Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated:
1
00:00:00,070 --> 00:00:02,390
You're using ChatGPT wrong, but it's not
2
00:00:02,390 --> 00:00:04,430
your fault. Whenever a new feature gets
3
00:00:04,430 --> 00:00:06,430
released, self-proclaimed experts share
4
00:00:06,430 --> 00:00:09,190
their best practices to capitalize on the
5
00:00:09,190 --> 00:00:11,830
hype. That's all fine and well, except at
6
00:00:11,830 --> 00:00:14,270
that point most of their advice haven't
7
00:00:14,270 --> 00:00:16,870
gone through enough testing to produce
8
00:00:16,870 --> 00:00:19,150
consistently good results. Let's take
9
00:00:19,150 --> 00:00:20,590
custom instructions for example, A
10
00:00:20,590 --> 00:00:22,950
powerful feature available for both paid
11
00:00:22,950 --> 00:00:25,550
and free users you might have heard. It's
12
00:00:25,550 --> 00:00:27,710
best to be as specific as possible.
13
00:00:27,910 --> 00:00:30,320
Include your values and principles. Your
14
00:00:30,320 --> 00:00:31,600
learning style, The city you live in, the
15
00:00:31,600 --> 00:00:33,720
apps you prefer to use. This is terrible
16
00:00:33,720 --> 00:00:36,510
advice because custom instructions. Acts
17
00:00:36,510 --> 00:00:39,070
like an umbrella that should cover all of
18
00:00:39,070 --> 00:00:41,630
your use cases. For example, if my custom
19
00:00:41,630 --> 00:00:43,470
instructions were super specific and
20
00:00:43,470 --> 00:00:45,510
optimized for writing YouTube scripts,
21
00:00:45,590 --> 00:00:48,110
the relevancy of the chapter PT outputs
22
00:00:48,110 --> 00:00:50,470
for my work and personal life will
23
00:00:50,470 --> 00:00:52,510
actually decrease. Here's what to do
24
00:00:52,510 --> 00:00:55,030
instead. For the first box up top, keep
25
00:00:55,030 --> 00:00:57,670
it simple with three sentences. First, a
26
00:00:57,670 --> 00:00:59,990
snapshot of who I am professionally, I'm
27
00:00:59,990 --> 00:01:01,430
a Product marketing manager in tech.
28
00:01:01,830 --> 00:01:04,510
Second, insert your secondary persona.
29
00:01:04,590 --> 00:01:07,230
Who are you outside of work?In my spare
30
00:01:07,230 --> 00:01:09,230
time, I create online content on these
31
00:01:09,230 --> 00:01:12,189
topics and 3rd my personality
32
00:01:12,189 --> 00:01:14,069
and my interests. I'm all about
33
00:01:14,069 --> 00:01:16,189
efficiency, productivity and self
34
00:01:16,189 --> 00:01:18,709
improvement. That's it. The bottom field
35
00:01:18,709 --> 00:01:20,869
is extremely important and here's a hack.
36
00:01:20,909 --> 00:01:23,029
Find an online creator who's articulate,
37
00:01:23,029 --> 00:01:24,749
smart, capable, charming, charismatic.
38
00:01:24,789 --> 00:01:26,509
Basically someone you respect, admire,
39
00:01:26,509 --> 00:01:28,189
and want to emulate. Let's go with Ali
40
00:01:28,189 --> 00:01:30,149
Abdullah, head on over to his newsletter
41
00:01:30,149 --> 00:01:32,909
page and open up the last two links
42
00:01:33,149 --> 00:01:36,029
1-2, then head on back to
43
00:01:36,029 --> 00:01:38,399
Chachi BT. Start a new chat and input
44
00:01:38,399 --> 00:01:40,319
this prompt. I'm going to share two blog
45
00:01:40,319 --> 00:01:42,359
posts written by Ali Abdullah. Analyze
46
00:01:42,359 --> 00:01:43,759
the blog post and give me a set of
47
00:01:43,759 --> 00:01:46,199
instructions I can use to write in
48
00:01:46,199 --> 00:01:48,599
exactly the same tone, writing style,
49
00:01:48,599 --> 00:01:51,599
humour, reading level and delivery. Here
50
00:01:51,599 --> 00:01:54,149
are the blog posts. Blog post one
51
00:01:54,149 --> 00:01:56,789
paste, Blog post two paste and hit
52
00:01:56,829 --> 00:01:58,589
enter. Obviously it doesn't have to be a
53
00:01:58,589 --> 00:01:59,869
newsletter. It could be an online
54
00:01:59,869 --> 00:02:02,149
articles, LinkedIn posts, or even speech
55
00:02:02,149 --> 00:02:04,469
transcripts from your favorite comedian.
56
00:02:04,549 --> 00:02:05,749
You got to hit the like and subscribe
57
00:02:05,749 --> 00:02:06,989
button if you're also a fan of Trevor
58
00:02:06,989 --> 00:02:08,949
Noah by the way. If not, hit the dislike
59
00:02:08,949 --> 00:02:11,349
button twice. Now instead of copy pasting
60
00:02:11,349 --> 00:02:13,189
all of this into your custom
61
00:02:13,189 --> 00:02:15,189
instructions, pick and choose the ones
62
00:02:15,189 --> 00:02:17,909
you vibe with the most. For example, use
63
00:02:17,909 --> 00:02:19,789
Casual Language Baker Post into
64
00:02:19,789 --> 00:02:21,669
digestible sections. Definitely good ones
65
00:02:21,669 --> 00:02:24,639
to include. Light sarcasm, cheeky remarks
66
00:02:24,639 --> 00:02:26,359
might not be for everyone, right? You got
67
00:02:26,359 --> 00:02:28,439
the idea. Copy paste the ones you select
68
00:02:28,439 --> 00:02:31,399
into the second custom instruction box
69
00:02:31,399 --> 00:02:34,319
down here. Now ChatGPT knows just
70
00:02:34,319 --> 00:02:36,119
enough about your background and
71
00:02:36,119 --> 00:02:39,029
preferences. To provide relevant outputs
72
00:02:39,029 --> 00:02:41,709
for any and all of your use cases.
73
00:02:41,789 --> 00:02:44,109
Moving on, I'm 100% surethose of us with
74
00:02:44,109 --> 00:02:45,789
no coding experience have made the second
75
00:02:45,789 --> 00:02:48,349
mistake with Chachi BT not writing code
76
00:02:48,349 --> 00:02:50,709
to automate tasks. To give some context,
77
00:02:50,789 --> 00:02:52,389
I studied business in college and the
78
00:02:52,389 --> 00:02:54,549
closest I've ever gotten to writing code
79
00:02:54,549 --> 00:02:57,069
is typing in cheat codes in Warcraft 3.
80
00:02:57,069 --> 00:02:59,309
Not not that I needed to cheat to win. I
81
00:02:59,309 --> 00:03:01,469
want every game. I was a pro. That was
82
00:03:01,469 --> 00:03:03,629
great. And yet with the help of Chachi BT
83
00:03:04,069 --> 00:03:06,749
I've written a script that allows me to
84
00:03:06,749 --> 00:03:09,349
click a button in Google Docs. To
85
00:03:09,349 --> 00:03:11,949
generate a new page before every
86
00:03:11,949 --> 00:03:14,549
single one-on-one meeting I have with my
87
00:03:14,549 --> 00:03:17,549
manager and I have no idea what
88
00:03:17,549 --> 00:03:19,389
any of this code means. All you have to
89
00:03:19,389 --> 00:03:22,349
do is to tell Chachi BT hey write a
90
00:03:22,349 --> 00:03:23,869
Google app script for my one-on-one
91
00:03:23,869 --> 00:03:25,669
meeting notes in Google Docs that does
92
00:03:25,669 --> 00:03:27,589
the following. Add a new page with
93
00:03:27,589 --> 00:03:29,828
heading two that says Date today's date.
94
00:03:30,188 --> 00:03:32,828
Add a heading 3 notes followed by three
95
00:03:32,828 --> 00:03:35,508
bullet points. Add a heading 3 action
96
00:03:35,508 --> 00:03:38,508
items followed by a numbered list. Create
97
00:03:38,508 --> 00:03:40,308
a button I can click before every meeting
98
00:03:40,308 --> 00:03:43,108
to generate this new page and not only
99
00:03:43,108 --> 00:03:45,588
will Chatty BT generate the entire script
100
00:03:45,588 --> 00:03:47,868
without you having to make any edits, it
101
00:03:47,868 --> 00:03:49,548
will also provide step by step
102
00:03:49,548 --> 00:03:51,948
instructions on how to implement this for
103
00:03:51,948 --> 00:03:53,428
yourself. I won't bore you with the
104
00:03:53,428 --> 00:03:55,628
details, but basically within most Google
105
00:03:55,628 --> 00:03:57,068
applications you can click on the
106
00:03:57,068 --> 00:03:59,028
extensions menu to open up the app's
107
00:03:59,028 --> 00:04:01,268
script, copy and paste the code GHGBT
108
00:04:01,268 --> 00:04:04,188
generated, rename it, click save
109
00:04:04,188 --> 00:04:06,668
and run the script. It's that
110
00:04:06,668 --> 00:04:09,038
easy. Pro tip You can even make simple
111
00:04:09,038 --> 00:04:11,038
tweaks yourself. For example, this grips
112
00:04:11,078 --> 00:04:13,118
outputs a date whenever you click on the
113
00:04:13,118 --> 00:04:15,908
button and this long date format. Month,
114
00:04:15,908 --> 00:04:18,748
date year I can change this to a shorter
115
00:04:18,868 --> 00:04:21,428
month date year save.
116
00:04:23,238 --> 00:04:26,238
Run We can now apply
117
00:04:26,238 --> 00:04:28,118
for software engineer roles at Google.
118
00:04:28,278 --> 00:04:30,318
You're welcome. Jokes aside, we're now
119
00:04:30,318 --> 00:04:32,438
only limited by our imagination. For
120
00:04:32,438 --> 00:04:34,358
example, I also created a button, Google
121
00:04:34,358 --> 00:04:37,278
Slides that standardizes the font
122
00:04:37,278 --> 00:04:40,118
and font size for all speaker notes,
123
00:04:40,118 --> 00:04:42,758
since different colleagues might copy and
124
00:04:42,758 --> 00:04:45,518
paste their slides into one
125
00:04:45,518 --> 00:04:48,078
presentation. Super handy. By the way,
126
00:04:48,078 --> 00:04:49,958
this video is not sponsored, but it is
127
00:04:49,958 --> 00:04:51,518
supported by those of you who subscribe
128
00:04:51,518 --> 00:04:53,358
to my paid product TV newsletter on
129
00:04:53,358 --> 00:04:55,548
Google Tips. Link in the description to
130
00:04:55,548 --> 00:04:57,748
learn more. The 3rd way we're using Chat
131
00:04:57,748 --> 00:05:00,028
GPT wrong is falling into the first try
132
00:05:00,028 --> 00:05:01,908
fallacy. This is the unrealistic
133
00:05:01,908 --> 00:05:04,188
expectation that we get perfect answers
134
00:05:04,188 --> 00:05:06,788
from Chachi BT after just our initial
135
00:05:06,788 --> 00:05:08,948
prompt. Yes Jeff, I know we should follow
136
00:05:08,948 --> 00:05:10,788
up, but it takes so much time and there's
137
00:05:10,788 --> 00:05:12,628
no standardized framework for following
138
00:05:12,628 --> 00:05:14,908
with catchy BT. It's just too much extra
139
00:05:14,908 --> 00:05:16,668
effort. Good news, instead of all that
140
00:05:16,668 --> 00:05:18,468
extra effort, all you have to do is to
141
00:05:18,468 --> 00:05:20,588
add this sentence at the end of your
142
00:05:20,588 --> 00:05:23,028
prompts. Ask me 5 questions that would
143
00:05:23,028 --> 00:05:24,948
improve the response you'll be giving me.
144
00:05:24,948 --> 00:05:26,948
For example, as a marketing manager, I
145
00:05:26,948 --> 00:05:28,908
pay my vendors in a variety of
146
00:05:28,908 --> 00:05:30,948
currencies. And to keep track of all
147
00:05:30,948 --> 00:05:33,628
these payments in Google Sheets, I asked
148
00:05:33,628 --> 00:05:35,628
Chatty BT is there a way for me to
149
00:05:35,628 --> 00:05:37,828
automate tracking so that even if I input
150
00:05:37,828 --> 00:05:39,828
the amount in its original currency,
151
00:05:39,948 --> 00:05:41,588
there will be a column that automatically
152
00:05:41,588 --> 00:05:44,268
converts the amount to U.S. dollars USD
153
00:05:44,468 --> 00:05:46,588
asking me 5 questions that would improve
154
00:05:46,588 --> 00:05:48,588
the response you'll be giving me instead
155
00:05:48,588 --> 00:05:50,507
of responding to my request immediately.
156
00:05:50,507 --> 00:05:52,827
ChatGPT clarifies its understanding by
157
00:05:52,827 --> 00:05:55,147
asking me questions like hey, are you
158
00:05:55,147 --> 00:05:57,627
cool with using Apps Script? If yes, I'll
159
00:05:57,627 --> 00:05:59,947
write you a script, you know Code genius.
160
00:06:00,107 --> 00:06:02,227
If not, we'll stick with formulas, you
161
00:06:02,227 --> 00:06:04,267
lowly peasant, or other questions like do
162
00:06:04,267 --> 00:06:06,067
you need real time currency conversion?
163
00:06:06,067 --> 00:06:08,427
If yes, we'll use option A, if not, then
164
00:06:08,427 --> 00:06:10,227
option B. After I provide the secondary
165
00:06:10,227 --> 00:06:12,267
round of input ChatGPT gives me an
166
00:06:12,267 --> 00:06:14,627
extremely relevant response. If you don't
167
00:06:14,627 --> 00:06:16,147
believe me, try using the same prompt
168
00:06:16,147 --> 00:06:18,027
without the hey ask me 5 questions
169
00:06:18,027 --> 00:06:20,627
sentence and compare the results. The 4th
170
00:06:20,627 --> 00:06:22,467
mistake I see is what I call the summary
171
00:06:22,467 --> 00:06:24,667
only shortfall, where the majority of
172
00:06:24,667 --> 00:06:27,027
users are still only using chatty BT to
173
00:06:27,027 --> 00:06:29,107
summarize information instead of
174
00:06:29,107 --> 00:06:31,427
distilling actionable insights. For
175
00:06:31,427 --> 00:06:32,947
example, let's say you come across this
176
00:06:32,947 --> 00:06:35,787
article on Holiday shopping insights. You
177
00:06:35,787 --> 00:06:38,467
can select all text, copy,
178
00:06:38,547 --> 00:06:40,467
ask Chatty BT to summarize the article
179
00:06:40,467 --> 00:06:43,307
paste and it'll do a fine job. But there
180
00:06:43,307 --> 00:06:45,147
are two small changes you can make to get
181
00:06:45,147 --> 00:06:47,867
exponentially better outputs. First, to
182
00:06:47,867 --> 00:06:49,587
get personalized insights that are
183
00:06:49,587 --> 00:06:52,117
specific to you. Try this. I'll share an
184
00:06:52,117 --> 00:06:53,797
article below on holiday shopping
185
00:06:53,797 --> 00:06:55,677
insights. How does the information in
186
00:06:55,677 --> 00:06:58,357
this article apply to a Insert your role
187
00:06:58,597 --> 00:07:00,597
paid Media Marketing Manager responsible
188
00:07:00,597 --> 00:07:03,267
for increasing brand awareness?Here's the
189
00:07:03,267 --> 00:07:05,267
article and you paste the article. I'm
190
00:07:05,267 --> 00:07:07,787
now provided with actionable next steps
191
00:07:07,827 --> 00:07:10,467
like focusing on Genji shoppers using
192
00:07:10,467 --> 00:07:12,227
YouTube because more than half of Genji
193
00:07:12,227 --> 00:07:14,747
use YouTube for holiday shopping. Wow. I
194
00:07:14,747 --> 00:07:17,347
guess that's not too surprising. Pro tip,
195
00:07:17,347 --> 00:07:19,707
instead of copy and pasting, paid users
196
00:07:19,707 --> 00:07:22,627
can upload APDF onto Chat GP directly
197
00:07:22,627 --> 00:07:25,227
and the prompts work in the exact same
198
00:07:25,227 --> 00:07:27,267
way. Use case number two is my favorite.
199
00:07:27,267 --> 00:07:28,987
It saved me so much time and it's to have
200
00:07:28,987 --> 00:07:31,587
a dedicated Q&A session with the written
201
00:07:31,587 --> 00:07:33,827
content. Type this into chatty BT. Assume
202
00:07:33,827 --> 00:07:35,747
the role of a Senior Business Analyst
203
00:07:35,747 --> 00:07:37,507
with over 20 years of experience for the
204
00:07:37,507 --> 00:07:39,307
rest of this conversation. I like to
205
00:07:39,307 --> 00:07:41,027
engage in AQ and a session where you'll
206
00:07:41,027 --> 00:07:43,547
provide insights, analysis and answers
207
00:07:43,547 --> 00:07:45,347
based on the uploaded report in PDF
208
00:07:45,507 --> 00:07:48,067
format. If you understand, please respond
209
00:07:48,067 --> 00:07:50,547
with yes. Great. Now instead of spending
210
00:07:50,547 --> 00:07:53,187
30-40 minutes reading this entire report
211
00:07:53,187 --> 00:07:56,067
by LinkedIn, I can simply ask targeted
212
00:07:56,067 --> 00:07:58,627
questions. Hey, does the report provide
213
00:07:58,627 --> 00:08:00,547
any evidence there'll be layoffs because
214
00:08:00,587 --> 00:08:03,107
of AI and chatty BT responds citing what
215
00:08:03,107 --> 00:08:05,347
it has read in the report. I can even
216
00:08:05,347 --> 00:08:07,347
stress test this by asking for something
217
00:08:07,387 --> 00:08:09,946
I know is not included in the report.
218
00:08:10,186 --> 00:08:11,826
Hey, does the report mention AI tools
219
00:08:11,826 --> 00:08:13,506
like Google Bart or Perplexity AI? Since
220
00:08:13,506 --> 00:08:15,706
LinkedIn is owned by Microsoft, I highly
221
00:08:15,706 --> 00:08:17,506
doubt those two tools will be mentioned.
222
00:08:17,506 --> 00:08:19,546
And here we see that yes, the report
223
00:08:19,546 --> 00:08:22,226
focuses on the impact of chatty BT, but
224
00:08:22,226 --> 00:08:24,426
the report does not specifically mention
225
00:08:24,426 --> 00:08:26,506
Google Bart or Perplexity. The 5th
226
00:08:26,506 --> 00:08:28,466
mistake I see people making the prompt
227
00:08:28,466 --> 00:08:30,906
overload paradox where we consume a lot
228
00:08:30,906 --> 00:08:33,066
of chatty BT related content. We save
229
00:08:33,066 --> 00:08:35,066
them offline somewhere, but we forget to
230
00:08:35,066 --> 00:08:37,746
optimize or even use those prompts in our
231
00:08:37,746 --> 00:08:40,146
daily tasks. Here's a simple solution,
232
00:08:40,146 --> 00:08:41,866
and no, it's not to click out this video,
233
00:08:41,866 --> 00:08:44,226
you gotta keep watching. One tell Chachi
234
00:08:44,226 --> 00:08:46,786
BT2 Assume the role of an experienced
235
00:08:46,786 --> 00:08:49,106
prompt engineer with over 20 years of
236
00:08:49,106 --> 00:08:51,626
experience. I am a insert your role
237
00:08:51,626 --> 00:08:53,746
working on Insert your responsibility
238
00:08:53,746 --> 00:08:56,466
with the goal of your objective. Give me
239
00:08:56,466 --> 00:08:58,506
10 chatty BT prompts that would help me
240
00:08:58,506 --> 00:09:01,146
become more productive in my job. Step
241
00:09:01,146 --> 00:09:03,586
two pick a maximum of three prompts you
242
00:09:03,586 --> 00:09:05,266
can see yourself using every day at work.
243
00:09:05,546 --> 00:09:07,226
For me that would be this competitive
244
00:09:07,226 --> 00:09:09,906
analysis one for sure. This audience
245
00:09:09,906 --> 00:09:12,786
insight one and this
246
00:09:12,786 --> 00:09:14,826
create a survey to get feedback from my
247
00:09:14,826 --> 00:09:16,946
audience. Then focus on using those 3
248
00:09:16,946 --> 00:09:19,156
prompts that work for two weeks. And
249
00:09:19,156 --> 00:09:20,596
refine them using the perfect prompt
250
00:09:20,596 --> 00:09:22,956
formula on an entire video here so that
251
00:09:22,956 --> 00:09:24,876
you can actually become more productive.
252
00:09:25,426 --> 00:09:26,466
See you on the next video in the
253
00:09:26,466 --> 00:09:28,946
meantime. Have a great one.
18518
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.