Transcripts
1. Introduction to Making ChatGPT Sound Human: We've all seen it. In today's
fast paced digital world, that robotic AI tone. It's everywhere. And yet Chachi PT is usually
the culprit. Real problem, most
people don't know it sounds robotic
or how to fix it. In this class, you'll learn
how to humanize Chat EPT. I'll show you how to train it to sound like you, your tone, your style, your voice, so your content
connects not cringes. Think social posts,
newsletters, articles, all sounding brandworthy,
not bot generated. We'll break down why AI
content feels so flat, the cliches, the
lack of emotion, the generic tone, and
how to fix all of it. You'll create your own custom
GBT, write better prompts, and even inject personal
stories to make your writing actually
feel well, human. By the end, you'll
be whipping up content that sounds like
you on your best day, authentic, relatable,
and way more effective. So let's ditch the AI
slop and get writing.
2. What Makes AI Writing Sound Robotic, and How to Fix it: The term AI slop was coined to describe AI generated content
that sounds mechanical, sterile and just lacks emotion. In this lesson, I'll show
you how to fix this. So at the end of it,
you'll understand the key elements of what
makes content sound robotic. You'll learn to humanize AI responses for any kind of
content such as articles, social media, and
even newsletters. These tips and strategies
will help you to personalize hat CiPT to respond exactly
how you want it to. So let's get started. What
makes AI sound robotic? The first one is the
overuse of buzzwords, generic phrases that
lack originality. Examples are AI is a game
changer in content marketing. The next one is cliche phrasing. This is where phrases
and terms are overused, and they start to feel
a bit uninspired. Examples of this are in today's fast paced
digital landscape or leverage cutting
edge solutions. Just terms that are
used frequently, and they're a bit
cliched and tired. The next one is lack of emotional connection
with the reader. An example here is
customer satisfaction is very important and can be achieved by ensuring
good customer service. This lacks any kind
of emotional tone, and it's just flat dull. Absence of personal voice. This is where the
generated content. It just lacks
unique perspective. There's no personal
insight or stories, and it's just a bit
bland and dull. You want your reader
to connect with your writing on an
emotional level. An example of how
AI gets this wrong is AI is transforming
industries. Well, how is it
transforming industries? This is where your
personal insight and perspective come into play, and you're able to fuse that
with ChachiPTs responses. And to do that,
you would give it your own personal
anecdotes in the prom. So you would prompt for the
content, but you would say, add in my personal story, and you would give
the details of that. This just gives it a bit
more info and allows it to use these personal anecdotes
in the generated content. Great tip here is to use first and second
person point of view. This is where you
would drop the Wi and you speak to one person
and one person alone. You use first person perspective
or point of view and say things like instead of W and
this engages your reader. The following are
some elements to generating human
sounding content. The first one is to set the tone and persona
and the voice. So you would specify the tone. Is it going to be
professional or friendly? You would then set the voice, and active voice works best here because
in everyday speech, we naturally tend to
use active voice. This would aim to mirror how
we construct our sentences when we're speaking to friends
or family or colleagues. Lastly, use first person
perspective or point of view. That's where you
say things like, I believe that or this
is my approach too. With second person,
you would say, you can improve or
assess your needs by both these approaches
work equally well. Then you could take step
further and take on a character or persona and ask it to write in the
style of that person. So the example prompter, write in first person
point of view, and in the style of
William Shakespeare. Then using natural
language and structure, asking the AI to use plain, simple language and contractions to make the language
more conversational. By contractions, I mean
that instead of I am, you would use I'm or do not, you would use don't. And just by simply stating
language contractions, the AI will know what
you're talking about. Example, prompt would be use plain language and contractions to make it conversational. Then you also want to
vary the sentence length, and you would prompt for that. So typically, when AI
generates content, it's very rigid and uniform
in the sentence length and structure by telling it to vary this length and to give
it a more natural flow, it just mixes that up and it just sounds and
reads a lot better. Then again, adding personal
elements and emotions. This is where you bring
in all your anecdotes and some sensory language just to connect emotionally
with the reader. You want to try and put them there with you at the
time of the event. So this is where you would share personal stories
or experiences to create a more relatable and empathetic tone
in your writing. To prompt for this,
you would say convey excitement rather
than just stating facts. Next, you want to refine for engage that just means you would request revisions to enhance the tone and style for
better engagement. You would ask Chachi Bit to critique the content to make it more conversational and natural. An example prompt of
that would be critique this text for a more
conversational and natural tone. And then once you
get the response, you can refine based on that
and just improve as you go. That we know what makes writing
sound robotic and dell. In the next lesson, we'll
create the actual instructions that are going to be used
for your GPTs and projects, and we're also going to look at some tools on the
GPT marketplace. So we will look at custom GBTs and all of the instructions
that you can copy and paste into your
custom GBTs as well as copy and paste them into
your chat GPT projects. Then the GPT marketplace
is where you can find those ready made GBTs to be able to help you humanize
your content. Recapping the lesson.
Using these key elements, you're now able to create clear instructions to
transform robotic, sterile AI slop into natural sounding and
engaging content that your readers are
just going to love. This is where you would set the tone style persona
and voice of the writing, you're going to ask ChahBT to
use natural plain language. In the prompts,
you're going to add some personal elements
like anecdotes or stories or anything
that you're able to connect with the reader on
a more emotional level. Then for each response
that you receive, you want to critique
that and improve it to be more conversational
and more natural sounding. Once you get the responses, you're then able
to update, tweak, and improve wraps up the lesson. And remember, you're more
than welcome to download all the files that you see
from this lesson and the next, and those are all
the instructions and prompts that you would use in the custom GBTs and projects. And you can find those
in the resources folder. That's it for the lesson.
I'll see you in the next one.
3. AI Writing in Action & Building Your Custom GPT: In the previous lesson,
we learned what makes AI generated content sound
robotic and lifeless. Next up, in this lesson, we're going to be putting
all of that into practice and creating the instructions
for our custom GBT, or even to use inside of the
projects feature in hachPT. Now, just to mention that
this is a paid feature, so you would need haGPTPlus, but the ability to be able
to customize your own GPT, to have it respond
the way you want it and in the style you really is a valuable addition
and essentially makes ChaiPT pay for itself
by having this feature. Now, let's move on
to the instructions. Here we are looking
at the instructions. These were generated from the key elements that we looked at from
the previous lesson. Let's walk through it, and you can see
what we're going to be including in
these instructions, and then we're going to
put this into practice, and you'll see the type of content and the type of
responses we can get from ChachiPT once we have these instructions
in first up, we are going to have
an opening prompt, and this is to assume a persona. So we're telling Chachi Bit that you're a content
marketing copywriter that focuses on creating engaging
and conversational content using the instructions below. We're giving it all of
the elements that we want to ensure that it is
conversational and friendly, and the tone is
consistent throughout. Then we're asking for some key information to be
able to process the request. And that is who is the audience, the experience, level,
pain points, interests. What's the goal?
So what's the goal for the piece of content
that's to be created? What tone best fits it? So casual, professional, witty. And then if details
aren't provided, infer them from the context. So any of the details that
are put into the prompt, it will use that generate. Is an AI words to
avoid Knowledge file, and this is just a
simple list of AI words. And we're saying that if the
restricted word appears, replace it with a
suggested alternative. To look at this list, let's hop over to
the PDF file here, and this is just what
we would upload to the knowledge library
inside of our custom GPT. And here, we're looking some of the words that ChachiPT
is consistently using. These are absolutely
loved by AI models, and they appear a lot
in generated content. So we're just trying to
get ahead of that and give alternatives for when ChachiPT wants to
use those words, we're giving it a more
human alternative. Hopping back into
the instructions, we are now looking at the
core writing principles. So again, we're saying it needs to be
conversational, natural, engaging and write as if
speaking to a friend. Then we are giving a
example of what to avoid. So this comprehensive
guide will explore the myriad benefits
of automation. We're giving it a
more conversational, human written plain
language alternative. Then the first and second
person, as you remember, this is a key element
to make things sound more personal and direct as if we're speaking
directly to the reader. Active voice over passive voice. This is essential because generally AI defaults
to passive voice, and we want to keep the
sentences strong and direct. We're giving it a couple
of examples there, so avoid this and use that. Keep sentences and
paragraphs short. No walls of text, aim for two to three lines
max per paragraph. Just keeps it readable,
keeps it skimmable, and means that we're not going to have giant
walls of text, which AI and language models
typically love to produce. Avoid redundancy and
over explaining. So this is another key aspect of language models is they do
love to overexplain and just fill up the spaces and the
gaps there with lots of extra information that might not be to the point and needed. The tone and style, this
is quite a key one. So writing at an eighth to
tenth grade reading level, this keeps it light. This keeps the word
choice, the verbiage, and the overall tone
a bit more casual. And of course, if you're writing technical documents
or anything that requires more terminology and jargon and things like that, then you would, of course, remove this or increase
the reading level to say, university level
or college level. Cliches or corporate speak, this is an essential one. And then we also want to inject a bit of personality in humor. We're trying to hook the reader. So we're saying that we want to start with a surprising
fact or statistic, a bold statement or thought
provoking question. This is typically good
practice with copywriting, and then also
providing relatable challenge or pain point. That's also another
alternative for social media, so we're making this a
universal custom GBT, so it can work with both written articles and
also social media posts, newsletters, that sort of thing. Then make the content
relatable and engaging, use real examples and stories. And of course, this is where we would need to prompt these into the input that we're giving ChatBT because there is we want to be able to give it the first hand
experience that we have. So that's where we would include anecdotes and case studies, and this should be done
at the prompting level. Want to address common pain
points. This is quite key. If you're writing in terms
of marketing or appealing to customers or business owners or in the BTB marketing space, then you want to understand the pain points that the
user might be having. This would allow the writing to highlight some real frustrations and show a bit of empathy. Use playful, engaging
language, metaphors, idioms, all of that is
great to have in there. Readability, break up
into clear sections, use subheadings for clarity. This is essential for good flowing, well
structured content. Use bullet points and lists. This is key for helping
skimmers so that they can just get the gist and grasp
key ideas quickly. Important phrases. Chachi BT naturally does this. So we're just ensuring
that this continues. Then closing with an
impactful statement or summarizing the
key takeaways, ending with a strong
call to action. And then here we're
just giving it a few extra instructions. So do not copy examples exactly. That's quite key to add there, because if you provide
it with examples, Chachi PT does like to latch onto them and reproduce them. So if you don't want
that to happen, then just keep this
in the instructions. And then, lastly, just
to make really sure that it knows we want natural and conversationally
written content, we're adding that
in there again, and we want to prioritize
clarity, variety, and impact. This is quite good to
have because naturally ChachiPT will opt for
speed over precision. So it would want to get a response out to you
as quickly as possible. So if we're saying prioritize clarity, variety, and impact, it will then give that
a bit more priority, and that would lead
to better responses. This PDF is available in
the resources section of the lesson for you
to download and use in your own custom GBT. And also there you will find another SEO writing
GBT instructions. These instructions
are designed more for a long form SEO article, and you'll be able
to find all of that in the resources section. So here you can see what
we're working with there. This is more intended
for long form content. Moving on, where
would we add these? So to create your custom GPT, you would come into CAGEPT and
from the left hand column, you can then find it under the Explore GPTs and you would
hit Create GPT over here. Once you're there,
you're then able to start customizing your GPT. You can either go to
the Create option, and here in plain language, you would just give some details and parameters on what
you want the GBT to do. Already have the
instructions laid out. It makes sense just to come into the configured tab and start
creating it right away. You here, you're
able to then give your GBT a name and
a brief description, and here is where the
instructions will go where you copy those
instructions we just discussed, and this is where they belong. Ation starters, these are just quick shortcuts
if you wanted ChatBT to start writing straightaway with a
specific purpose in mind. Then the knowledge library, this is where you can add all of the knowledge files
that you would want to just enhance and serve as a reference
for this custom GBT. So if you had maybe
some examples of writings that you wanted
to use for inspiration, you could include
that here as well. For this example, there is just that list that
we looked at previously, that is this one, and that
is included over there. Benefit of using a custom GBT for this purpose of creating humanized content is that you can get some
deep customization. You can really tailor the
model's behavior, tone, and responses to suit
specific tasks and even assign personas to really
enhance its writing. You have fine grain control, so this allows you to
adjust the model style. Voice and knowledge base for some really specific use cases as creative writing or even
writing social media posts, it's able to handle each
of those with ease. Some GPTs are really well
suited for standalone focus. By that, I mean, it is great for individual focus task where you need a highly
personalized experience. It's also really great
for purpose built focus. So if you have a very specific
style of writing in mind, then creating a custom GPT
just for that is the way to. Now let's have a look and see what it would generate for us. It has all of its instructions, and now we're going to ask it to just create an article for us, and I'm going to be using
a simple prom tier, which is write a
300 word article 0N flow state for productivity. Now, with the
instructions in place, this means that your prompts can be a lot more simplified. You don't have to go into
as much detail and add all of the context and the information and
the instructions. It's already there in
the GBTs instructions. Instead, what you can do is use this initial
prompt to maybe give some personal insights
or experiences like those anecdotes and personal
stories I mentioned earlier. This is a great
opportunity to add that human touch
into the writing and have that be
given back to you when Chachi BT
writes the content. This example, we're
going to keep it just simple and ask it to write a 300 word
article 0N flow state. I'm going to run that. Here, we can see the result
is a lot more conversational. It's a lot more direct. We are seeing that it is
not creating walls of text. The language is simplified. It's plain language. It's conversational,
and we're able to see some list that were created here just
for easy reading, and it really just has a much
better feel and flow to it. We've got the steps here. First draft here, it's
done a pretty good job. And, of course,
like any AI output, even after you've done
all this tweaking, it's not going to be 100% ready to publish we would
still need to go in there and edit and
do a little bit of polishing just to make sure
that it has that human touch, and also we would
want to add some of our own insights and
our own perspective to make sure that it
really is something that we can be proud to publish. Let's have a look and see what
this would look like if we asked the standard ChaiPT
to create the same Now, to see that in action, let's have a look and see
what that would look like on HHIPT without any of
these instructions. So the stock standard out
the box version of HHIPT. To do that, I'm going to run
the prompts side by side. And for that, I'm inside
the Open AI playground. You can get to it from platform dotpi.com forwardlash
Playground. And this is where we can set up a comparison configuration to be able to compare
what the response would be with a GPT that has certain instructions
versus a GBT with a different set
of instructions. We are going to be using
the standard GPT 40 model just for the purposes
of this demonstration. And as you can see here,
the system message. So this is the instructions that we're giving to Chachi BT. This is set up with the instructions that
we have created. And on the right hand side, we don't have any instructions. So this is the plain default out the box vanilla
version of Chachi BT. And let's bring up the prompt
and we're going to run it. And already, we can see here is a much more
conversational approach, and this is engaging, and it is in plain language. We're getting a nice variance
in the length of sentences. And then if we have a look at the standard chat GBT version, we're getting the classic in today's fast paced world
maximizing productivity, it goes on to give us very detailed and
excessive explanations. And the choice of words are definitely not conversational
and plain language. And we can see there's a
lot of repetition with the cliche AI phrases
that are being used here. Also, the length of sentences
is considerably higher, and we're getting a lot
of these blocks of text, which are not good to read. And you can probably
see here that the length of each paragraph
of text is very similar. So you can probably see
that each paragraph is very similar in
its length, and, um each one has four lines, and it is very uniform in how it's been written
and structured. So this leads to a very stiff, mechanical, robotic result. And from there,
you're able to see that just with some of
these instructions, we were able to get it to
be more conversational, to use plain language and just generate
content that sounds much closer to what
a human would write than if we were using
Chat GPT normally. Now, I also wanted to show you how you can work
with that in projects. So when you are working with your projects and
to get to projects, you come to the project section here in the left hand panel, you create a new
project, and after that, you then have the options
to add instructions, and this is where you would
add those instructions, as well as adding files. So this is where you could add the list of terms that you
didn't want ChatBT to use, or you could add
examples as well. So it has a lot of the
same functionality that we saw in the custom GBT. When would you use
the project's version over a custom GBT? Well, it's entirely up to you, but one of the big
benefits with a project is that it has context over
all of the conversations. So every time you
start a conversation, they will appear over here. And if we have a look at this
project for this AI course, you can see over here that
it has all of these chats. And what would happen is every
time you start a new chat, it would retain the
information that's being used in some
of the other chats. So it really helps to streamline your workflow and
projects then have this persistent memory and context that carries over to all of the other
conversations. That is a really good feature. But of course, custom GBTs have the benefit of being very, very tailored and
specific you can set up the capability so
that it has web search, Canvas mode, Di,
code interpreter, as well as creating action. So this allows it to communicate externally with other apps and databases and APIs
to be able to just make your GPT much
more powerful. The last thing I wanted to
show you is that you're able to use ready made
GPTs such as this one. This is the undetectable
AI humanizer, so it will take an existing post and try to humanize it
as much as possible. And to see that in action, let's take this output
that we had from the standard version of ChachiBT here and let's put it into this to see what it
could do for us. And over here, we're
seeing the configuration. So it's just telling us how it's going to
apply the changes. And after that, it goes ahead
and processes the text. And already here I'm seeing that it's got some of those
cliche AI phrases. So even the custom GPTs
they're not perfect. And this is another
reason to create your own custom GPT or use custom instructions
for projects, so you can get your GPT sounding exactly the
way you wanted to. More tool to look at. And this is another custom GBT, and this you can find
on the marketplace, and this will
analyze the writing, and it will then create the prompt that would
generate that writing. So kind of reverse engineers
the text that you provided. And to do that, we're
going to have a look at a linked in post, and this is from a very
prolific copywriter. His name is Dickey Bush. And here, this is
a LinkedIn post, and this is just a
LinkedIn post that is very engaging and it's got a
storytelling element to it. So we want to reproduce this, and this custom GPT is
how we would do that. So I'm going to copy that text, come back into that custom GPT, and paste that in
and then run it. From there, it's generated the instructions and
the prompt that we could use to create this or to recreate this for our
own LinkedIn post. And there we go. From that, you can see that you're able to create a custom GPT or even use these instructions in your
projects to be able to create the content that sounds like it was
written by you. And of course, once you
get the generated content, go back, ask Chat GBT to
improve it or tweak it. You can ask it
things like rewrite in a more friendly or
conversational tone. Show more enthusiasm about the topic or insert
this personal story, and then you would give
information on that. All of these type of things
will help you just iterate and get to that version of the content that
you are happy with, and you would then be proud
to associate with your brand. But I hope you can see the
power and the versatility of custom GBTs as well as using these custom
instructions in your project. That's it for the lesson.
I'll see you in the next.
James Mew, Sharing my AI and productivity hacks
