Transcripts
1. Introductoy Video: Pt got released a while
back yet there is not a single comprehensive
course based on proved information provided
by open EI itself. That's why I'm here.
I've condensed all the most relevant
information regarding GPT. This information comes from open I block post Discord server a lot and I mean a lot of experience but what is
it actually teaching? Well, in the first less
song you're going to log in and learn a basic
introduction about GPT. Then we're going to talk about basic prompting
techniques so that even if you know
nothing about GPT, you're going to be
able to catch up. After that, we're going to dive into advanced prompt
engineering and we're going to uncover
12 prompting techniques which are going to explain
to you how to leverage GPT 3.5 the free
version to its max. For example, in lesson seven we're going
to talk about how to give GPT 3.5 Internet access, even if it's a feature
that is theoretically designed only to be
available for GPT four. As we finish the actual
ChagPT master class where you're going to learn
everything about GPT. We're going to then move onto three practical ways
to use GPT right now. We're going to get
onto Instagram, SEO, so that you can optimize your Instagram
description in order to appeal to your potential
customers if you're a business. We're then going to
move onto Youtube CO, where we're going to combine
the power of GPT with another extremely powerful
AI tool which is called Que. And you will learn step
by step how to create a video which has high demand
for and low competition. We're going to give an
natural practical example and proof it works
in the lesson. After that, we're
going to script an entire ebook from scratch so that you will have your
perfect outline for your ebooks if you're
interested in learning GPT, increasing your productivity as a student or business owner, learning how to problem
solve with GBT, and leverage this extremely powerful technology to its max, I'll wait for you in the inside.
2. Class Project: Welcome and thanks for rolling. Before getting into the lessons, I will quickly go over the
class project of this class. In the assets of this course, you're going to be
able to download a prompt list which
we're going to create together in the
last three lessons. And it's made of
three prompts which will help you in specific tasks. What I want you to do in this class project
is go ahead and create a prompt of your O by using the knowledge
that I'll give you. In this lesson, I will break down 12 different
prompting techniques which will allow
you to understand how to craft a perfect
prompt for GPT. What you will do is
go ahead and create a prompt for a specific
task of your choice. Maybe a prompt which will
help you in math problems. You get to pick the topic of
your prompt so that there is going to be a lot of variety in the prompts everyone is
going to come up to. I'm going to review every
single class project myself so that I can suggest you any corrections that may help you with the output that GPT is going
to provide you. How should you submit
this class project? Well, to submit it correctly, you should type in the
class project section, a message divided
into three sections. The first one being
a small title for your prompt, let's take, for example, the Instagram SEO prompt that
I'm going to provide you. In the prompt list,
you're going to see that I could title it
Instagram SEO prompt, then a small description
of what it does. So in this case it
would be it will help me search engine and
optimize my Instagram descriptions on my profile
in order to attract more people to my profile which will be interested
in my content. And then prompt, and I
will type the prompt, that's how you should submit this class project so
that I can review it. And they will then help
you optimize your prompts. And without further
do getting to the first lesson where you're
going to learn the basics. And I wish you a good learning.
3. User interface explanation, customization, and shortcuts: Get ready for your
mind to be blown. Because in the first two lessons it's going to be pretty
basic and stuff, but after that we are going to skyrocket with insane
prompting techniques and without wasting
any more time, let's log into chip. We do that, we
simply type in open eye.com We are created
with this interface. You can either click
here on GPT and then TrichPT or right here trip. It will ask you to log in, you enter your e mail address, or you sign up if you've never done that or you
continue with Google, whatever you prefer and
you will be in chip. In this first lesson,
we will talk about the interface because
it looks pretty clean, but there are some
features which are slightly iden and not
really easy to see. First, let's start by saying
that there are two models, GPT 3.5 and GPT four. P 3.5 is free and
GPT four is paid. We are only going to use
GPT 3.5 for this course. The plus has access
to GPT four which is connected in real time with the Internet
and the information. Meanwhile, the free version stops his knowledge
at January 2022. However, we show you a way to basically bypass this obstacle. Then in the GPT you can
also use custom GPTs. You can also browse
pre made GPD's, create your own GPT's, or buy your own GPTs if you want, and therefore also
sell your premium GPT. I will consider making course on how to make a GPT
further down the line, but for now let's stick
to the free version. So we close it right here. The first thing that
we're going to do is acknowledge this prompt bar. This is where we type
in our messages. There are also four
recommendations from open AI. We are going to go ahead and click one of them to
see what happens. Let's go come up
with concepts for a retro style arcade
game. You referred to me with my name. I'll show you
how to do that in a second. And as we can see, we have
five names which also have a description below them which describes the games
that we ask for. Perfect. Once GPT has
sent this message, you have four options
at the bottom. The first one being copy, which allows you to copy and
later on paste a message. Then we have good
response which tells GPT that it did a good
job. And it will keep replying in that same
style bed response, which does the opposite, and regenerate,
which allows you to regenerate the answer if
you weren't happy once. If you regenerated, you can see if it was better,
worse, or same. And you can also go back to your first or second
response as much as you, as you can see what GPT tries and do when
you regenerate it tries and switch up a bit the style of the response
as you can see here, it was one bullet point
with a description here to try and give
you more information. It also has given you a genre, you can see and change the
style, a bit of the response. Then here we have a
button to share the chat. If you go Hydrate here
and click Copy Link, you will be able to copy
and paste the link. And if you open a new
tab and paste it, you will be able
to see the chat. And the other
features of the chat can be found in the sidebar. Then by clicking
on the three dots, you can choose again, if you want to share it, rename
it to whatever you want. And then if we go ahead
and click Archive, our chat disappears,
but it's not deleted. And here we can click on
New Chat to open new chat. Now let's look at where
are the archive charts, how to delete them,
and much more. Let's now look right
here in this section. If you click on your name, you can see Logout, which of course locks you out. And then there are custom
instructions and settings. Let's go on settings first. Here you can choose
the team if you want it light, dark, or system. You can check your
archived charts. You can archive or delete. I'll archive so that
it appears here. And then you can also delete all the charts that you've
previously created. Keep in mind that
if you delete it, you will never have
access to it again. And even if you use the
link that you had before, as you can see, if I copy in
the link four under four, this page could not be found. Now, regarding the other
parts of the settings, there is data control here, there is chat
historian training. I highly suggest
you to have this on because what this allows the AI to do is use
the charts that you have had to
improve own training. And it also allows you to save your charts for
more than 30 days, otherwise they will
just get cleaned out. And then you can also delete
your account if you want, but I don't want to.
So let's close this and let's go on
custom instructions. Custom instructions
allow you to get better responses from GPT and
you have two boxes to fill. The first one says,
what would you like GPT to know about you to
provide better responses? For example, if you're going
to use GPT as a student, for example, in this
case you could type, I'm a student 80 years old which circles with science and history, however strongly met. So this will allow GPT to tailor his answers to your needs as
a student in high school. And then you can also
click here on Show Tips, which will show you
the basic questions to which you can reply
to get better responses. And then there is how would
you like Cha GPT to respond? So for me, as I already said, I'm an ego maniac. So I have chosen
address to me as Buzz. However, you could type
something like I like my responses to being bullet
points extremely detailed. Not necessarily professional,
but with a simple language. You can also, if
you want reply in which language should
chachiPT respond. So for example, let's
just say that you do French as a second
language is cool. You could also type,
whenever I type two in French, apply
to me in French. So if I now it's Save and
they say hello in French. As you can see, it's
applying in French. This is the power of
custom instructions. Just remember to click
Enable for new chats. Otherwise, every
chatter you will create will not have
these properties that we have talked about, the basics of the
interface we are about to move into basic
prompt engineering. I don't want you to be
scared from the names because even though
this sounds like scary, I guarantee you
that I'll put it in a simple way so that even
a rock can understand it. Last thing before
leaving this lesson. Right here on the
bottom right there is this small question mark. You can tap it and you can see the shortcuts
which can help you. However, the interface is so
clean that you don't really need shortcuts regardless.
I'll see you in the next on.
4. The 4 fundamental prompting techniques: What is a prompt? A
prompt is a set of instructions that you give
the AI language model. There is a direct
correlation between how good your prompts are and the
responses of the language model. That's the whole reason
I'm having discourse. First, let's start with
the basics of prompting. Let's create a
really basic prompt. Give me the five key
points I need to know while studying
the Roman Empire. Once I enter, I will know that the main
points are geographies, so how vast it was,
government and leadership, the military power,
the engineering, and the cultural contributions. This is the most
basic type of prompt, which as you can
see, it's really similar to what you would
type in a Google Bar. However, if you type
in a Google Bar, the same prompts where are
you going to be directed? For example, the first website
that pops is this one. If I click here, you're
going to see studying Ancient Rome is an incredibly fascinating and
rewarding experience. Et cetera, et cetera, blah, blah, blah, blah. That's not what we want. We want extremely coincise
responses without anyBS, any time loss, any of that. And the reason a website does
something like this is so that they keep you more on the website and
you watch the ads. However, that's not
what GPT is doing. Chgpt's objective is to give you the more value in the least
time possible. Therefore, it's the best way of studying or asking for advice
on the Internet. To expand on a prompt like this, let's create a new chat. And this time let's take, for example, a language. Let's go ahead and ask it, what's the difference between
yes and no in Italian? And then we're going
to expand on it. We provide me two examples. So I enter it and it
says yes means yes. No means no. Here
are two example. Do you want a coffee?
Yes, Thank you. Are you hungry? No. I just ate. So you've just gotten
the idea of what yes and no means in Italian
with a simple bran. Let's now talk about the
third way of basic prompting. This type of prompting, you
have a basic information. Let's say for example,
that in this case, I have received an e mail
from someone which is interested in buying
my editing services and I am a video editor. What I can do is paste the e mail that
I've received where someone asks me to get
on a call to sort out some business. So I'm telling
GPT reply to this e mail, politely accept the offer. Say that I am free on
Tuesday 6-630 UTC. And also convert the time
to the EST time zone, which is his time zone
because he's in America. And now GPT is going to vary. As you can see, it left a placeholder to add the
client's name, and it says, we are free on this time, which is this time for you in EST time zone.
And then again, polite, please let
me know if this time suits you or if you have
an alternative suggestion. And this is the last basic
prompting technique, which is simply asking GPT
to write stuff for us. So I'll type write a five underbods article on
online courses marketplaces. And it is going to give
it a title, Paragraphs, as you can see it
as a conclusion and an introduction,
which is perfect. It mentions a couple
of platforms, like Demi Corsera doesn't
mention skillshare, which is really big. The issue may be
that this GPT is from 2022, which is pretty old, so it is not really
updated to date. However, this is how
you write articles, essays, subjects. And from the next lesson on, we are going to talk about specific strategies
which are going to enhance your results in Chap. The best
has yet to come.
5. Advanced prompt engineering explained, and first technique: All right, we're
going to analyze over ten techniques
which will allow you to announce your prompting
to the next level. We'll start from easier methods and we'll move on to
more advanced methods. But first, what's
prompt Engineering? Well, a prompt is a
set of instructions. As we have said in
the last lesson, prompt Engineering
is making sure that our set of instructions
improve every time. And to do this, we use a
variety of techniques. And the main reason this is important is because
it will allow you in the long term to save time while asking information
to Chachipeti. Because the thing that
you need to know is that Chachipeti knows
basically everything. So it's up to you to ask
the right questions in the right way so that ChachipT can give you
an appropriate answer. Usually when people
type a prompt, they only type the
ultimate goal. So for example, it may be
write a 500 word article or tell me the five
key points which are necessary to study
the Ottoman Empire. So what I do is, for example, if I type a prompt like correct
the sentence I learning, prompt engineering,
and I send it, I have only talked about the
goal that I want to achieve. And in fact, GP is going
to simply reply with, I am learning prompt
engineering corrected. Okay. However, this
response isn't complete as it's lacking
some information. For example one, what
did I type wrong? I don't really know
what I've typed wrong. Now in this case I
can clearly see it because I forgot M and
misspelled engineering. Which you can see visually because the sentence
was a short one. But if I'm correcting
a long E mail, I would like GPT to give me more detail on why he
corrected what it corrected, so that I can improve from my mistakes on
my own next time. And also make sure that GPT
is not doing the wrong thing. Because sometimes even in
math and physics problems, sometimes GPT is slightly wrong. It does have wrong calculation, it never fills a number. If I ask eight times three, it's always going to be 24. It doesn't do mathematical
errors in that way, but sometimes it misunderstands a formula and stuff like that. So I want to make sure to
create an extensive prompt, which I will be then able
to reuse in the future. Or at least copy
the style so that I will always get the responses
that I want in one try, instead of going back
and forth with GPT. The first way of improving your prompts is
extremely simple, and I will provide a
couple of examples. In this lesson, I will simply go ahead and add context
and examples. Let's start with
the context first. Instead of saying correct the sentence I'm learning,
prompt engineering. I will go ahead and say correct
the sentence and type it. And then consider
that I'm a student, which is to create an essay. And I am not sure if this
sentence is grammatically correct so that I am directly
addressing my problem. So as you can see, it has
corrected my sentence and given an explanation
considering that I'm a student. Gpt said, this version maintains the present tense by adding
the word M as you can see, which wasn't present in the first version
of the sentence. Let's go ahead and watch
this content adding methodology from another
more advanced perspective. Let's go ahead and say
that I want to create a text inviting my
neighbors to a barbecue. Let's copy this
idea from open AI. As you can see if I typing
this simple message, write a shortens we text message inviting my neighbor
to barbecue. I'm going to get this really
general and broad answer. However, let's go ahead and improve this prompt
even further. So I've gone ahead and kept the first sentence
the same as before. And then I've typed, the
barbecue will be ousted at my house at 01:00 P.M.
on Sunday. Explain that. The weather looks
like it's going to be sunny and it's a perfect
opportunity to me. I've already purchased the
meat for the barbecue, asking me if you can bring some beverages as
well by doing like. So GPT is going to add way
more detail to the message. Let's go ahead and enter. And as you can see, we went
from a GPT vibe message. We are throwing up the grills this weekend and we'll love to have you join us for a fun
barbecue gathering good food, good vibes, and good
company guaranteed. Looking forward to catching up, which sounds free, GPT. However, in the second part
we have we are roasting a barbecue at my place
this Sunday at 01:00 P.M. and the weather
forecast looks sunny, which is way more specific and provides actual
information to the other person instead
of would love to have you join us for a
fun barbecue gathering, which is of course the case. Otherwise you
wouldn't be messaging your neighbors anyway. Perfect for grilling. I've got the meat covered. Would you be able to
bring some beverages? And then it says, it's going
to be a great chance for us to catch up and enjoy
some good food together. Let me know if you
can making and then I would probably remove this
part, But as you can see, we have already an
improved response just by adding some context. Now instead of adding context, let's add an example. To do this, let's
create a new chat. Let's take once again
the language example. If going to add and typed a prompt, which
includes example. In this case it is correct, the sentence following the
structure of the example. I love GPT because and here's
the reason I've typed. So if I out, I would love
to eat pizza every day. It's going to transform into I would love to eat
pizza every day because correct spelling of
pizza becomes pizza and capitalized day to day for
consistent capitalization. And as you can see, it has
completely copied and pasted the structure of my prompt in the response which
is what we wanted. Now in this lesson, we've talked about the first
bit of prompt engineering. This is just a tip of the Asper. Now we know out what
context and examples. In the next lesson, we're
going to get on 01.2 shot, prompting. I'll see you there.
6. Zero, One and few shot prompting: We're now going to talk
about 01.2 shot prompting. These are common prompt
engineering techniques which we have also used
in the previous lessons. We are now here to define what they are and
which ones are better. Zero shot prompting
is when we ask GPT something in a more
conversational way, providing actually zero context. And GPT has to respond with some information
that it doesn't have. I'll give you an example because right now it's a bit abstract. Let's give a practical example
of zero shot prompting. So I'll go ahead and type the bas ket ball has
a lot of weight. What's the sentiment
of the sentence? As you can see, JGPs going to simply reply
that the statement is neutral because it's just indicating that the ball
is heavy or substantial. There's no clear positive
or negative sentiment as staged to the statement. However, the reason GPS
replied like that is because I have zero shot prompted the sentence and it has
given me a wrong answer. Inside char GPT, there
is enough information to understand that the
basketball has a lot of weight is a
negative sentence. And the reason for that is if a basketball has
a lot of weight, it doesn't bounce well. However, ChargiPT seems not to recognize this in this
specific sentence. So if I go ahead and type, what are the consequences for a basketball having
a lot of weight. And Ch Chip replied that it's difficult in
handling fatigue, injures and impact or technique. So if I don't, go ahead and call and paste all
of the things that Ch Chip has just told me about what happens when you
have an heavy basketball. And I then ask GPT, considering the
difficulty in handling fatigue, increased injury risk, and the impact on technique, which is exactly what
Chip has just told me, the basketball has
a lot of weight. What's the sentiment
of this sentence? Gpt is going to
add and reply that the sentiment of the sentence
is generally negative. It implies that the
basketball being avered and the standard regulation weight is problematic and undesirable. And this is one shot
prompting when you give more context to the
large language model, which in this case comes
directly from ChagPT. I've literally created this
paragraph in another chat. And that's the reason
why you should always add more context
to your prompts. If you create a
zero shot prompt, just like this with a sentence, and then you ask a question, whether it may be a
sentiment or something else, it might simply give you
a wrong answer because it does not know precisely where to pick the
information from. Even if I didn't know what are the consequences
for a basketball, for having a lot of weight, and I'd like to know
what's the sentiment, I'd simply ask GPT, then copy and paste
the information. And then simply say, considering all of this
that you've just said, what's the sentiment
of the sentence? And as you can see, Judge
PT is going to give you a precise answer that considers all of the data that you have
previously given him. I want to emphasize that Judge PT sometimes doesn't
know which data to pick. And the reason is simply add short and not precise prompt. If you make your prompt longer, give context that you don't
even need to write yourself, I want to highlight once again, you can simply ask for context
to GPT and then ask you to critically think about the sentiment or
whatever you need. This is one shot prompting, providing context and
then asking a question. When it comes to F shot, prompting is as simple
as adding more context. For example, in this case, instead of only adding what happens to a basketball
when it's heavier, I could also add what happens to a tennis ball or whatever
other ball in any sport. Or I could also add some physical laws which explain what happens
when a ball is heavier. The more context
you provide to GPT, the more precise and accurate
your answer will be. If you provide a short prompt, you're going to have a small and not detailed
nor correct answer. Another way to use
fu shot prompting is to teach something to
the model from scratch. One of the first ideas that
come up to my mind is if you're trying to create
a book for kids or something like that
in a fictional word, you might want to create new words which have no
meaning in real life. To create these, you simply need to provide context
as we did before. However, instead of
providing real context, you're going to
provide fake context. In this case, I could say the verb frondling, I
completely invented. This means jumping up and
down in a happy manner. Provide me an
example of this word being used if I now
go ahead and enter. As you can see, Buzz
couldn't contain his excitement when he heard about the news about
his promotion. He began frondling
around the office. His joy evident in every bounds as you
provide context and information to the large
language model he is going to understand and
apply as soon as you ask. You could also go ahead and type more context
around the verb. You could ask it
to conjugate it to all of the tenses that exist
in the English language, or you could even ask the
large language model, I'm inventing a verb. Please help me come up with
a definition for the verb. You can create a verb together
and then feed once again the description to
GPT and is going to flawlessly implement your verb. I could also guide in
custom instructions, copy it here, save by, of course, enabling
for new charts. I go right here and I ask it
what does frontling mean? And as you can see, he's
going to type like that, my definition and also
a slight add on that. He decided that's so
you can teach char GPT, new stuff from completely
scratch and how to get good responses
which are accurate by using one and fuso prompting.
7. How to use output templates: In this lesson, we're
going to talk about one of the most important
things that you can do, especially when asking
for information to char GPT and it is providing an output
template in your prompt. What is an awful template? It happens really
often that when you ask GPT for something, you're not happy with the way it provides you
certain information. For example, if I want
to go ahead and type in, give me the ten best
books that talk about war and they enter.
As you can see. I've gone ahead and received ten books in a numbered list
as you can clearly see. And there is written
the writer and Azure Small
Description, two lines. But maybe that's not what they
want because for example, the art of war is by Sanzu. Maybe I don't know who Sanzu is. And then it says a
timeless treatise on military strategy
and tactics, offering valuable lessons on leadership planning, and
conflicting resolution. But I've read The
Art of war myself, and this description
doesn't really resonate with what it's
actually in the book. So what I will go ahead
and do is ask for a different output from ChachiPT because I am sure it
has the right information. I just didn't get fed
what I actually wanted. And that's because my
prompt wasn't good enough. So I'll go ahead and give me the five best books that talk about war and follow
the output template. So basically the same prompt. However, I'll add
the output template, title, date, name of the author. Small description
of the author and his achievements in
the military field. Because I want to make sure
that the person I'm learning to was actually someone
that has fought. How is the information provided
in the book, for example, if it's small snippets, quotes multiple chapters,
few and longer chapters. How many pages is the book and what's the average
review on Amazon? I'll go ahead and enter. I've also added include
the Art of Work. So that I've read it and I can verify that the output
template was followed. As you can see, we
have the title, the date, the name
of the author, and then his achievements are. The fact that he was a Chinese military strategist
and philosopher. How is the information
provided in the book? Concise chapters,
each focusing on disparint aspects
of war strategies and tactics, which is true. It provides insights
through short, insightful passages and dotes, which is really, really true. You can find a lot of quotes online that are
taken from the book, and that's how the
information is delivered. Then on many pages, it's of course always
going to say that it varies depending
on the edition, but typically around 100 pages. Then we have the average
review on Amazon. Then we have all quiet
on the western front, which is from a soldier in
World War One, as you can see. Then we go below, there is this book from a
Vietnam War soldier, et cetera, et cetera. As you can see, we have also
average reviews on Amazon. And they could ask
GPT, for example, to sort them out by
the highest reviews, or I could ask them
to sort them out by the shortest or
the longest book, or by the oldest, the newest. As long as I provide a
good output template, I'm going to get the information that I want as I want it. You can either do it
by literally outlining the output template or
explaining it by word. I could have said make sure
to include title, name, a small description, how the
information is provided, just in a normal sentence. However, what I found is that by creating an output
template like this, you're going to get way better
and more outlined results. Let's go ahead and
provide another example. So I go ahead and
type I'm learning prompt engineering
correct sentence providing what I've
written becomes correct snippet because
reason of the correction. And then I'll outline in
a bullet point format all the single corrections that you have done to the sentence. Following the output template, I'll go ahead and enter, and as you can see, I have
a bullet point format where it says this becomes this. Because I is capitalized
to I as the pronoun should always be capitalized in English becomes for
proper grammar, learning becomes learning
for the correct spelling. I've misspelled testing. Engineering becomes engineering
for correct spelling. Once again, as you can see, it has followed the
output template. So now I know what was
corrected in my sentence and why this is really useful when you're correcting of
course, longer essays. It's not that useful when you're correcting
a single sentence. You could probably have seen the difference
between this one and this one with your own eyes without asking for
review like this. But when you're creating
longer says, it's really, really useful to ask to provide a description for each small
thing that is modified. Another really cool
feature that you can use regarding
output templates is asking for CSV files
and table formats. How do you do that in now? Go ahead and Ty you ordered
the following numbers in two different columns
in a table format. One is for odd numbers and
one for the even ones. I'll go ahead and enter,
and as you can see, I have the odd numbers
and the even numbers. I could also ask you to
slightly modify these and say at 4,356.34 And
as you can see, you now have your table. I could also ask
you to order them from the smallest one to the
biggest one in the table. As you can see here,
I didn't provide the right output template by telling GPT to keep
them in the table. I just told them to pick
the numbers from the table. It has given me the
numbers like that. But then after a small
correction here, it has given me the output
template that I wanted. They are now ordered from
the smallest to largest. And I can also ask you
to create a CSV file. As you can see, I'm going to get this CSV file that I
can copy like this. The reason CSV files are useful, which by the way stands for
Comma Separated Values. The reason this is useful is
for creating Excel files. For example, I can copy the data and past
in the spreadsheet, which is going to create two different columns
that I can use. That's how to use
output templates. You can use them in a variety
of way, it's infinite. You can use them for
tables, bullet points. It's really useful for outlining or highlighting
important information. I will now see you in the
next lesson where we'll talk about roles and
perspective prompting.
8. Roles and perspective prompting explained in depth: In this lesson, we
will talk about giving roles to GPT
and perspective, prompting the air techniques
that go end in end. So we'll treat them
both in this lesson. Let's start by giving roles. Let's go ahead and use one of the examples that
we've used earlier. But by implementing the
roles, let's go, for example, with the grammatical correction example that we've
used multiple times. This is a prong
that I usually use and it allows me to
prompt certain charts in a way so that I
can have a list of charts that I can use when
I have specific needs. And it basically
works so that you give a prompt at the start
of the chart which says, in this case, act as
a language tutor. From now on in this chart, I'm going to send you e mails, messages, sentences, essays. I ask you to act as a
language tutor and provide me with the grammatically
correct version, explanation and
eventual vocabulary suggestions which would
elevate my writing style. Once I send this message, GP is going to acknowledge this, and as you can see, it's
language tutor assistance. I can even modify the name
to correct my e mails. For example, if I'll use this to correct my e
mails grammatically. And I can now go
ahead and type in my message without
having to prompt it. I can simply type in the
message, copy and paste. Or for example, let's use the sentence that we
have used until now. So if I enter, that is going to
correct the sentence for clarity and
grammatically accuracy. And as you can see, it
corrects it and it explains. We pull a points
and as you can see, it has acted as a
language tutor. I missed the contraction of IM, which is the appropriate
subject pronoun for yourself. That's what the English
teacher would say. Learning is the present
of the verb learn, indicating an ongoing action. About is a preposition indicating the subject
or aerial study prompt. Engineering is capitalized
as it seems to be a specific field of
topic you're studying. And as you can see, it also
has an encouragement message. And the main point
that I want you to take is that you
can give any role to Ch GPT and it is going to enhance its performance
every time. If I go on a new
shot, for example, let's go ahead and
type the same prong in one shot I will not ask it
to follow a certain rule, and in another one I
will so that we can confront the
different responses. So on one side I'll say
act as a business coach. The sales of my e commerce
stores have lowered recently. How can I enhance them? So if you confront
the two responses, let's start with the
business coach one. As you can see, the
first thing that it tells you is to analyze
customer behavior. Utilize analytics tools to understand customer
behaviors on your website. Look for patterns in browsing, shopping cartapandument,
and purchase history. Understanding your customer's
preferences and pain points can guide you in tailoring your offerings and
marketing strategies. As you can see, it's an extremely precise
and detailed answer which tells you to first understand what's
the real problem. Look for the patterns
in browsing, what's the shopping
cart abandonment, trade, and the purchase history. All these kind of
starts which will help you understand
what's wrong. Then it says optimize
user experience so that once you have understood what's the problem
on your website, you can fix it just like so. Then it talks about
marketing CO, so once you have analyzed what's wrong with your
website and you fix it, it's time to expand your reach. So you start going with
content marketing, search engine optimization,
high quality, relevant content, et
cetera, et cetera. It talks about target audience and search
engine optimization. Then we have the idea
of discounts and incentives to enhance
the amount of sales. And it also talks about loyalty programs so that it incentivize
people to buy more. And then it also talks about
limited time offers and free shipping and exclusive
deals that create a sense of urgency which motivates
the customer to take action. And then it talks about
customer engagement. So once again, analytics,
review and personalization. Then reviewing the
pricing strategy compares to competitors, but adjusting it to remain competitive without
compromising profitability. Then it talks about
customer support and to offer multiple channels for communication such as live chat, email, phone support, et cetera. And then it also
talks about KPIs, Key Performance Indicators
such as conversion rates, average order value, and
customer acquisition costs. So that you can understand what you're doing prong and
what you're doing right. And as you can see, this is an extremely
well studied response crafted step by step to
improve your eco store. As soon as you read
this message instead, if we go on the message here without asking to be
a business coach, as you can see, the response
is a bit more broad and less defined
and less precise. Here, it talks about optimizing your website loads quickly, both desktop and mobile, all these kind of basic stuff. But once again, it doesn't talk about analyzing
your customer data. So where should I even
optimize my website? Am I going to improve my sales just by making it load quicker? That's not really
a precise answer. And then if we go below, improve product descriptions
and images because it has a significant
difference in converting visitors
into customers, it says, I should invest in professional
photography and writing compelling product
description that highlight benefits and features, which is something that
surely is important. But maybe my problem is
getting people on my website. I haven't tracked the data, so I can't really
know if my problem is the conversion rate or
attracting people on my website. And then it talks right away
about customer service. However, this is something that you should do
more at the end. As you can see, it talks about e mail marketing after
customer service. But it's useless to prepare
a customer service if you don't even have clients in the first place that
visit your website. Then it talks about
social media content. So once again we have
client acquisition which is of course necessary. Seo referral program,
which is a good idea actually that the business
coach hadn't yet formulated. And then it talks about analyzed data metrics and then
it says stay competitive. But as you can see, let's
compare, for example, this point with the one provided by the business
coach right here. So if I look at this one first, as you can see it says, keep your competitor and
industry trends continuously innovate and adapt your
strategies to stay ahead of the curve and meet the evolving
needs of your customer. Well, this sentence doesn't
really tell me much. I already know that I
have to follow trends. Yes, that's true, but it's
an abstract response. Like it doesn't tell me
that to follow the trends, I should go on Tiktok and a certain ash tag or go on Instagram and search
for a certain account. It's just really,
really broad and just saying make sure
to innovate and adapt, but it doesn't tell you how. Instead, if you go to
the business coach, it's way more precise. As you can see, it says, review your pricing strategy. Action step, review
your pricing strategy. It's already an action step
in comparison to competitors. So you can search competitors of your store which sell the
same product or similar. And adjust pricing where
necessary to remain competitive without
compromising profitability. So it also gives you guidelines on how to review your
pricing strategy. Consider dynamic pricing
strategies based on demand, season, and market trends. So not only it gives
some guidelines regarding profitability
and competitor research, but also regarding the amount of demand the season we are in, and the market trends, which is a way more
precise answer than this one, as you can see. And that's why you should
always give a role to GPT if it's relevant
to what you're asking. Of course, this
case it was because I was asking how to boost
my e commerce sales. Now let's move onto
perspective prompting. At the start of this lesson, I told you that
perspective prompting and roles go hand in hand. And that's because with
perspective prompting, you ask Ch GPT to impersonate
a certain person, which is similar to the role. However, when you
ask for a role, you're asking it to act as a business coach or a language
tutor, something general. Instead, when it comes to
perspective prompting, you're asking him to be
someone way more specific. For example, I have received an e mail and I'm going to
ask him to impersonate me, or impersonate my client, or to impersonate the one
that sent the e mail. I'll give you once again, a practical example so that you can understand right away. Let's go ahead now and type an hypothetical
scenario where I am a video editor and there is someone that is interested
in my services. So let's play out
this scenario in GPT so that I can get an e mail
to respond to this guy. So I'll go ahead and
type the first prompt. There is a guy named Bob
that has typed to me because he's interested in my
services as a video editor. If this was a real scenario, I would simply copy
and paste the e mail that Bob sent so that I
don't have to type anything, that I can just let
GPT do his job. However, since this is
a fictional scenario, I'll go ahead and type this. So we'll go ahead
and say a reply to him saying that
I'm interested. And I would love to set up a call to discuss
the opportunity. Further, I have a spot on
Monday from 06:00 to 06:30 P.M. Italian time. Respond politely and
when you say the time, also at the time
in his time zone, which is American EST. So now I'll go ahead and enter. And as you can see, Hey Bob, thanks a lot for reaching out. I'm thrilled to hear that you're interested in my video
editing services. I'm definitely interested in discussing the
opportunity further. How about we set up a call? I have a spot available on Monday six to 06:30
P.M. Italian time, which would be this
time in ESD in the US. I work for you. Looking
forward to chatting you. As you can see, I
have completely automated my e mail response just by outlining the context. Or I could have done it simply by copy and pasting the
e mail, the dissent. And that's the power of
perspective prompting. You can use it in any way to chat or also for active
problem solving. So let's go ahead and take
this scenario a step further. I've gone ahead and typed, I've sent the e mail to Bob, and he has seen the e mail but didn't reply for seven days. What could be the
possible problems that caused him not to reply, even if he was interested
in the first place? Act as if you were Bob. Which problems did you find? So go ahead and enter. And as you can see, he's
going to say, as Bob, there could be several reasons why I didn't reply
to your E mail, even though I was
initially interested. Busy schedule, perhaps I have other tasks and projects and
I'm postponing this e mail. I forgot technical issues. As you can see, there might have been some technical issues with the e mail account
considering options. So he's considering
various video editors at the same time
or changed plans. The priorities or requirements for the projects
might have shifted, leading me to delay
responding until I have a clearer
idea of what I need. Regardless of the reason, it's important for me as
Bob to follow up with Buzz. So I'll now take
this again a step further and say now act as Baz. Considering these
possible reasons, what's the next step
you would take? So go ahead and enter. And as you can see, it provides me with four things
that ChachiPD would do. The first one is following up politely in the
follow up e mail. It would also offer
alternative ways to communicate if e mail
isn't convenient for him. For example, I could suggest catalin a call at a
different time or even communicating via
different platform such as phone call or Peterson. I would also be
patient and understand that maybe as a valid
reason for not responding, you're getting GPT's L in order to understand what problems could have caused
him to not respond. What should you do and basically problem
solve this scenario. This can work in any
kind of situation, any kind of communication, whether you're a
student with a teacher, if you're at work
with a colleague, a video editor with his
clients, whatever it may be. Asking always for
Cha GPT is always a good idea because
it might provide insights that you
haven't thought about. For example, I've had this
scenario happen to me once, and I haven't thought about the possibility of technical
issues with the e mail. In fact, you could try to deliver the message in
another way, for example, via Instagram or Twitter, DM, and you're going to have a higher chance
of responding. Chagpt's opinion is always extremely good because
it's completely rational. There are no emotions
involved with GPT. And when it comes to
professional scenarios, acting rationally is one of the key things that
you should do. I'll see you in the next lesson, where I'll talk
about personality prompting and descriptors.
9. Personality prompting, descriptors and internet access: Let's now talk about adding
personality to your prompts. This is extremely useful when it comes to writing
with Chachipeti, whether it be blogs, articles, essays, Youtube
scripts, whatever. Adding personality is
extremely important to make your content not
sound like a robot, because Chagipti tends to make anything sound
like a robot. I'll give you an example.
Welcome to the bad zone, where curiosity
meets creativity. From thrilling adventures
to insightful discussions. This is where inspiration
takes flight. You get ready to explore, learn, and ignite
your imagination. This is bad zone. I think this is one of the worst suitable interests
I've ever seen. And that's why you need
to add personality to your prompts to avoid
this kind of general, boring kind of content
that GPT tends to create with these GPT
words like thrilling, or sometimes it uses catalyst. All this kind of stuff that people can instantly
tell, it's GPT. How do you add personality? Well, personality by adding
a style and descriptors. Let's start by explaining
how to add style. To create a style,
we're going to create an e mail from a newsletter
that we have just invented. And we have a
newsletter that talks about online course
marketplaces, auto monetize them, auto,
make money with them. So we'll go ahead and
create this e mail by first typing the content
that I want to outline. I've gone ahead and
create an e mail that talks about auto
monetized courses on online courses marketplaces
such as Udemy and Skillshare keep it relatively short and do not
include a final CTA. Now here's where the
style comes into place. I'll go ahead and t create an email in the style of and
here I have two options. The first one is to type a name of someone that
has a newsletter. You can type any copywriter
or any person you know. But the way I prefer
to do this is by copy and pasting one of the actual e mails that is
used in the newsletter. So in this case, I will take an e mail from Tate newsletter. I'll go ahead and copy and
paste one of the e mails that was sent in under
head newsletter. And I'll say make sure to copy the spacing vocabulary
style and pacing. I'll go ahead and it send.
And would you look at that? Now we have subject
maximizing profit from all anchor marketplaces. And then we have the email
spaced as and rotates. One, as you can see, 12 lines
per paragraph saying here. And as you can see, there
are short sentences. Earnest, the power of
compelling content craft. Your courses embrace
relentless innovation. Remember, success in
this domain hinges not only on the quality
of your content, but also on your ability to captivate and engage
your audience. Embark on this no monetization, channel your inner
entrepreneur and seize the boundless
opportunities that lie ahead. Forge ahead with each course, you creates a testament to
your ingenuity and vision. As you can see, it has copied
the style and the spacing, pacing of the e mail
and the vocabulary. Now let's also add
something else. Instead of making an
e mail like this, I'll go in a new chat. Copy the same prompted before. And now I will add
some descriptors. Instead of adding a style of another copywriter or someone
that has written an E mail, I'll go ahead and type make it sarcastic with
simple vocabulary. As you can see, this is
a descriptor, Sarcastic, which means slightly funny
and a bit provocative. As you can see, it
starts so you ready? Cache knowledge, Huh? We're about to dive
into the wild world of it in courses on
platforms like them, in Skillshare, without
losing your sanity. Know your niche.
Pick a topic that's hotter than a yellow
pena in a heat wave. Think how to train your
cats in opera level niche. As you can see, it's a funny
way to make an e mail. It's not serious at all. Your course title should scream, click me louder than a
toddler in a candy store. None of that introduction
to basic stuff, nonsense. Content is king or queen. Make sure that your content is chooser than a ripe watermelon. Get your Ssl on social media, e mails, and carrier pigeons. Use whatever you can
to spread the word. Beg borrow or steal
those five star reviews. They're like gold nuggets in the wild west of
online learning. Cheers to your future fortune, or at least a decent site. So as you can see, it is given a fun angle to this e mail even though
the content is still there. As you can see, it
has still given the insights that you need
to create online courses. So you have to know your niche titled content engagement,
the marketing, the reviews, updating
your courses, the pricing, all of that stuff. The information is still there. It's just packaged in
a way that is funnier, sarcastic, and not that serious. Although I want to repeat it, the content of the e mail
is still there and true. Asking for something
and then adding the sarcastic descriptor will add this kind of
twist which will make your content way more enjoyable and digestible
for the audience. If I go ahead and
type the same prompt, but without any style or descriptors, this
is what we get. You're going to find some GPT words as I like to call them, which are some
words that are used in basically every GPT response, such as tapping into
it's usually used. Compelling is used
a lot by AI Foster. As you can see, these are really basic and
unpersonalized answer. That's the reason why by using the prong techniques
that I'm showing you, you can answer your results and stand
out from the competition. Because nowadays a
lot of people are using GPT to write their
newsletters or their articles, blogs, or at least
to help them out. But the thing is that
if you can actually use prompts which would generate
you a really good result, you have already a really
good starting point. Personality prompting
is really important because it allows you to
stand out from anyone else. It allows you to give
custom instructions to GPT regarding the style and the way it should say your
certain information. And that's what's going to help you stand out from anyone else. Because while anyone
else has all of these basic descriptions and newsletters and all the stuff, you can still create some
good results in GPT. If you give the right
descriptors and stylizations, I highly suggest you to
try a couple of things. Pick a certain style that
really works for you and then use it every time you're creating a newsletter
or something similar. And that's going
to help you a lot. I will end this lesson
by talking about how to access any information on
the Internet with GPT 3.5 And the reason I'll do
this is because we have already done it 2 minutes ago. But I'll explain all
these works basically. Let's take, for example, an event that has happened
after January 2022, for example, the Ukrainian
versus Russian work. If I go ahead and type, give me a brief summary of the events that
are happening in the Russian Ukrainian work.
Go ahead and send it. And as you can see, it says that this last update is
in January 2022, but it cannot provide
specific details because it doesn't have access
to real time information. Therefore, we are
going to fix this. To do this, I'll
go ahead and type, there has been an
escalation in conflict. I'll go ahead and copy
the Wikipedia page. Please. Go ahead
and summarize it. Gpt is going to
say that I cannot access or browse the Internet,
including Wikipedia. But the thing is
I'll go on Wikipedia and simply just
copy all of this. I'll copy it until here
because I already need the history and I'll
go ahead and paste it. And GPT is going to now
summarize everything. Is now thinking and getting
the information in order. As you can see, it's
going to start typing. As you can see, it's
now saying that everything started
in 2014 and then in 2022 there has been a full scale invasion of Ukraine and it has kept
going throughout 2020. 3.20 24. So we have now updated to date information regarding the Russian
Ukrainian war. And I could do this with
everything happening. May be some beef
from some tubers. I could simply download
a Youtube video, transcribe it with some website that you can find everywhere, and then just copy and paste. Transcribe and ask GPT
to summarize the beef, or I could do this with a book that has been released
after January 2022. I would simply copy
the summary in the chat and ask Ch
GPT to summarize it. In that way, I can access real time information
from the Internet even if I have GPT 3.5 Now we know
how to access the Internet. Give personality to our
prompts at descriptors. I'll see you in the next lesson where we're going to finish our prompt engineering
lesson by talking about laser prompting
and prompt leaking.
10. Prompt Leaking and Laser Prompting: In this lesson, we will talk about two prompting techniques. The first one will increase the accuracy of the content that you create thanks to GPT. Meanwhile, the second
one will increase the rate at which you can learn prompt engineering to
create better prompts every time you
interact with ChgPT. The first one is
laser prompting. Laser prompting allows you to create specific
answers which aren't general and allow you to focus on specific parts
of a certain topic. That's why it's called
laser prompting because a laser is really strong in a really
small area or surface. So let's go ahead and
create some examples to make this less abstract and give you actionable
ways to use this. Let's say that for example, I asked PT to create a 500 word article about the
evolution of technology. This is a broad
topic and I could include more detail as we have
done in the first lesson, which might be, as an example, include the invention
of the smartphone. However, the way laser
prompting works, Instead of adding context for the large language model
to assess certain topics, we are going to reduce
certain topics. So I'll try and explain
it in a new chat. In this specific case, instead of saying include the invention of the
smartphone and the first PC, I could also create a
negative prompt where I say, make sure to start only from the first industrial
revolution and ignore agricultural and physical innovation
and inventions, rather focused on digital
inventions and innovations. So if I go ahead and type these, as you can see,
it's going to start from the first
Industrial Revolution. So as you can see, it's not
going to start like here where it just talked about
the evolution of technology. It starts from the prehistory where it talks about
rudimentary tools. Because GPT simply
interpretates technology as the advancement of
also physical tools and inventions,
not only digital. If I want only digital stuff, I have to make it clear
in the first prong. As you can see right here, it already starts
with 18th century. Here it gives an introduction
to the digital age, already from the second
Industrial Revolution. As you can see, we have the telegraph and the
telephone right away. We have networks. We already get into the
computer in the 20th century. Then it talks about electric
computer from room sites mo, to leak personal devices. Talks about the main companies
that have created them. And as you can see, it
limits his knowledge to only the technical
inventions which are related to computers
and everything digital. So smartphones and
stuff like that, we get to mobile phones, social media, everything which is digital and not physical. The main reason this
prompting technique is important is because GPT
contains so much knowledge, which makes it hard for the large language model to understand Which knowledge
specifically do you want? In this case, the problem isn't like in zero
shot prompting, which is the fact that the
answers aren't correct. Because in this case,
this article is correct, all of the information is right. However, it doesn't really link together and it's
not focused at all. Gpd had to manage the whole technological
advancement created by mankind in 500 words, from the prehistorical to today, which is basically impossible. Which may GPT come up with a broad article which
talks about prehistory, and then it talks about the
ancient Egypt and then Rome. And it starts to skip until
the industrial revolution, just mentioning
the things before. And then it skips once again, two centuries, in five lines, because it has to go fast. Which ends up in an article
that says almost nothing. Because it's almost
impossible to represent all the achievements created by the human race
in a matter of 500 words. Another way to do
this is instead of asking to start from the
first industrial revolution, I could ask it to
start from 2020. So now it has only two years
to create an article on, because the last update
of ChGPT was in 2022. And in this case going
to already talk about Covid and it's going to
talk about five G network. And we have zoom teams met,
instant messaging platforms. We have AI Machine
Learning, LP's, OpenIchGPT. Unprecedented
innovation. Of course, because we're talking about AI. We have block chain technology. We have Bitcoin,
Etherium finance. And as you can see,
it has talked about extremely niche stuff which it didn't even mentioned
in the first article. Stuff like block chain
and all the platform, et cetera today, Netflix, Disney, Amazon Prime Video. As you can see, it has
managed to space into more detail from
the entertainment to the communication industry, all of this stuff, because it had only two years to cover, which is still a lot. Let's not talk about
prompt leaking. For prompt leaking, I'm
not actually going to go into a new hot but
into the old chats. The reason for this is prompt leaking is a technique
which allows you to take an output that
you have received from one of your chats
that you actually liked. For example, let's
go right here into the boost E commerce
sales, all right? And we've gotten this
amazing response, as we've said in
the last lesson, this was a good response which
gave me actionable steps. And all of this
stuff, prompt leaking is divided into two parts. The first one is
complementing GPT. I'll go ahead and say
your response was extremely compliment. Extremely compliment. I love the specific language that you have used,
other compliment, once you have complemented the response of GPT
four or five times, GPT is going to understand that response was actually
good and you really liked it. So now you're going to ask it, what's a problem that I
could have asked you to get a similar
response in one try? In this case I had actually
used one try only. But this is really useful
where you're trying to, where you have gone through
a long process with GPT, of back and forth and you just want to get it
done next time quicker. So you'll go ahead
and send this. So as you can see, we have, as a business coach, so we have the role
of concrete steps. So actionable steps
can I take to improve the sales performance
of my struggling income? Restores. In this case, GPT suggests me not to say that the sales of my commerce
stores have lowered, and how can I announce them? It simply says,
how can I improve the sales performance of my
struggling e commerce stores? Go ahead, copy and check if GPT said the right thing by
copying it in a new hot. As you can see, the two
responses are extremely similar. If I look at the first one, we have comprehensive analysis and here we have analyzing
customer behavior. Same thing, user experience, we have it right here. It's the third instead
of the second one, but basically the same thing. Content marketing and SEO, which right here is the seventh, but basically the same thing. Here we have customer
understanding and here we have the same thing
with customer engagement. Then we have performance
tracking and optimization, which is the same as here
monitor and iterate with KPI's. As you can see,
the responses are extremely similar because we have used the prompt that
ChchiPT suggested us to use, and that's why prompt leaking
is extremely important. Let's summarize this course. We have analyzed how
to create content with Ch TPT and we have started from adding extra content and examples to laser prompting, to give roles, to
ask GPT to provide more specific and
competent responses to contextual prompting, which allowed Ch PT to gain internet access and obtain every single information
on the internet, to even laser prompting, which allows us to control specific parts of
our topics in order to create a more niche down article blog script,
whatever we need. We have then talked about
zero shot prompting, one and few in order to obtain the maximum accuracy
from our prompts. We have also analyzed output
templates in order to obtain the information
in the way we want with whether
it may be tables, CSV files, bullet points, text paragraphs, whatever we want From an outlining
perspective. We have also understood how to tailor a GPT to our
style with perspective prompting adding personality and descriptors in order to ask
GPT to have our own voice, or the voice of
someone that we know, by providing samples or
descriptors which will allow GPT to create content which
is framed in a certain way. For example, we have
used the example of the Sarcastic descriptor
which allows you to create Sarcastic text. Then we have also talked
about the best way for you to improve your
prompting techniques, which is asking Cha GPT itself, how was he so good at
providing me an answer so that I can then recreate
it in one single prompt. You will find the prompt
leaking prompt in the word document as well as three prompts that we
are now going to craft. The first one is going to be for Instagram description, SEO. The second one is going to
be for Youtube descriptions, titles, search
engine optimization. And the third one is going to be a prompt which will allow
you to script books. I'll see you in
the next lessons, where we're going to take our GBD masterclass
into practice.
11. Using ChatGPT For actionable Instagram BIO SEO: We are now in the second part of the course where
we're going to create a prompt list which is
going to help us in social media search
engine optimization, both for Instagram and Youtube. You can find these prompts
that we're about to create in a Word document in
the document section, you can copy them, tweak them. I'll show you the
framework now to create these prompts by using all the information that
they have just taught you. First, let's analyze
our winning description of a profile on Instagram. Looks like I've
taken five examples. I'm not going to take more
because they're all the same. As you can see, a
good description on Instagram is divided
into three sections. The first section is a
simple line and it's just describing what the
account or the agency does. As you can see right
here, it's well, business owners build and
monetize personal brands and then there is the social section which
is the second one, generated 1 billion
client views. And then there is
a call to action. In this case it's download free video templates right here. It might also be a direction to the paid website right away. But I'm sure if I
click on this link, they're going to either ask for my e mail so that
I can subscribe to their newsletter
so that they can funnel me into their
paid programs. Or it's a comprehensive website which includes all
the free content, paid content, and paid services. Probably. Then we have
Nato Zeal, for example. Same structure, what do we do? Social proof, pre value or
redirection to a funnel. As you can see, there
is the link three here. When you click, for example,
on this link three, you get redirected to two free resources and then you get funneled into
their discord server. Then we have another
content creation page. As you can see, who are the two guys helping you
grow your personal brand? 0-100 again in less
than two months. So here is the social perforce
again, as you can see, there is free value once again
and their paid program or free value that finels into the newsletter and then finel
you back into the program. But if I also change niche, for example, in this case I
have a motivational niche. As you can see, it's
voz en right here. As you can see, the structure of the description is still
the same. What do we do? Social proof and how to learn
our secrets are basically funneling into paid programs or free values, et
cetera, et cetera. As you can see, for example, the Vosen formula which is $100 and then there's
a lot of offers. I'm not going to go too deep about all the offers
that Vos and, but as you can see,
you're getting funneled into different
options that they have. There is a mentorship, one on one calls, et cetera, et cetera. And that's the point
of the call to action. And then we also
have bold motivation still in the motivational niche, unlockraal potential.
So what do they do? Motivation once
again, social proof, mastering, editing, and socials. They have a lot of
followers here on you to probably even Tiktok and
then join our community. Again, you get funneled
in their course. We have now the structure of
a good Esagram description. However, what does
it take to make it search engine optimized? Let's look, for example, at Miyato. What is he aiming at? He has an agency and he
wants to give value and sell services to people
which can afford him. So what does he
talk about is bao. He claims that he's
going to build and monetize your personal
brand for business owners. Then in the social proof part, he talks about 1
billion client views, which is a really big number. And then this little part
is not really important, he just talks about how
he's going to give you free vita templates,
but as you can see, there are a lot of
keywords right here which are business owners, build, monetize, personal
brand, 1 billion video. Every time someone searches
up one of those keywords, he's going to be in the top
results as he has a lot of followers and he has those
keywords in his bio. If we go, for example, on this page as you can see, what are the keywords. Once again, personal brand grow, 100 K, grow once again. If fluke, for example, Vosen, he has
motivation motivated. So every time someone
looks up for motivation, which is what Boson
is known for, he is going to pop up
as the first result. And then if we
look, for example, at Bold, they have more
of a editing approach. As you can see, if they
go on their website, they're trying to sell
you a course where they teach you how to
edit specifically. As you can see, what
they're trying to do is put master editing and socials in their boo
more than motivation. As you can see, you can't
find motivation right here. You can only find
it in their name. And that's because they want
to be found for the editing. More than the
motivation part because their main selling point
is that they're good at editing and they went
viral over and over, as you can see from the views. So now that we have an introduction of
winning Eastern bias, let's create a prompt which will perfectly create a
description for us. First, let's start
with the goal. I want to create an
Instagram description full. The description should be
divided in three parts. What do I do, social
proof, and three B. Then I'll also add, here is one example from
a content marketer, and I'm going to paste Devin Atos description as
you can see right here. Then I also in here's one example from a
motivational page and type in the Vos Ens bio. I'll finish it by saying, if you have understood
the template, I will now send you the
details of my account so that you can tailor the template
to my niche and services. As you can see, GPTSaid. Sure, I understand the template. Please go ahead and provide
the details of your account, including your
niche and services, so that I can tailor the
template accordingly to you. And now we'll go ahead
and I am a video editor, which has generated over
50 K views for my clients. And as you can see here
is what we have got. Elevating your storytelling
through video editing, over 50 K views generated
for our clients. Unlock free editing
tips and traits. Below here, you can have your
website or contact info. And what should you do here? Well, you can either create a simple ten page book or a
simple 30 minute tutorial. A free course,
something really small, but that will provide
value to your followers. It's not art at all,
but as you can see, we have a good description. Let's now try for another niche. And now I'll go ahead and I am a motivational page that aims to motivate people into
becoming better. Let's see what GPT comes up to. As you can see, uniting
transformation for better lives. Inspiring over 50 K followers
to reach their potential. Diving into our
empowering secrets below your website,
Contact info. Now let's try and change
our subject once again. We are no longer a
motivational page. Let's say that now I am
a wedding photographer. However, I'm pretty
new in the industry and I only have, let's say, six months of experience, which doesn't really
look too great in an Instagram bio because you usually want to
exceed confidence. And six months is
not even a year, so people might not
take you seriously. So how do you improve that? Well, let's say you have
two clients per month, which is extremely possible
for a wedding photographer, which turns out to
be around 12 clients in the whole six month period. What will I go ahead and do? Instead of saying with
six months of experience, I will say over
ten P clients and then show their
testimonials in my website. I will go ahead and
type these to GPT so that it can create
a good description. I'll type, let's change now. I am a wedding photographer
with over ten Ep clients and their testimonials are on my website. I'll go
ahead and do that. And as you can see, capturing timeless moments on
your special day, delighting over ten plus P
clients with lasting memories. And then there is the
section of the testimonials, however I go ahead and type, that's a good description. However, search
engine optimize it by including more words relating
to wedding and photography. And as you can see, wedding photography expert trusted
stunning memories, ep couples finished
the last line from focusing on the testimonials
to my previous work which is on the website and
as you can see there is view our portfolio on
our website if I want, I could also change the last
sentence to something like, instead of directing
to the portfolio, direct them on a one
on one free call. And as you can see, book your free one on one
consultation today. That's also a really good call to action to add.
That's perfect. We now know from scratch how to create a
good, a grand bayo. You will find this prompt right here in the word document. I highly suggest you to
modify these examples to something more related
to your niche so that GPT can understand better
what you're talking about. As you can see, it struggled a bit with photography
because it had nothing to do with the first descriptions
that we gave him. However, with content
and motivation, it was way better
from the start. That's because I gave him descriptions regarding
content and motivation. Therefore, I suggest you
to tailor the examples. It's going to take
your second to find the best descriptions
in your niche. You just have to type
wedding photography and you're surely going to find them in the first results. So that's how you create
Instagram descriptions. And I will see you in the next lesson where we are going to talk about Youtube search
engine optimization.
12. Using ChatGPT and VidIQ For actionable Youtube titles and descriptions SEO: We are on GPT and we are
now going to look at how to search engine optimize our titles and descriptions
for Youtube videos. We're going to use GPT only to write the description
and the videos, but to do the research on what actually works
and what doesn't, we're going to use
another tool which is VQ. Vdq offers you way more options. If you upgrade to
their premium version, you can find some affiliate
links on the web, on Youtube, which
will allow you to use VdQ premium for one month for only $1 otherwise it's
going to be 50 per month. Regardless, we can still use the basic version
to do basic words, research tags, and all that
kind of stuff. Here we go. I am currently in an
incognito window so that I get completely new
recommendations from the algorithm which aren't influenced by my previous
searches and history. And how does CEO
work on Youtube? Well, there are three things
to take into consideration. The first one is the title, the second one is
the description, and the third one is Stags. However, this video
doesn't even use Stags. I'll later show you
how to create tags. But it's a really quick and
not really important process when creating a video. It's important to look at keyword research because
it's going to allow you to evaluate if your video has the chance of becoming
successful and viewed. Let's talk about the
most important metrics when it comes to
keyword research. The first one is search volume, so I'll go ahead and type
mid journey as an example. As you can see, there
is a search volume of 1,300,000 searches, which is extremely high. And then the second important
metric is the competition. As you can see, the
competition is really low. There are only 35
videos on average, which allows journey to be a
really, really good keyword. Now let's go over and try with
something else, Mr. Best. As you can see, for Mr. Best, there is a really
high search volume, but a really, really
high competition. And when you're trying
to make a video, especially if you
have a small channel, or it's your first video, you shouldn't go in a market with high competition
because the people with more subscribers and better quality videos are
going to stand out better. Let's go back to Journey. As you can see, VDQ says it has a good overall score because it has high search volume
and low competition. So we could go ahead and
make a video journey. However, I am not going to title my video Id journey
simply me journey. So let's go ahead and niche down once again on the keyword. As you can see, we have
mid journey version six, how to use Journey
and a Russian title. We're going to ignore
the Russian title. Let's go ahead and click
on Journey version six. And as you can see, there is a really high search volume and an extremely
low competition. Why is that? A simple Google
search will tell us that the version six of
Mid Journey was released two months ago only. Which makes it pretty new. Therefore people are not
really updated and want to find out about the new
features and all that stuff. We can niche down once again to mid journey version
six tutorial. As you can see, it still has around 20 K searches
and low competition. However, I would stick to mid journey version six and
maybe change the keyword. So if we go right now on
the related keywords, we are going to find
mid journey tutorial, the features, and the rumors. Let's try and click
on the tutorial. Tutorial still has high
volume and low competition, which could be a good
idea to make a video on. If I look for example and
Mid Journey I tutorial, once again, we have
some really good start. Mid journey looks like a
really hot topic on Youtube. Let's take, for example,
Mid Journey V Six tutorial. It has a medium search
volume and low competition. Let's now check if this is true on actually Youtube.
So how do we do that? Well, we open an incognit award, we go on Youtube. And now I type mid journey. And as you can see there, all the keywords that we
have found before. Ai Tutorial, Tutorial six. I'll go with V six. And then the keyword
we were looking for is tutorial with
a lot of search, so it should be
on the first one, and in fact it's a second one. And if I click, I'll find that there is a lot
of search volume. As you can see, 24 K, 235 K, 103 k, and these videos
are all pretty new. So how would I go about this? I will go ahead and check if these channels which
are ranking first, have really high subscribers or they're just small channels. So I'll open them on a new tab by clicking the scroll
wheel on the mouse. So if I check it, the AI grid, it has a bunch of subscribers. Cyber Jungle as 40 K, however is video that
we have just watched as made five times a sub count,
which is pretty good. This guy has 41 K
subs and he has made double a sub count from this
video that we've just seen. And this last video
that was just posted two days ago already
has 14 K views, which means that
the search volume is actually extremely high. Now we are sure that this keyword is extremely
good to create a video. So if I had a channel in any way related
to generative AI, I would surely make a video
on mid journey version six. I will go ahead and
type this prompt, which you will find
in the word document. What I said is, I want to create a Youtube video about
mid journey V six, which is a topic that
was just released. It's an AI image generator. I want you to create a
title for the video. I will send you four titles to take inspiration from in order from the one that performed the best to the one that
performed the worst. And then I've just copy and pasted the titles from Youtube. And then I'm also going to
expand on my video by saying my video is going to be in our in depth tutorial about
the update has affected me, journey and the best way to leverage it to create
amazing style. And then going to
finish it by saying, give me ten title ideas. As you can see, we
have the first one which is Mastering
Mid Journey V six, Unleashing the power of AI Image generation,
which is pretty good. Exploring the latest
mid journey B six features a
comprehensive tutorial. This one might look
really well harnessing the full potential
of mid journey V six, advanced
technique revealed. This one is really good as well. The ultimate Journey six
tutorial from basics to Protyps, I've gotten some
really good results. I could also simply send a new prompt like Imam more
and he's going to give them. But really there are
some that are really good and I think they would perform really
well on Youtube, especially this one if I'm
doing a complete tutorial, or if it's just
advanced fissures, or if it's talking about
the update in the specific. It honestly depends from which angle I'll
take the video from, but I got some
really good results. Now let's go ahead and create a description regarding
the description. The prompt Te I'm up to is simply create a
fluent description. Now make sure to keep it
around eight lines long. You can merge keywords and then I'll also say which includes
the following keywords. And there we go, we have it. Are you ready to navigate the exciting word
of mid journey Six, The cutting edge AI tool offers a seamless mid journey
expert tips and tutorials. Mid journeys innovative features harness the power
of mid journey AI. Comprehensive tutorials,
step by step guides. This is a really
good one as well. Crafting prompts,
Let journey six, be your trusted companion. These are really good
description which includes basically all of the keywords
that we have said earlier. And we'll help our
videos rank high on the search results
and regarding the tugs, I'll keep it really simple. I would simply copy and
pastes into the tags part. Tags are not that important,
especially in Youtube. They are a bit if you're trying to create a new channel and the algorithm wants
some direction regarding who it should
suggest your videos to, But if you have an
established channel or even if it's new,
it's not that important. You can just copy
these ten tags, which are the basic
keywords. You're good to go. So now we have completed the
keyword research on Youtube. Make sure to download the
prompts in the word document.
13. Using ChatGPT To Script an eBook: Let's go over the final
lesson of this course. Here we are going to
talk about scripting. Any book with Chagepet. We are not going to create any book from scratch
as that would take probably another
full on course as creating an E book
requires a lot of time, but it's always the
same repetitive action. So what we will do
in this lesson is go ahead and script it and
create the chapters. The rest of the work with simply repeating
the same prompt, by simply changing the
chapter every time. And I'm not going to
bore you with that. I will probably
create book anyway because that's something I've
wanted to do for a while. And I'll do it of course
by using Chagept and rewriting some parts
because that's the most efficient and
best way to do it. Book is really good
for two reasons. The first one is you can use it and gift it from every social
media platform around. Just gift that ten
page book on whatever you're doing if you
are in the cooking nh just make a book with ten different recipes and just
gift it to your following. And that's going to
allow you to gather their e mail addresses through
a website like Gam Road. And once you have
their e mail address, you can add them to
your e mail list and sell them via e mail. Another way is directly
selling the ebook. I suggest to go
for a cheap price, around five to $10 I think it's the best price
to sell any book at. But I'm not going to focus on the marketing
aspect of any book. We're just going to
script it right now with these general kind
of long processes. By creating an E book, which is going to be around 50 pages or something like that. I usually do a two step
scripting process with GPT and I let him know in
advance so that he knows how to reply to me
and what to answer me. Because if he tries to script
the whole thing at once, it's going to just
give me a too broad, not really defined,
not specific answer. And we do not want that, we want the outline to
be extremely clear. So I'll go ahead now and type. I want to write an E book. So I've gone ahead and type now the subject subject is going
to be social media content. The ways to approach it created what's required to set it up, how to be successful in it, how to analyze your past
results in order to improve, but not the ways to monetize it. In fact, that's going to be another book
completely on its own, because monetizing on social
media is plenty of options. And you need to choose
carefully the ways to approach the marketing and the moneymaking on
your Youtube channel. Because anyone above
1,000 subscribers on Youtube or 2000 followers
on Instagram can make loads of money every
month if he's smart about it. But that requires another
50 pages on his own, at least to outline all of the possible scenarios
to set it up. I'm going to create an E book
problem on that as well. But for now let's focus on
auto per social media content. Now that we have the
subject and the goal, let's go ahead and type what we actually want from Ch GPT. So I've gone ahead and type. I want you to create
ten chapters which subdivide the information
listed in ten bullet points. So we've put also
the output template. Now I will add a clarification
and then I typed, do not write the chapters. I only want the title
to figure out if we are on the same page about the content we are
going to type. So I'll go ahead and send. And as you can see, we now have understanding
your audience. So target our study pretty good. And we're going to
add this crafting, compelling content
strategies for engagement and
impact. Really good. We're going to have to edit as well platforms and presents. Choosing the right channels for your message. Extremely good. We're going to have to leverage some specific social media depending on our work
message art of consistency. So how important it is to
be consistent every day. We're going to add this as well, leverage individuals, so how to create eye
catching content. However, we have said kind of the same thing
in 0.2 How to maximize, reach important
engagement strategies, analytic study, and the
power of storytelling. And I will avoid the part of
trends as this is an book, digital product,
that you type once and you don't reuse once again. So I'm just going to
delete this part. So after a problem like this, I would go ahead and retype GPT. So I've gone ahead
and type, all right, we'll start the book with this chapter that PQ
already provided, which is a target outer study. And then I said we're going to proceed with this
one right here. After that, we'll go
with chapter three, The Recipe to be successful. And I've added some of the
chapters that he has said, just Justina molar and compressed first
on as you can see. And then I simply added chapter four with the
technicalities of gear, editing assets all the stuff. And then I also want to create a small master class
on short form content. I've added this and just a brief description
in chapter six. I have applied the Chapter eight that P has provided,
which is analytics. I'll now go ahead and
type create an outline following the six
chapters and known as small description
per each chapter. Perfect. Now we have six chapters for our book and we have a small description. How would I go about
creating the ebook now? So I'll go ahead and expand on this chapter, which
is the first one. And then I'll say, make sure to provide valuable information and leave placeholders
when you don't know the answer or you
believe it's outdated, be extensive, but don't waffle. So I'll go ahead and
enter and look at this. We now have a small
introduction and we have demographic analysis,
interests and preferences, analysis of behavior,
personal development, content personalization, and also optimization and adaptation to changes
of the audience. This is an extremely good
outline for our first chapter. So what I would go ahead
and do is simply say, now put all of this
information in a cohesive paragraph without
it being bullet points. And I would simply get the chapter one and
then repeat it for Chapter 2,345.6 And
we will have our Bo. Regarding the design part, I suggest to use Canva because it's really
easy to use it, just concrete a design
search for Evo. You click it here you have
a lot of book templates. As you can see, I could
paste this right in. And then if I go forward
and create more pages, I can go ahead and copy, for example, this
template right here. And get already pre
made templates. I can change what's
written here, so for example,
social media content. Social media. And then
go right here and say instead of a journal about fitness, health
and lifestyle, an E book about growing
on social media, I would research the more
I would probably ask a Jp once again what could be a
good subtitle for this book. However, as you can
see, for example here, I wouldn't use a
page like this for my E book as the information in providing
it's really extensive. And I cannot just describe it
in this small box of text. If I go back, as you can see, there are plenty of
templates, There's a lot. You can also go on evil pages and you will find straight up pages instead of
full on templates. I could copy this right here. And it's really straightforward. It's not going to take too
long to create an book. The biggest part is
just creating the text. We have now learned also
how to create an book with GPT and how to design it
pretty quickly with Canvas. I didn't go really
too deep in that because I don't think
there's too much need. The hardest part is just to get the information going and that's where you're
going to use TragPT. Then you can simply use canvas templates to
create an ebook. They have templates
for business, for time management, productivity,
all that good stuff. So it really depends
on your niche. So thank you for following this course and I will see you in the next
lesson when I'm going to wrap up all
the course and tell you where you can find
more valuable information.
14. Get Your Free Prompt List!: I'll now show you how
to get the assets. So I'll go down here
and as you can see, there is project and resources. I'll click on this and
then I will scroll down on Download Resources
and just download it. And now we'll be
able to have this in my download folder on the PC. So I'll just open it up
and as you can see here, we have our prompt
list that you can use. The first one is for
bios for Instagram. Then we have the Youtube
titles and description SCO, and then the E book structuring. So if you enjoy this course, I would really appreciate
if you can leave a review and that we should
have an amazing day.
Buzz, AI Video Editor
