmers break big problems into smaller
functions and loop. I think AI does something very similar when
it generates text. Let me explain to
you what I mean. After reading millions of
AI generated paragraphs, I started to realize it's
writing and blueprints, like someone taught it
to organize thoughts the way a software
engineer organizes code. It's not surprisingly
really because AI was also trained on
tech heavy dataset, shaped by clarity
ups knowledge basis. The model naturally gravitates toward patterns that
work and by work, I mean make predictable sense, which is helpful for
answering math problems. Less so for writing with
soul or creative writing. And here is where
it gets strange. I writes and what I can only describe as function
shaped blocks. A clear input, a dramatic pause, and then an output that
pretends to surprise you. Here is the real problem and
what you can do about it. The truth not what you think. You thought it was this. Turns out it's that. After you notice that pattern, you start seeing it everywhere. Because AI is simply behaving like a magician.
With one trick. With one trick, you're
changing the deck of cards. Then there is this most
frustrating thing I noticed in so many
AI generated essays. Looping, as I've
already told you, this habit of repeating the
same sentence shape with just enough variation to sound like progress,
but it's not progress. It's more like recycling. For example, AI helps marketers, AI helps writers,
AI helps teachers. This tool works for emails, works for scripts, works
for presentations. It's not that it sounds bad. It just doesn't build. It feels like a holding pattern when you are landing in
bad weather conditions, it's like circling around the airport but not
quite landing anywhere. There is also this if then
thinking that creeps in, like AI is running
conditionals in a script instead of
expressing a thought. Traditional methods were slow. AI speeds them up. People used to
struggle with content. Now with AI, they don't. If you want efficiency, then you'll want this. It's writing logic disguised
as insight and show. Logic can be beautiful, but only when it's earned. This just feels like a shortcut, dressed up as a revelation. Oh, oh, and I would forget. And there is this AI
metaphor obsession or what I think of as variable assignments
where every concept has to become a cute
little comparison. Think of AI as your copilot. Proms are like spells
for your ideas. You tone and voice, basically the style sheet of your writing. Here
are some more. That are a real examples
of what TGPT generated. And it's not the
metaphors, I mind. I love metaphors myself and I love coming up with
new creative ones. It's the neatness,
I hate, I think, the way AI flatten something complicated
into a punchline, like metaphor only
for a metaphors sake. And as I've already told
you, the MDs inside. AI uses MDs as an escape. When in doubt, AI drops
an MDS and moves on. It doesn't always know
how to transition, so it doesn't is changing, and AI is at the center. These tools are everywhere in work clothes in classroom
and your inbox. It's not just a
trend, it's a future. And I've told you, once you start noticing
these patterns, they are hard to unsee. They show up like a
little watermarks across AI generated writing, which fascinates me honestly because AI writing is an echo. Very often it's very
symmetrical and hollow. I what I'm saying feels like a neat little wrap up
with a reflection, well, maybe that's the point because
maybe I've been reading too much AI writing and too
much AI generated content. And I slowly begin
to sound like AI. Please tell me I don't. To sum up this chapter. You already know AI
has a thing for MDAhes I also have to mention
AI loves italics, especially when it wants
to sound dramatic. It's also obsessed
with lists of and it loves circling back to its main point for some reasons, it keeps shaping paragraphs into the oddly the same lengths like it's following an
invisible ruler. If I had to summarize this section chapter with one big takeaway, I
think it would be this. When writing is too perfect, too vogue or too repetitive, or it starts to feel
like someone is performing language instead
of using it to connect. That's the most important clue. It might be written by AI. Let me end this chapter
with what I think is really useful a
vibe scan checklist. When you are reading
something and trying to tell if something
is AI written, tune in to this. Does it feel too
clean to rehearse? A phrases repeated just
enough to feel suspicious. There many cute and
unusual metaphors, even though the author wasn't using metaphors,
for example, two years ago, are emotions mentioned but never really felt. Is there a strong voice and strong opinions or is the
voice weirdly neutral? Like trying to not
to offend anyone? Do the stories feel real and specific or could they
belong to literally anyone? Maybe, most importantly,
does this sound like something someone would love to share and would
love to write, even with tired eyes and does it have a real story and
real experience to tell?
14. AI’s Favorite Words (and How to Catch Them): Words that make it obvious, your text is written by AI. Okay, here's the part
where I will share something I really
wish wasn't true. There are certain words
that the moment I see them make my trust level drop
to about three out of ten, not because the words
themselves are about. Also use them and have
used them before. They are just not how
most people talk. When they show up too often, especially in a very
casual writing, they feel a little suspicious, like someone asks GPT to make it sound professional and then just copy paste
whatever came out. These words in the realm of
additionally groundbreaking. Shed light on, delves into and unfortunately
also, however. I'm not saying these words
should be banned forever or you shouldn't use
them. No, not at all. But we can and we
should swap them out for something
simpler, something more. And suddenly your sentence will feel ten times more alive. Here's how I like
to think about it. AI now loves to use the
phrase in the realm of, but we can just say in the world of or to be more specific, for example, for the
design industry. Additionally, another word that AI is solely obsessed with. We can try also another example. Honestly, just start
a new sentence. Another favorite
of AI delved into. I can't count how many times I have seen this one delve into. I would swap that for delve into or got curious about much
simpler, much better, much more human another one
AI loves shed light on. But we can just say
highlight or made clear, or even brought into focus. Those are the phrases that
AI isn't using so often, so they are much safer. However, yes, of course,
AI loves however. We can say that set or but
and if you are like me, you can just pause dramatically
and shift your tone. The main idea behind that is if a word feels like it belongs
in a formal press release, but you are writing
an Instagram caption, an email or a personal
post on a blog or a subs. We should swap
these words out and you will sound more like
you and less like AI. Because honestly, the
best way to sound smart is to stop trying
so hard to sound smart. I have to tell you I
have a soft spot for phrases like empower
and transform. Until I see five of them
jumped into the same sentence, and suddenly I feel
like I'm reading a start up mission
statement written by CLGPT who's never
had a real job. And here is a sentence
I came across lately. We leverage innovative
solutions to empower creators and drive
meaningful impact at scale, which sounds very
cool. I got to admit. But what do you actually do. When AI writes, it often strings together
important sounding words. They are usually a little
bit abstract, vogue, and just load in the air without
anything ground in them. Here are some of AI favorite go to words that I
haven't mentioned yet. Innovative game changing. I love this one before. It was training. Impactful. Empower, scale Drive
results, leverage, navigate, enhance expertise,
offerings, valuable, leverage, unraveling,
intricate interplay. When you see three or more of these living in one paragraph, yeah, it's right
to be suspicious because AI loves these words. AI makes a sentence
sound professional, polished and very often
completely hollow. A human, if they're
being honest, would probably write
something more like, we help small businesses
stop wasting money on stuff they don't need and get their
message across clearly. I would be much more real,
much more understandable. And that's the thing.
AI is great at saying nothing but beautifully. Humans, we always say something. We always have some emotions
behind what we say, even if we say it in a
little bit messy way. Yes, there are words
that just scream, this was written by
AI and they are not. Exactly. Of course,
we also use them. They are normal words
and we had them in our personal
dictionaries before. But they are very official. They are very from
academic writing, so AI was trained on those sources and that's why it loves these words so much. Now let's break down some of my personal favorite examples in detail to call out
transformative. Here is what AI said. By harnessing the power
of big data analytics, businesses can unlock
transformative insights that drive strategic decision making and enhance operational
efficiency. I can't lie. Now every time the moment I see the
word transformative, I start looking for more clues because it's one of those
words that tries really hard to make things sound
word changing when most of the time it's just talking about a newsletter redesign or
some fancy dashboard. And now let's analyze
why it gives AI away. It over hypes everything. It adds sparkle where
there should be clarity and specificity and
more often than not, it just fluff in a cap
and what to do instead. Honestly, try
removing it entirely. Most of the time, the sentence makes just as much
sense without it. If you must replace it, go for something more grounded,
like helpful, meaningful, useful actually made a
difference because we need to remember that simple
words are not boring. They are just honest and people use them in
real conversation. That's how we talk.
Tapestry, a tapestry off. Example, from AI,
the museum displayed a stunning tapestry depicting scenes from ancient mythology. And I think hGPT seems to be
obsessed with this world. You could ask it to write
about online banking and it can say something like
it's a tapestry of security, convenience, and modern finance. Why it gives AI away. It's trying to sound
elegant and meaningful, but it's doing that
thing where it borrows death instead
of creating any. Also, we don't Talk like this. I don't sit across
from a friend and say, Oh, how is my weekend? My weekend was a tapestry of
laughter and homemade pasta. If I do talk like this, please take my phone away and what to do instead I
sound more natural, how to edit the sentence. All things what they are. If something is connected,
say it's connected. If it's layered,
say it's layered. If it's a little chaotic but
somehow working, say that. For example, the museum had this wild room full of old
stories stitched into fabric, mythology, mets,
embroidery, basically. That's how a human says it more casual and
much more specific. This is about it's
all about and here is the example I once saw in one of the AI generated results. The ultimate gamers guide is all about giving you
the strategies to conquer any gaming challenge and level up your skills
without breaking the bank. If you've ever use this
phrase, don't panic, if it's one of your
personal favorites, also don't panic because
we all use this. But here is the thing. Now, unfortunately, the moment
I see this is all about, I get a little bit suspicious because nine times out of ten, it means the writer didn't know how to actually
say what it's about. It's a fill in the blank
move that sounds helpful. Dogs, the real explanation and
why it might give AI away. AI uses this construction a
lot because it's formulaic. It sounds like it's
making a point, but it's not actually
saying much. It's describing Lasagna as a
celebration of layered food. Okay, but what was in it? What you do instead? Get
more specific instead of. This guide is all about
helping you save money. Try. This guide is
basically a cheat code for not getting destroyed online and keeping your wallet intact. One version is very
generic and it has been repeated
million times online. The other sounds like something someone might say with chips in their hand and that's what we should be going for fostering. And here I have an example. I establishes a coercive
visual identity, fostering a sense of reliability and professionalism.
Why do I know I? Because it sounds intentional. But it tells me?
Absolutely nothing. Fostering what exactly? A vibe, a culture, a polite nod
from a potential investor. And when we use the word foster, it's often because we are
trying to sound thoughtful, but haven't quite figure
out what we actually mean. It's a placeholder for a
feeling that we can't name yet. And why it gives AI away? Because as you already
know, AI loves vagueness. Fostering is a
comfort word for AI. It's literally the equivalent
of things and stuff. And what to use instead, what to do instead say
what's actually happening. Instead of fostering a
sense of connection, try helping people
feel seen or creating a shared language or making the brand feel like
a real person. Not a corporate void, the more specific you get, the more human you sound. Think of it us. I also used to write
like this a lot. Think of this us mostly because I thought I was
helping the reader. Giving them a clear way in, offering a metaphor, making things feel smart
but accessible. But at some point, I realized that every
time I saw that phrase, especially in a blog post or Substack post or
articles written by I, it made me stop and
question the whole thing. I I stopped using that
often that often in my own writing because of the fact that AI
loved this pattern, it started to feel
like a shortcut. Like it's trying to get clarity without actually sitting
in the ness of the idea. Think of influential blogs and forums as your digital mentors. I mean, sure, that's a
tiny little sentence, but it also feels like something you'd read in a linked in carusel with too many emodis
and not enough essence. It sounds like a filler, like a sentence that
has no soul behind it. And the giveaway, why we know is AI because AI loves
telling you how to think. But the human writing
doesn't usually do that. We don't walk around
telling our friends think of my ex as the
learning opportunity. We just say he was the worst, but I also weirdly learn
something from him or from him. If you catch yourself writing, think of it as pause. Ask yourself what I'm
actually trying to say. What's the situation I'm reaching for and
say that instead. For example, we could transform the sentence from AI Think of influential blogs and forums as your
digital mentors into. I learned more from
the comment section of that forum than I
ever did in school. It was like a free mentorship,
but yeah, messier. It's like the AI
favorite metaphor. It's like having a front
row seat of the future of crypto or from the
comfort of your couch. Okay. I get it. It's meant to be
visual, familiar. Relatable. But the moment I see it's lie in this
perfect little setup. I get this weird
feeling in my chest. Like, someone is trying to sell me something
I didn't ask for. The thing is AI
loves this phrase. It's AIs go to structure, easy formula and
instantly recognizable. So much that sometimes when a human writes it,
it reads robotic. I also used to lean on
it's like too much. Too much when I didn't quite
know how to transition in a sentence or when I was
trying to sound clever. But now I see it for what it is. A little sign that says, I didn't sit with
this long enough to make it feel more specific. So what you can do instead? Well, you already know,
just say the thing. Make it sound like you
actually thought about it. Instead of it's like having a front row seat in the future. Try you were watching something shift right in front
of you on your screen. Now it's not a metaphor
and feeling they are way harder to fake by a eye
and it shows in the writing. Not only, but also. Yes, it's giving high
school essay energy and it's time to talk about one of those phrases
that sounds like a eye is trying too hard
to impress a teacher. Not only do pillar pages
help users navigate better, but they also position your
as an industry leader, sure, it sounds neat
tidy and balanced. Yeah, too balanced,
too tidy, too neat. And you already know it. AI loves symmetry. It really likes when
a sentence walks in a straight line because it was trained on so much
academic writing. So not only, but also is
one of AI's go to moves. And why it gives AI away? Because it sounds
formal and scripted. Like a headline from
a corporate brochure. And to be fair, it's not
wrong. It's totally okay. If you use it, I just want to say that AI loves
this phrase too. It just doesn't feel
like something a person says when they actually care
about what they are writing. Unless, and this is important. These two things
you are comparing are kind of together
or surprising. That's when I think it works. For example, not only
did she ghost me, but she also sent me
a Christmas card. Six months later. Yes. This is the normal usage of this phrase, and absolutely use it there. Not only was it raining, but also somehow the Uber
driver was my old math thir. Also aloud yes. But if you are just listing two benefits of
marketing strategy, no, it instantly tells us
that AI was used there. Be how real people
would write it? I think humans drop
the performance. They just say What's true. For example, using pillar pages makes it easier for people to find what they
are looking for. And yeah, it makes your brand look
like it has its together. Oh, pillar pages do two
things really well, help your users and make you look like you know
what you are doing. It's not this,
it's that pattern. You've probably
seen this format. As always, it tries
to sound wise, mysterious, and a little poetic. This is the sentence structure. I see everywhere now in car results, in
Instagram captions, AI generated essays on sop song, and even well meaning advice posts that were also
generated with Chad GPT. It's not about being selfish. It's about protecting your
piece. You get the picture. At some point, this
pattern went from clever and grounding
to so overused, it started sounding fake. Of course, AI loves this structure because
it's clean, predictable, parallel there is something satisfying about it
like sentence symmetry. But there's also exactly
what makes it feel robotic. In the second, you
see that structure, you can swap the nouns like refrigerator magnets and
get the same effect. It's not attention,
it's the intention, it's not writing, it's
resonance, don't get me wrong. Sometimes this structure
works really well. I also use it in my
personal writing, which has never seen an AI. But when it shows up in
back to back sentences, it really starts to feel like a ted talk never
quite lens because real people usually
don't phrase things so tightly when they care
about something deeply. They stumble, they
try to explain, they side halfway through. It's also important to mention that when you
know another personally, you know them in real life, it's really easier to
recognize they wrote it themselves if they could write a thing you are
reading themselves, and you can easily tell if
it's them or if it's Cha JPT. Because you knew them before, because you know them,
their talking style, how they think, how they
articulate your ideas, and you know their
natural manners. But it's so much more difficult. When you don't know an author
personal in real life. And you see someone's
first article, first post, and the author is someone
new and anonymous to you. Then for most people,
it's really difficult to distinguish whether
the vocabulary, the post is or isn't at a level appropriate and
natural for a given author. And yes, that's exactly where the knowledge
from today's comes in handy and can help you more easily recognize the
auspicious patterns. But I also can't stress this enough and I need
to repeat it one more time that AI writes like the sources
it was trained on. Which means, yeah, AI mimics us. AI mimics humans. But just not the
writing style of the humans we work and collaborate
with on a daily basis, not the casual, very spoken
language we are used to. Of course, it's changing with time changing for the better as AI is in the constant state of being trained
on new datasets. But what is funny that
now AI is also being trained on articles and
posts that TGPT wrote. So it's being trained on
the texts that AI wrote. It's like it's
learning from itself.
15. AI Detectors: I detectors. One of the newest GPTs
passed the turing test, it was no big deal. Researchers at UC San Diego
run this experiment where people chat it just text for 5 minutes with either
human or the AI. Then they had to guess
which was which, and you know what happened. How the GPT got mistaken for a real person 73% of the time. Not 50%, not even close. That's well past the
random guess line that we literally
couldn't tell anymore. And that changes a lot more
than people want to admit because now that post you are reading on LinkedIn
could been AI. The video that made you
emotional could been AI. The essay you are grading, the code you are reviewing, the pich deck you are being
with your actual money. Yes, you get it. What about those shiny AI detectors
everyone spent money on? Unfortunately, many of them are pretty much useless
at this point. Sorry. I think it's also
the question we will have to ask ourselves
sooner than later. How do we build systems that still work and
maybe even thrive, no matter who or what? Created the input. Because if you want results
that AI can't replicate, then you need inputs
AI doesn't have access to lived experiences, deep understanding,
unusual contexts, just stuff that doesn't
come from a training set. Maybe that's where the
real opportunity is now. But that's another topic
back to AI detectors. The tricky thing about
AI writing now, yes, it's getting harder to
catch by AI detectors, even the most expensive ones because people have
learned how to hide it. They are mixing and matching, prompting, editing,
paraphrasing. Running it through rewriters. It's not AI or human
like it was anymore. It's both, and that's what
makes detection messy. Even I teach advanced
prompting techniques in my other courses
because I didn't want to gate keep
all that knowledge, hoping people will use it
to make their lives easier, that it would be
helpful white magic for them to the lives
a little bit better, a little bit more enjoyable
because they have more time for the things that matter most to them
for their loved ones, not something that
ends up being used for the wrong or
unethical reasons. I'm constantly reading something that might be AI generated, blog post, emails, pitch decks and because it's
part of the job, I've tested just about every
AI detection tool out there, GPT zero GPT, copy leaks, many, many more that were brand new and promised
excellent results. And also some chrome extensions
and safari. So were okay. So were completely crazy and unreliable because
they were flying my own personal untouched by AI writing as I and most were
somewhere in the middle. But the main issue, they are too sensitive
when something sounds too polished and too blind when something's been
lightly rewarded. If the sentences are short, the detectors freak out. If the sentences are long
and more often than not, they don't tell you exactly why they are flagging something. Anyone who wants to sneak
it through probably will because I know the
weaknesses of AI detectors. If I ever come across something
truly worth recommending, something I truly trust, I will update this chapter right away and post an
announcement about it. Don't worry. I don't plan on stopping the
tests anytime soon, even though I've already tried more tools
than I can count. Honestly, it's wild how
much money some of them make despite being so far from
being perfect or reliable. This escaped from the future
and I need to add something very important though
a little bit set. There are already thousands of cases where students
were falsely accused of using I ClsenI detectors
work by measuring perplexity. By running your text for
a large language model. And considering the probability that the model would have
chosen the same words, the same word
options as you did, the same words as and
what is funny is that some AI detectors
like things like US Constitution as
being AI generated, which is an obvious sign
that they are unreliable and they aren't measuring what we think they
are measuring. Why does it happen? Because the US Constitution has appeared so many times in
the EIs training data. So now obviously, it's very easy for large language model to use the words from US Constitution because it was strained
on it so many times. So what do remember from here? AI detectors aren't
telling you whether AI wrote detext you are reading or you are
pasting into it. But whether had bought ChatGPT or other lag
language model Hoot. Could have written the text. Also as a bonus, here I'm showing you some
algorithmic clues which AI detectors rely on and
measure when judging the text. Some people also worry that artistic writers
might experience their text being flagged as AI generated more often than neurotypical
writers because they naturally like writing using patterns and
they prefer concise, perfect sentences and
perfect structure. Also, this is very important. This speaks volume. Even the authors of AI detectors repeat and write officially
that their tools aren't reliable tools
for recognizing dishonesty and arrangement for official or academic purposes. I think that speaks volume.
16. How to Make AI Writing Feel More Human? How to Trick Everyone Into Thinking You Wrote It?: How to trick everyone into
thinking you wrote it. How to make AI writing
feel more human? Because maybe you
are using AI to help you write and
honestly, I get it. I use it too, but not
for my personal writing, not for the honest
authentic writing I love. But in work, I think it can be a really solid thinking partner, especially on those
super busy days. But here's the thing. Just because the writing
is technically correct, doesn't mean it sounds
like you or sounds human. If you've ever read something
back and thought, wait. Why does this sound
like I've been possessed by a customer
support agent? Let's talk about how
to keep the help but lose the AI is vibe. These are a few dead
simple ways to humanize AI writing without needing another tool extension
or paid prompts. Rewrite in your actual voice. As you already know
from this course, a lot of AI writing
is overly tidy, polish to the point where
it feels just blank. AI might give you
a sentence like AI generated content typically follows a highly structured
and predictable format. But you could just
say the eye writing all sounds the same
after a while. They see the difference. One feels like it came from
someone with a personality. The other feels like it came
from a brochure or PDF. Of course, you don't need
to rewrite everything, but just check if
something sounds too safe, too netural or too formal, tweak it until it feels
something you would really say in a conversation
and protip, read it loud. If you cringe a little, you know it's time
to make it more you. Inject something real, a
story from your real life. Because AI writing is very
often like bash painting. It's inoffensive,
clean, and forgettable. But if you want
your words to land, you have to bring them to life. And real stories, no
matter how small are one of the fastest ways to do
that because AI might write. Imposter syndrome is common among creatives and
professionals alike. Which okay, is true, but sounds very dry and
you could say instead. A few months ago, I rewrote the same paragraph six
times before realizing it. Oh, this is just
imposter syndrome playing dress up as
my perfectionism. Even just one sentence can flip the energy
of a whole post. AI doesn't know what
your Tuesday felt like. You do. Say it, use this. Use those emotions
and those memories. Losing the rhythm. Yes, I've already repeated
it. I don't know. Five times or more, you tell me, and I also
loves less of free. Like you will gain clarity, boost productivity, and
grow your business. Okay. But would you ever
say that loud? Try breaking the rhythm, leave a thought hanging. Add a pause because your sentences don't
need to be perfect. For example, honestly,
I just wanted to feel a little less
crumbled every morning. Clarity came later,
and as you can see, it's much less about
the perfect flow and more about feeling through. Here is another very
sounding paragraph. Time management is essential for professionals and
fast paced industries. With proper scheduling, task prioritization,
and delegation, individuals can
maximize productivity and achieve their
goals efficiently. Now, let's humanize it. Look, if you are to do this, it looks like a CVS
receipt and you keep skipping lunch because
meetings don't stop. Yeah. Time management
is something you should figure out because you don't need to say it the
way a textbook would. You just need to say
it the way you would. Say goodbye to B speak. AI has a weird crush
on corporate language. It will spit out stuff like this cutting edge
solution delivers enhanced value for
cross functional teams. But come on, you could just say this tool makes things easier for people who
actually do the work. Be okay with some mess. AI wants to close every loop. But people don't always
talk and finish fuss. Sometimes we hesitate. We very often hesitate. We rumble. Sometimes
we Tiva mid sentence. Sometimes we end a sentence with a question and
don't answer it. So let your writing have edges, weird metaphors, but
really weird and creative. You ones, not the
metaphors that is domain, quiet confessions, because that's where
being human lives. And I would say edit
with your intuition, not Taji PT or grammarly. Because tools wants you
to clean everything up. Capitalize every subject line, remove every passive phrase,
shorten every sentence. But the more you obey the rules, the more sterile your
writing becomes. Don't worry if something
is slightly off. If it feels like you, it
probably reads like you too. What's actually bothering us? I'm still thinking about
it and I'm still doing more research among my
colleagues and my clients. But I think many times it's not the fact that something
was written by AI. It's the way it feels
too clean, too boring, to save, too, like it's
written from formula. Stripped of any real voice built to sound like
everyone else, someone or no one. And in a scroll happy world
full of same sounding stuff, blending in is the
quickest way to disappear. What can you do instead? What can we do instead? I really, really can't
stress this enough. Skip the vogue navigating
change phrases. Share that story about the
day you nearly walked away. Swap delivering results.
For the moment, it all went sideways and
won't pull you through. Don't say you empower
your audience, like they are actual
people because they are. I think the creators who will
stand out tomorrow aren't the ones writing the
most advanced prompts. They are the ones with the
most unique and honest voice.
17. Final Words and My Question to You: Words and my question to
you. Okay, we made it. If you made it all the way here, I hope your reader for
AI generated writing is just a bit sharper than it
was before or maybe a lot. Whether you are a curious
reader, a content editor, a content creator,
or someone who just really values, knowing was real. Thank you for going
on a slightly nerdy, but hopefully fun
journey with me. This wasn't about blaming or
shaming anyone who uses AI. It's about, no, I'm using
the formula that AI loves. It's about awareness that pause when you read something
and suddenly you think, Oh, something is off here and now you
might even know why. Also, if this topic
got you curious, I'm currently working on a separate course about
spotting AI generated images, videos and photos because yes, I've been tracking
patterns there too, and some of them are wild and I haven't seen
anyone mention them. If that sounds interesting, stay tuned and if you enjoy this course or learn something new, which I
really hope you did. I'd love to hear your thoughts. What surprised you the most? Did any example really
click with you? Have you spotted anything
suspicious recently, or maybe you have a friend who suddenly started
using Em Dash. So I'd love to hear from
you, leave a review, or share your reflections in the class project section and
in the discussion section, and I read every single one. Thanks again for being here with me and for caring about words, truths, and attention to
detail as much as I do. Susan,
Kasia Pilch, Online Strategist & Marketing Specialist
