1
00:00:12,533 --> 00:00:13,366
Stellan?
2
00:00:13,500 --> 00:00:14,833
Kang, I wanted you!
3
00:00:14,833 --> 00:00:17,233
Look! Come here
4
00:00:17,233 --> 00:00:18,866
I have this super fancy new tool
5
00:00:18,866 --> 00:00:20,333
It’s called bolt.new
6
00:00:20,333 --> 00:00:22,200
It’s both designing and coding for you
7
00:00:22,200 --> 00:00:25,500
So I’ve prompted the new start page
for the website I'm thinking of here
8
00:00:25,900 --> 00:00:27,600
So it’s like, you know, just
9
00:00:27,600 --> 00:00:30,166
Boom! That’s prompt engineering for you
10
00:00:31,233 --> 00:00:34,233
Look
11
00:00:37,600 --> 00:00:40,033
Yeah, it’s really impressive
12
00:00:40,700 --> 00:00:41,766
Yeah, it’s like
13
00:00:41,766 --> 00:00:44,266
anyone can code now, so…
14
00:00:44,266 --> 00:00:47,166
I’m kind of guessing developers
might soon be out of a job
15
00:00:52,666 --> 00:00:55,500
Okay, so you want me… like,
16
00:00:55,500 --> 00:00:57,966
implement this, right now?
17
00:00:57,966 --> 00:01:00,066
More like copy paste it into our website
18
00:01:00,066 --> 00:01:01,500
It’s already working, look
19
00:01:01,500 --> 00:01:03,300
I can click here, and I can
slide this thing here
20
00:01:03,300 --> 00:01:06,566
And it’s updating the pricing there
21
00:01:06,966 --> 00:01:10,600
So, you know, I’ll have a fika, and you…
22
00:01:10,600 --> 00:01:12,666
Go ahead, take my place
23
00:01:13,100 --> 00:01:14,100
Kang
24
00:01:14,100 --> 00:01:15,866
It’s no longer a game of
“the more you know”
25
00:01:15,866 --> 00:01:17,866
It’s everyone’s game
26
00:01:18,133 --> 00:01:19,533
Future is looking bright
27
00:01:19,533 --> 00:01:20,933
See you
28
00:01:40,666 --> 00:01:43,766
[Chinese flute playing]
29
00:01:43,900 --> 00:01:45,466
Kung fu
30
00:01:45,466 --> 00:01:49,500
—a term often associated with Chinese
martial arts
31
00:01:49,500 --> 00:01:55,366
—traditionally refers to any discipline
achieved through dedication and time
32
00:01:55,900 --> 00:01:57,500
Tea making
33
00:01:57,500 --> 00:01:58,766
Cooking
34
00:01:58,766 --> 00:02:00,466
Calligraphy
35
00:02:00,466 --> 00:02:02,500
Even programming
36
00:02:02,800 --> 00:02:06,933
And after 30 years of straining my eyes
and fingers
37
00:02:07,033 --> 00:02:10,266
My level is well over 9000
38
00:02:10,266 --> 00:02:13,266
A master’s degree, literally
39
00:02:13,600 --> 00:02:16,100
But a new threat is rising
40
00:02:16,100 --> 00:02:17,900
In the face of AI
41
00:02:17,900 --> 00:02:22,700
My proud skills are but hands and swords
against guns and bombs
42
00:02:23,266 --> 00:02:27,000
What was once
the aura of the wise old master
43
00:02:27,366 --> 00:02:30,833
is now… just old age
44
00:02:41,233 --> 00:02:43,100
Welcome to the Secret Tech Club
45
00:02:43,100 --> 00:02:46,800
A video series where
we make tech fun and approachable
46
00:02:46,800 --> 00:02:51,666
I’m Kang, and today I have been joined
by our one and only Stellan
47
00:02:52,000 --> 00:02:52,766
Thank you, Kang
48
00:02:52,900 --> 00:02:57,633
And this probably means that I, for once,
don’t have to talk that much
49
00:02:57,900 --> 00:03:02,866
because you are much more knowledgeable
when it comes to AI, as proven before
50
00:03:03,433 --> 00:03:06,133
In our skit
51
00:03:06,133 --> 00:03:07,966
And as you might guess
52
00:03:07,966 --> 00:03:11,233
We are going to talk about
AI development today
53
00:03:11,966 --> 00:03:14,400
I don’t know about you,
Stellan, but from my perspective
54
00:03:14,400 --> 00:03:16,600
This topic is a bit dangerous
55
00:03:16,600 --> 00:03:18,600
Cuz whatever we might say today
56
00:03:18,600 --> 00:03:21,133
Might not be true in a few months
57
00:03:21,200 --> 00:03:24,266
Or weeks, or days, or seconds
58
00:03:24,266 --> 00:03:25,200
And secondly, the future is
looking really grim for us developers
59
00:03:25,200 --> 00:03:28,533
And secondly, the future is
looking really grim for us developers
60
00:03:28,533 --> 00:03:31,666
So, I think, I want
to record this with you
61
00:03:31,666 --> 00:03:33,466
And hopefully you will reassure (us)
62
00:03:33,466 --> 00:03:35,066
And hopefully you will reassure (us)
63
00:03:35,066 --> 00:03:36,733
That it’s okay
64
00:03:36,733 --> 00:03:40,733
Or we actually
might end up somewhere else
65
00:03:40,733 --> 00:03:43,400
Where everyone loses their job
66
00:03:43,400 --> 00:03:47,666
So what do you think
about the AI future, Stellan?
67
00:03:47,700 --> 00:03:51,566
I think it’s
kind of interesting in a way
68
00:03:51,566 --> 00:03:55,066
Cuz if you look at design,
for example
69
00:03:55,066 --> 00:03:58,500
At least in our trade, it’s always been
that everyone has an opinion
70
00:03:58,500 --> 00:04:02,666
about design,
and everyone, in some shape or form,
71
00:04:02,666 --> 00:04:07,166
think that they are a designer,
and have an opinion about design
72
00:04:07,166 --> 00:04:10,566
And so, now to a much larger
extent than before,
73
00:04:10,566 --> 00:04:15,300
you can actually have an opinion
about code and development
74
00:04:15,800 --> 00:04:18,800
Because if someone says it can’t be done
75
00:04:19,000 --> 00:04:21,766
You could potentially test it
76
00:04:21,766 --> 00:04:24,766
And it’s like “maybe it can”
77
00:04:25,233 --> 00:04:29,000
So that opens up interesting
doors, I think
78
00:04:29,900 --> 00:04:31,600
That’s one aspect of all the things
79
00:04:31,600 --> 00:04:34,700
Another aspect is this whole thing of
80
00:04:34,900 --> 00:04:36,966
I read on X that someone said
81
00:04:36,966 --> 00:04:40,566
Essentially, we’re moving
up another obfuscation layer
82
00:04:40,566 --> 00:04:46,200
So, you know, no one
is coding Basic anymore
83
00:04:46,200 --> 00:04:48,266
You know, so whatever
it took Bill Gates
84
00:04:48,266 --> 00:04:51,633
some six weeks
or whatever to do that in the 70s
85
00:04:51,933 --> 00:04:56,800
And we moved like a long way up
from the ship level, already
86
00:04:56,800 --> 00:04:59,633
And now we’re just moving up
one more level
87
00:04:59,633 --> 00:05:01,800
In terms of what we’re coding
88
00:05:01,966 --> 00:05:04,466
I always said that front-end development
89
00:05:04,466 --> 00:05:07,600
is harder than back-end development
90
00:05:07,600 --> 00:05:11,566
not because it is harder,
but because nobody has opinions
91
00:05:11,566 --> 00:05:14,666
on back-end development.
But front-end development, you see it!
92
00:05:14,700 --> 00:05:17,933
So you can have an opinion, like you said
93
00:05:17,933 --> 00:05:22,500
But now you can have even more opinions,
because you can try it out
94
00:05:22,500 --> 00:05:26,700
Look here! It’s already done
in a few seconds
95
00:05:27,833 --> 00:05:28,400
Exactly
96
00:05:28,400 --> 00:05:28,966
Yeah
97
00:05:28,966 --> 00:05:31,933
And then when it comes to like, okay,
more or less work, etc …
98
00:05:31,933 --> 00:05:36,133
I think that parallel can be drawn
from historical events like
99
00:05:36,633 --> 00:05:40,600
development has been
a scarce commodity in a way
100
00:05:40,600 --> 00:05:45,033
It’s like been way too few developers
101
00:05:45,033 --> 00:05:49,833
compared to what people has,
you know, demanded for a very long time
102
00:05:49,833 --> 00:05:53,766
And now all of a sudden,
we can sort of supercharge developers
103
00:05:53,766 --> 00:05:56,900
to maybe meet the
demand of development that’s out there
104
00:05:56,900 --> 00:05:59,900
That’s more
how I see it than anything else
105
00:05:59,900 --> 00:06:03,800
So, I think we
will see much more being done
106
00:06:03,800 --> 00:06:06,366
than what we have historically seen
107
00:06:06,366 --> 00:06:08,766
Which should
ultimately be good for humanity
108
00:06:08,766 --> 00:06:12,600
If we take it the full
step of what this will lead to
109
00:06:12,600 --> 00:06:13,233
Yeah
110
00:06:13,233 --> 00:06:19,466
And some have
compared AI to how we before,
111
00:06:19,466 --> 00:06:24,300
didn’t have calculators, and now
we have calculators, but we still do math
112
00:06:24,300 --> 00:06:25,200
Yeah
113
00:06:25,200 --> 00:06:30,100
But I can’t help but still
think that AI might be a threat
114
00:06:30,100 --> 00:06:31,500
regardless, you know
115
00:06:31,500 --> 00:06:34,266
But, if I compare it to art
116
00:06:34,266 --> 00:06:37,233
Throughout history,
for example, photography
117
00:06:37,233 --> 00:06:41,500
Instead of trying to replicate
the reality, they started to do more
118
00:06:41,500 --> 00:06:43,800
absurd stuff
119
00:06:43,800 --> 00:06:44,633
And we have always managed to
120
00:06:44,633 --> 00:06:47,000
And from Rembrant to Picasso
121
00:06:47,000 --> 00:06:48,500
Yeah
122
00:06:48,500 --> 00:06:53,433
And we have always managed
to innovate and find new expressions
123
00:06:53,433 --> 00:06:56,300
At least know I feel like
124
00:06:56,300 --> 00:06:59,733
the jobs that we
used to have that required a lot of
125
00:06:59,733 --> 00:07:02,433
basically just routines and repetition
126
00:07:02,433 --> 00:07:06,033
They will be eliminated first
127
00:07:06,033 --> 00:07:07,633
And that is a good thing
128
00:07:07,633 --> 00:07:10,400
But I’m also worried because
129
00:07:10,400 --> 00:07:12,766
a lot of entry level positions
in programming
130
00:07:12,766 --> 00:07:16,366
They rely on these type of
tasks to learn
131
00:07:16,366 --> 00:07:17,366
Right
132
00:07:17,366 --> 00:07:20,033
And now we automate it away, or
133
00:07:20,033 --> 00:07:23,500
let a senior developer do it with
the help of AI
134
00:07:23,500 --> 00:07:25,133
So we don’t need junior developers
135
00:07:25,133 --> 00:07:27,833
And that leads us to the next topic
136
00:07:27,833 --> 00:07:31,066
“Is AI good or bad
for developers, long term?”
137
00:07:31,066 --> 00:07:35,300
Because it seems like we don’t
need junior developers anymore
138
00:07:35,300 --> 00:07:38,700
When we’re moving
towards such a future
139
00:07:38,700 --> 00:07:42,000
I think essentially,
if you’re interested, this is a blessing
140
00:07:42,000 --> 00:07:45,600
because you can
now have your personal teacher
141
00:07:45,600 --> 00:07:48,033
that will teach you everything you
need to know much faster
142
00:07:48,033 --> 00:07:50,900
than you could learn by doing,
you know, trial and error before
143
00:07:50,900 --> 00:07:54,166
have a colleague try to explain
things and they explain it like,
144
00:07:54,166 --> 00:07:57,200
you know, Einstein level,
and you are not at Einstein level
145
00:07:57,200 --> 00:08:00,100
so you have no idea what they’re saying,
but, you know, you nod anyway, and go like
146
00:08:00,100 --> 00:08:04,166
“That sounds good” because you don’t want
to make anyone disappointed or whatever
147
00:08:05,200 --> 00:08:06,933
So I’m, you know
148
00:08:06,933 --> 00:08:10,700
with my kids, for example,
constantly using this. They’re
149
00:08:10,700 --> 00:08:14,333
in different subjects,
so I had, like, just last night
150
00:08:14,333 --> 00:08:18,300
my oldest was
preparing for a math test
151
00:08:18,300 --> 00:08:22,733
And is doing equations
and didn’t understand a certain thing
152
00:08:22,733 --> 00:08:26,300
I took a picture of it,
sent it to ChatGPT and said “explain this”
153
00:08:26,300 --> 00:08:30,166
for a 16 year old,
and they did in, like, 10s
154
00:08:30,166 --> 00:08:31,633
And I was like, oh, okay
155
00:08:31,633 --> 00:08:32,433
Yeah, that’s it
156
00:08:32,433 --> 00:08:35,200
Okay, now I remember from class or
whatever it was
157
00:08:35,200 --> 00:08:38,000
So, I think the same, like
158
00:08:38,000 --> 00:08:41,900
use it as a teacher and
it will sort of supercharge
159
00:08:41,900 --> 00:08:45,500
junior developers into becoming
senior developers faster
160
00:08:45,500 --> 00:08:48,066
Yeah,
I think this is a very interesting topic
161
00:08:48,066 --> 00:08:50,633
because I’ve always
tried to crack the code:
162
00:08:50,633 --> 00:08:54,633
How to teach junior developers,
and help them upskill, you know
163
00:08:54,633 --> 00:08:56,866
and I read theories about
164
00:08:56,866 --> 00:09:00,866
in larger companies, you need to have
junior developers to do the routines
165
00:09:00,866 --> 00:09:03,866
but they cannot directly
communicate with the senior
166
00:09:03,866 --> 00:09:07,133
because the level is, you know,
the gap is too big
167
00:09:07,133 --> 00:09:09,900
so we have to have mid-levels
168
00:09:09,900 --> 00:09:11,300
But with AI
169
00:09:11,300 --> 00:09:13,466
they can explain in any level
170
00:09:13,466 --> 00:09:18,000
So I think, AI for learning is so great
171
00:09:18,000 --> 00:09:20,266
I read an article, a blog post
172
00:09:20,266 --> 00:09:21,600
The blog post is called:
173
00:09:21,600 --> 00:09:24,633
“New junior developers can’t actually code”
174
00:09:24,633 --> 00:09:26,333
Kind of negative
175
00:09:26,333 --> 00:09:28,566
But I’m gonna summarize a little bit
176
00:09:28,566 --> 00:09:32,266
So, he had been observing that
177
00:09:32,266 --> 00:09:35,433
the junior developers today rely
heavily on AI tools
178
00:09:35,433 --> 00:09:37,733
Copilot and ChatGPT
179
00:09:37,733 --> 00:09:39,333
And his concern was that
180
00:09:39,333 --> 00:09:42,666
many can produce code but
don’t understand what they're doing
181
00:09:42,666 --> 00:09:46,233
When I started to learn code, it were books
182
00:09:46,233 --> 00:09:50,166
The rest you had to just guess
and like, find out yourself
183
00:09:50,166 --> 00:09:52,600
And then came Google,
and StackOverflow
184
00:09:52,600 --> 00:09:54,100
So the pre-AI era:
185
00:09:54,100 --> 00:09:54,766
You had a problem,
186
00:09:54,766 --> 00:09:56,833
Your searched StackOverflow,
187
00:09:56,833 --> 00:10:00,433
you read the docs, tried out
different solutions, trial and error
188
00:10:00,433 --> 00:10:02,033
Finally, you learnt something
189
00:10:02,033 --> 00:10:05,466
This whole process creates some kind of
deeper understanding and growth
190
00:10:05,466 --> 00:10:06,133
Yeah
191
00:10:06,133 --> 00:10:08,966
And now, AI just provides fast answers
192
00:10:08,966 --> 00:10:10,500
with minimal effort
193
00:10:10,500 --> 00:10:12,800
And there’s a temptation
to copy and paste
194
00:10:12,800 --> 00:10:15,566
without truly
learning the concepts
195
00:10:15,566 --> 00:10:18,466
We might develop a
generation that doesn’t have this
196
00:10:18,466 --> 00:10:19,833
habit of problem solving
197
00:10:19,833 --> 00:10:24,466
And then if the AI fails
for some reason or starts hallucinating,
198
00:10:24,466 --> 00:10:27,000
which it eventually
will do at some point,
199
00:10:27,000 --> 00:10:30,666
then they lack a
lot of knowledge to solve it
200
00:10:30,666 --> 00:10:33,166
Okay. So that was the general idea
201
00:10:33,166 --> 00:10:34,466
and how
202
00:10:34,466 --> 00:10:38,300
But I think the comparison
with StackOverflow is good, cuz
203
00:10:38,300 --> 00:10:41,233
you could potentially
just take that piece of code
204
00:10:41,233 --> 00:10:42,566
and you have no idea how it works
205
00:10:42,566 --> 00:10:45,233
paste it in, and it works.
It’s the same with AI
206
00:10:45,233 --> 00:10:47,500
You take that piece of code,
paste it in, it works
207
00:10:47,500 --> 00:10:48,466
You have no idea
208
00:10:48,466 --> 00:10:50,366
But if you’re interested
209
00:10:50,366 --> 00:10:51,633
then you learn something from it
210
00:10:51,633 --> 00:10:53,233
and you try
to understand how it works
211
00:10:53,233 --> 00:10:56,066
regardless if it’s AI or StackOverflow
212
00:10:56,066 --> 00:10:57,900
I think that it remains the same
213
00:10:57,900 --> 00:11:02,300
Yeah, it’s just that you have to piece
together so much in StackOverflow
214
00:11:02,300 --> 00:11:03,600
That's the great thing with AI
215
00:11:03,600 --> 00:11:07,133
that you can just ask it and say
“explain this piece of code for me”
216
00:11:07,133 --> 00:11:10,700
Yeah, and that
is a very good point, because
217
00:11:10,700 --> 00:11:14,466
I believe that AI is
super great for teaching
218
00:11:14,466 --> 00:11:15,566
Just like you said
219
00:11:15,566 --> 00:11:23,333
Because I view development
as a master-apprentice kind of profession
220
00:11:23,333 --> 00:11:27,866
where when you’re
new, you should just observe
221
00:11:27,866 --> 00:11:32,133
when the masters are working,
and then you learn from that
222
00:11:32,133 --> 00:11:34,200
And you need a many years of experience
223
00:11:34,200 --> 00:11:38,866
And you can ask the master
—nowadays you say mentorship
224
00:11:38,866 --> 00:11:42,633
You have to find a mentor to help you
225
00:11:42,633 --> 00:11:44,600
move right direction
226
00:11:44,600 --> 00:11:46,766
And that is not easy to find.
Like, who’s going to help you?
227
00:11:46,766 --> 00:11:48,633
And that is not easy to find.
Like, who’s going to help you?
228
00:11:48,633 --> 00:11:53,666
But with AI, every time
you ask it to generate something
229
00:11:53,666 --> 00:11:57,500
It will try to generate the
code with the latest best practices
230
00:11:57,500 --> 00:11:58,766
and also explain it
231
00:11:58,766 --> 00:12:00,966
And you can also bounce
ideas with it
232
00:12:00,966 --> 00:12:04,600
So it’s like having this mentor 24/7
233
00:12:04,600 --> 00:12:08,366
But the key is what you said,
you have to be willing to learn
234
00:12:08,366 --> 00:12:10,400
because it’s much
easier to just paste the code,
235
00:12:10,400 --> 00:12:14,300
just use the code, not read what
ChatGPT says, and not being curious
236
00:12:14,300 --> 00:12:20,100
We need people who can fill
in that gap and make the judgment
237
00:12:20,100 --> 00:12:23,233
Yeah, I can just compare
with myself as well
238
00:12:23,233 --> 00:12:26,133
It’s like I’m using
it a lot to create these
239
00:12:26,133 --> 00:12:29,433
what’s called a “custom coded
workflow action” for a certain tool
240
00:12:29,433 --> 00:12:33,066
that we’re working with called HubSpot
241
00:12:33,600 --> 00:12:37,666
And it’s like, I don’t know a
single line of Node (JavaScript backend)
242
00:12:37,666 --> 00:12:41,166
So Node.js, I don’t
know anything, honestly
243
00:12:41,166 --> 00:12:44,033
But it took me,
244
00:12:44,033 --> 00:12:47,366
with the working script
from the beginning
245
00:12:47,366 --> 00:12:50,200
which I could understand
246
00:12:50,200 --> 00:12:53,366
because all the comments to the code
247
00:12:53,366 --> 00:12:58,166
that, and in the beginning
ChatGPT and now later Grok
248
00:12:58,166 --> 00:13:04,266
took me, I guess, 2-3 months
to find a good way to prompt the AI
249
00:13:04,266 --> 00:13:09,233
to get what I wanted to do
in terms of results, to be able to do
250
00:13:09,233 --> 00:13:14,133
lots of things that would
require a proper developer early on
251
00:13:14,133 --> 00:13:19,000
Because it’s so good
at explaining what it’s doing
252
00:13:19,000 --> 00:13:24,666
If you can prompt it well,
it will also return a good answer
253
00:13:24,666 --> 00:13:28,100
And that leads us to the next topic
254
00:13:28,100 --> 00:13:31,666
that something has happened
over the last week(s)
255
00:13:31,666 --> 00:13:35,500
I feel like it’s happened
over the last week(s) that a new verb
256
00:13:35,500 --> 00:13:36,733
was invented
257
00:13:36,733 --> 00:13:38,300
And that is “Vibe coding”
258
00:13:38,300 --> 00:13:38,733
Yeah
259
00:13:38,733 --> 00:13:44,566
And you are
kind of an expert vibe coder
260
00:13:44,566 --> 00:13:47,800
So Stellan is vibe coding
during the evenings
261
00:13:47,800 --> 00:13:49,800
and sends all the
new tricks that he learnt
262
00:13:49,800 --> 00:13:53,066
so he’s much
more advanced than I am
263
00:13:53,066 --> 00:13:54,800
when it comes to vibe coding
264
00:13:54,800 --> 00:13:57,000
But what is five coding?
265
00:13:57,000 --> 00:14:00,433
So I don’t know who invented this
266
00:14:00,433 --> 00:14:05,066
this term, but basically it is
that you are not a developer
267
00:14:05,066 --> 00:14:07,933
with the developer background,
but you are able to code
268
00:14:07,933 --> 00:14:10,900
pretty advanced
things with these new AI tools
269
00:14:10,900 --> 00:14:14,133
So you’re kind of vibing your way
270
00:14:14,133 --> 00:14:16,700
So I guess that’s where it comes from
271
00:14:17,766 --> 00:14:21,000
And yeah,
people are doing amazing things
272
00:14:21,000 --> 00:14:26,400
especially in like games, people
are doing crazy things with games
273
00:14:26,533 --> 00:14:30,133
I also see
a whole industry popping up
274
00:14:30,133 --> 00:14:33,233
around tools to create websites,
for example
275
00:14:33,233 --> 00:14:37,100
The big issues that people have had
so far has been around security
276
00:14:37,266 --> 00:14:38,766
So people are hacking the shit
277
00:14:38,766 --> 00:14:40,433
out of these things, basically
278
00:14:40,433 --> 00:14:43,600
And some just laugh
at it, and say like, well
279
00:14:43,600 --> 00:14:49,800
“Someone put some crazy thing in
my game, and I’m trying to solve it”
280
00:14:49,800 --> 00:14:55,000
and other people go like,
“Okay, this is genuinely a problem”
281
00:14:55,000 --> 00:14:56,966
But I’m pretty convinced that
282
00:14:56,966 --> 00:15:01,166
this again goes back to like,
what kind of prompts
283
00:15:01,166 --> 00:15:04,066
are you giving to the system?
What are you asking it to do?
284
00:15:04,066 --> 00:15:06,666
Are you asking it to do
these kind of security checks,
285
00:15:06,666 --> 00:15:09,633
or are you asking it to do refactoring,
and things like that?
286
00:15:09,633 --> 00:15:12,333
What I found just from my
hobby vibe coding
287
00:15:12,333 --> 00:15:16,200
is that most people who do this well
288
00:15:16,200 --> 00:15:20,400
they have a process for this,
so it’s just like
289
00:15:20,400 --> 00:15:23,466
I would imagine
a bit of a development process
290
00:15:23,466 --> 00:15:26,900
where you have certain steps
that you do that you approach
291
00:15:26,900 --> 00:15:30,500
Whereas if you
just randomly prompt it
292
00:15:30,500 --> 00:15:34,366
it’s not going to end well,
but you can have a process for it
293
00:15:34,366 --> 00:15:38,933
And that process,
people use ChatGPT or Grok
294
00:15:38,933 --> 00:15:42,266
to do this kind of development,
as a step by step process
295
00:15:42,266 --> 00:15:43,933
And they ask you to create the process
296
00:15:43,933 --> 00:15:46,066
They ask you to create the documentation
297
00:15:46,066 --> 00:15:50,333
They ask you to create, you know, folder
structure, necessary checks, etc.
298
00:15:51,400 --> 00:15:53,066
and also create the prompts
299
00:15:53,066 --> 00:15:55,833
And then,
then they use that in whatever tool
300
00:15:55,833 --> 00:15:59,733
that they are working with, whether
it’s Cursor (AI code editor) or anything else
301
00:15:59,733 --> 00:16:04,600
But the thing you are talking about
though, I think it’s quite unfamiliar
302
00:16:04,600 --> 00:16:08,433
or even foreign for people who
don’t hang out on X (ex-Twitter)
303
00:16:08,433 --> 00:16:11,800
I don’t see these kind of things
on LinkedIn, for example
304
00:16:11,800 --> 00:16:13,800
And I’ve been hanging
out on Twitter
305
00:16:13,800 --> 00:16:17,966
or “X” now, for a long time,
and I know that
306
00:16:17,966 --> 00:16:21,200
people are much faster
at adopting a technology on X
307
00:16:21,200 --> 00:16:22,733
So many early adopters
308
00:16:22,733 --> 00:16:26,833
And the games that you’re talking about,
they’re often done by indie hackers
309
00:16:26,833 --> 00:16:29,933
And solo entrepreneurs
and they are usually so quick
310
00:16:29,933 --> 00:16:35,633
to jump on new technology
and try to earn money on them
311
00:16:36,033 --> 00:16:37,200
and for them, like,
312
00:16:37,200 --> 00:16:41,066
they don’t have to think about the quality
and code maintenance at all
313
00:16:41,066 --> 00:16:44,400
You know, one of the best indie hackers
314
00:16:44,400 --> 00:16:48,300
The ones who made the game
— the flying game
315
00:16:48,300 --> 00:16:52,266
He only codes with jQuery (JS library)
jQuery and PHP (backend)
316
00:16:52,266 --> 00:16:56,400
And he launched so
many projects over the years
317
00:16:56,566 --> 00:16:57,266
But…
318
00:16:57,966 --> 00:17:02,466
How about Zooma?
Will vibe coding have a place
319
00:17:02,600 --> 00:17:05,333
outside of your late evenings?
320
00:17:06,000 --> 00:17:06,733
I think one thing
321
00:17:06,733 --> 00:17:10,133
that we have already done
a bit of is actually prototyping
322
00:17:10,133 --> 00:17:10,833
You know
323
00:17:10,833 --> 00:17:13,466
You move faster from a wireframe
324
00:17:13,466 --> 00:17:16,933
to an actual working prototype
so you can test all the interactions
325
00:17:16,933 --> 00:17:20,133
and what you actually want to happen
when you click this button,
326
00:17:20,133 --> 00:17:22,733
what is actually going to happen,
and so on
327
00:17:22,733 --> 00:17:24,833
You can simulate that much faster
328
00:17:24,833 --> 00:17:27,833
actually, I think, with AI tools
329
00:17:27,833 --> 00:17:30,600
I think for me, the thing would be
330
00:17:30,600 --> 00:17:38,266
If designers could get into this workflow
so they can test out their ideas faster
331
00:17:38,266 --> 00:17:39,366
Yeah
332
00:17:39,366 --> 00:17:42,133
I don't think we
have necessarily cracked it
333
00:17:42,133 --> 00:17:48,233
So we move from Photoshop to Figma,
and then you do static prototypes
334
00:17:48,233 --> 00:17:50,233
You add more functionality to prototypes
335
00:17:50,233 --> 00:17:54,200
The more functionality you add,
the more complicated it becomes
336
00:17:54,200 --> 00:17:58,100
Certain designers
will just say “Stop! This is my line here”
337
00:17:58,100 --> 00:18:02,933
“I’m not gonna do more. I just
wanna draw my designs”
338
00:18:02,933 --> 00:18:06,433
So the dream would be
if they could actually test out
339
00:18:06,433 --> 00:18:10,333
their designs
without the help of developers
340
00:18:10,333 --> 00:18:11,900
Yeah
341
00:18:11,900 --> 00:18:13,000
Or what do you think?
342
00:18:13,000 --> 00:18:15,600
Yeah, I genuinely think so as well
343
00:18:15,600 --> 00:18:17,300
That it's like, what do we want?
344
00:18:17,300 --> 00:18:22,366
Well, we don’t want AI generated
websites and apps and things like that
345
00:18:22,366 --> 00:18:25,666
that look like…
everything looks exactly the same
346
00:18:25,666 --> 00:18:29,533
We want this
sort of creative element to it
347
00:18:29,533 --> 00:18:32,600
So far,
humans are better at that
348
00:18:32,600 --> 00:18:36,000
Eventually, you know,
AI might be really good at this as well
349
00:18:36,000 --> 00:18:37,833
with the right input
350
00:18:37,833 --> 00:18:39,833
I’m actually seeing AI more as of
351
00:18:39,833 --> 00:18:43,100
like another teammate in the room
352
00:18:43,100 --> 00:18:44,200
than anything else
353
00:18:44,200 --> 00:18:47,033
So if you just imagine an empty chair
354
00:18:47,033 --> 00:18:48,933
and there’s an AI in it
355
00:18:48,933 --> 00:18:50,433
And I think that
356
00:18:50,433 --> 00:18:53,800
eventually this
might lead us to have a more
357
00:18:53,800 --> 00:18:59,100
of a conversational relationship
with AI and that ultimately helps us
358
00:18:59,433 --> 00:19:03,133
you know, develop things
that are higher quality, faster
359
00:19:03,133 --> 00:19:06,966
As you mentioned,
there’s a process to prompting
360
00:19:06,966 --> 00:19:09,933
It’s kind of straightforward, you
use natural language, but
361
00:19:09,933 --> 00:19:12,066
but it’s still not as
362
00:19:12,066 --> 00:19:17,466
For example, for a designer,
it’s not as straightforward as drawing
363
00:19:17,466 --> 00:19:19,133
You have to describe your design
364
00:19:19,133 --> 00:19:24,366
But when it comes to
design, you want to create it visually
365
00:19:24,366 --> 00:19:26,800
The same with development
366
00:19:26,800 --> 00:19:30,700
So if you get more advanced,
now you just described a process
367
00:19:30,700 --> 00:19:33,466
of how to prompt, and the structure things
368
00:19:33,466 --> 00:19:35,700
I don’t think anybody do it
369
00:19:35,700 --> 00:19:37,166
There’s a certain line where
people will say
370
00:19:37,166 --> 00:19:38,800
“Nope, that’s not what I do”
371
00:19:38,800 --> 00:19:40,033
“That’s not how I work”
372
00:19:40,033 --> 00:19:43,000
We might be looking
at this specific person
373
00:19:43,000 --> 00:19:44,066
So
374
00:19:44,066 --> 00:19:48,066
Will we eventually hire someone
who is like prompt engineer
375
00:19:48,066 --> 00:19:51,233
Like ten times
better than the normal person
376
00:19:51,533 --> 00:19:54,366
Well, it’s hard to tell
377
00:19:54,366 --> 00:19:58,566
at the moment, it’s a lot of talk
about, like, a new profession
378
00:19:58,566 --> 00:20:00,233
to be able to do that
379
00:20:00,233 --> 00:20:01,500
But, yeah, definitely.
380
00:20:01,500 --> 00:20:05,766
I think there are some skills going on,
in relation to that
381
00:20:05,766 --> 00:20:07,633
and that it’s constantly evolving
382
00:20:07,633 --> 00:20:09,533
It’s like every day
there’s something new
383
00:20:09,533 --> 00:20:12,966
Just last week
I learned that you can use JSON
384
00:20:12,966 --> 00:20:16,600
to prompt ChatGPT
for image generation, for example
385
00:20:16,600 --> 00:20:19,733
So you basically, with
code, you tell it, you know
386
00:20:19,733 --> 00:20:21,866
what kind of copy you want in the image
387
00:20:21,866 --> 00:20:25,333
where you want it, where
you want to place certain items
388
00:20:25,333 --> 00:20:28,266
the type of person that
should be in the image, and so on
389
00:20:28,266 --> 00:20:31,133
And it has
a better understanding of that
390
00:20:31,133 --> 00:20:33,133
than if you just write it
391
00:20:33,133 --> 00:20:35,466
So if you just typed the description…
392
00:20:35,466 --> 00:20:36,600
Yeah, exactly
393
00:20:36,600 --> 00:20:39,166
It’s markup in a way
394
00:20:39,166 --> 00:20:44,133
Again, it’s like , for these guys
that are like coding games for example
395
00:20:44,133 --> 00:20:46,166
Most of them seem to use Grok
396
00:20:46,166 --> 00:20:49,200
to create the instructions
and the prompts
397
00:20:49,200 --> 00:20:51,633
for example, for Cursor (AI editor)
398
00:20:51,633 --> 00:20:53,700
So they don’t just use a single tool
399
00:20:53,700 --> 00:20:56,133
They use Grok as a foundation
400
00:20:56,133 --> 00:20:58,966
for the idea of the game,
and the prompts and the process
401
00:20:58,966 --> 00:21:02,200
and everything, and then they
use that in Cursor
402
00:21:02,200 --> 00:21:05,033
So I think that’s also, like,
to find your process
403
00:21:05,033 --> 00:21:07,300
is also part of
what tools are you are using
404
00:21:07,300 --> 00:21:08,700
and for what
405
00:21:08,700 --> 00:21:11,966
Cuz it’s a combination of things
at the moment
406
00:21:12,433 --> 00:21:14,933
When I think about it,
it’s kind of sad, because
407
00:21:14,933 --> 00:21:17,033
you described the
whole process of using Grok
408
00:21:17,033 --> 00:21:20,133
to bounce ideas,
and write design documents,
409
00:21:20,133 --> 00:21:22,100
to do design and programming
410
00:21:22,100 --> 00:21:23,600
It used to be
a whole team of people
411
00:21:23,600 --> 00:21:26,566
Where you bounce and talk,
and now it’s just one person
412
00:21:28,300 --> 00:21:33,266
I saw a list somewhere of the
startups over the last year or so
413
00:21:33,266 --> 00:21:35,833
and the size of the teams,
they’re really small
414
00:21:35,966 --> 00:21:37,900
It’s like 10 to 20 people
415
00:21:37,900 --> 00:21:41,300
in these teams that are developing
this sort of SaaS tools
416
00:21:41,600 --> 00:21:43,800
And I also saw that someone in
Y Combinator
417
00:21:43,800 --> 00:21:46,166
the incubator in Silicon Valley
418
00:21:46,166 --> 00:21:50,900
known for a lot of famous companies having
gone through that program
419
00:21:50,900 --> 00:21:55,466
They said that like a third
of all applicants for the last round,
420
00:21:55,466 --> 00:22:00,433
did something like
95% of their coding with AI tools
421
00:22:00,433 --> 00:22:03,366
Didn’t write more
than 5% of the code themselves
422
00:22:03,366 --> 00:22:04,733
So
423
00:22:04,733 --> 00:22:06,766
Yeah, that’s the
direction where we’re headed in
424
00:22:06,766 --> 00:22:10,000
You’re kind of turbo changing
what you able to do
425
00:22:10,000 --> 00:22:12,533
I don't think it's gonna
lead to mass unemployment
426
00:22:12,533 --> 00:22:15,000
I’m kind of optimistic about it,
I think we’re gonna see
427
00:22:15,000 --> 00:22:19,200
more tools and more new cool stuff
than we saw before
428
00:22:20,366 --> 00:22:21,400
As a manager,
429
00:22:21,400 --> 00:22:25,166
what do you think is a good advice
to future developers
430
00:22:25,166 --> 00:22:27,600
who are, “afraid”?
431
00:22:27,600 --> 00:22:30,033
because I myself,
I wouldn’t advice someone
432
00:22:30,033 --> 00:22:33,100
“Let’s study development”
Like “it’s a good!”
433
00:22:33,100 --> 00:22:35,333
I think if you’re interested
434
00:22:35,333 --> 00:22:38,600
as we said before,
this is not going to be a problem
435
00:22:38,600 --> 00:22:42,200
This is going to be a the door opener
to so many exciting
436
00:22:42,200 --> 00:22:45,466
new opportunities
that we have not seen before
437
00:22:45,466 --> 00:22:50,233
We’re sort of on the breakthrough
in a lot of areas related to AI
438
00:22:50,233 --> 00:22:52,233
Whether it’s self-driving cars
439
00:22:52,233 --> 00:22:54,600
or whether it’s in robotics
440
00:22:54,600 --> 00:22:56,633
or in science in general
441
00:22:56,633 --> 00:22:59,466
What we can develop vaccines for
442
00:22:59,466 --> 00:23:01,633
and find a cure for cancer
443
00:23:01,633 --> 00:23:03,100
But you say “interested”
444
00:23:03,100 --> 00:23:06,866
So if you just want to learn
web development, or development
445
00:23:06,866 --> 00:23:10,666
just because you want
to earn good money, and have a
446
00:23:10,666 --> 00:23:13,200
flexible work time… is that…
447
00:23:13,200 --> 00:23:17,000
I think it’s a bad idea in general
to choose your profession
448
00:23:17,000 --> 00:23:20,133
after the potential salary
you might have
449
00:23:20,133 --> 00:23:23,000
I think you need to do something
that you’re really interested in
450
00:23:23,000 --> 00:23:24,500
Cuz if you just do it for the money
451
00:23:24,500 --> 00:23:27,566
that it’s like impossible to say
what's going to be
452
00:23:27,566 --> 00:23:29,166
worth your while, so to speak
453
00:23:29,166 --> 00:23:30,533
but again, like, if we take
454
00:23:30,533 --> 00:23:33,100
ATMs, for example came around,
and everyone says that
455
00:23:33,100 --> 00:23:36,000
Everyone working
in banks will be unemployed”
456
00:23:36,033 --> 00:23:38,933
Well, in fact, that didn’t really happen
457
00:23:38,933 --> 00:23:40,600
It’s like more people employed
in banking
458
00:23:40,600 --> 00:23:43,600
than ever before,
but you’re just doing different things
459
00:23:43,600 --> 00:23:45,566
You’re not
servicing people coming in
460
00:23:45,566 --> 00:23:48,566
asking for cash over the counter
461
00:23:48,566 --> 00:23:51,166
ATMs solved that problem a long time ago
462
00:23:51,166 --> 00:23:54,433
But you moved on to do more advanced work
463
00:23:54,433 --> 00:23:57,133
And I think it’s similar here
464
00:23:57,133 --> 00:24:00,400
It’s also like the industralisation
465
00:24:00,400 --> 00:24:05,600
Going back to the 1800s
and 1700s in England, and so on
466
00:24:05,600 --> 00:24:08,933
It’s like people
moving from farming to industry
467
00:24:08,933 --> 00:24:12,233
and then eventually from the industry
into white collar and so on.
468
00:24:12,233 --> 00:24:15,233
It’s an always evolving landscape
469
00:24:15,233 --> 00:24:17,533
But if you’re be interested
in what you’re doing
470
00:24:17,533 --> 00:24:19,133
you’re hopefully really good at it
471
00:24:19,133 --> 00:24:21,933
and then you’ll be able to adapt
472
00:24:22,233 --> 00:24:25,466
So that’s our optimistic key takeaway
473
00:24:25,466 --> 00:24:27,933
You should do what you think is fun
474
00:24:27,933 --> 00:24:31,266
What you’re interested in,
a not go for tech because you…
475
00:24:32,200 --> 00:24:36,266
Because it sounds good!
No, that’s a bad idea!
476
00:24:36,300 --> 00:24:37,400
Okay
477
00:24:37,400 --> 00:24:40,333
Thank you for the
fantastic discussion, Stellan
478
00:24:40,333 --> 00:24:41,633
We need to do this again
479
00:24:41,733 --> 00:24:44,033
And thank you, viewers, for watching
480
00:24:44,033 --> 00:24:47,033
Until next time, keep learning
481
00:24:47,200 --> 00:24:50,166
Bye bye — See you
482
00:24:52,866 --> 00:24:54,533
We kept the time, right?
483
00:24:54,533 --> 00:24:57,100
Yeah. So smooth.
484
00:24:57,100 --> 00:24:58,133
So smooth.
485
00:24:58,133 --> 00:25:01,100
That needs to be our first episode
486
00:25:01,100 --> 00:25:02,533
as well.
487
00:25:02,533 --> 00:25:04,333
So it's very intense.
488
00:25:04,333 --> 00:25:06,733
We. Oh.
489
00:25:06,733 --> 00:25:08,033
Okay. All right.
00:00:12,533 --> 00:00:13,366
Stellan?
2
00:00:13,500 --> 00:00:14,833
Kang, I wanted you!
3
00:00:14,833 --> 00:00:17,233
Look! Come here
4
00:00:17,233 --> 00:00:18,866
I have this super fancy new tool
5
00:00:18,866 --> 00:00:20,333
It’s called bolt.new
6
00:00:20,333 --> 00:00:22,200
It’s both designing and coding for you
7
00:00:22,200 --> 00:00:25,500
So I’ve prompted the new start page
for the website I'm thinking of here
8
00:00:25,900 --> 00:00:27,600
So it’s like, you know, just
9
00:00:27,600 --> 00:00:30,166
Boom! That’s prompt engineering for you
10
00:00:31,233 --> 00:00:34,233
Look
11
00:00:37,600 --> 00:00:40,033
Yeah, it’s really impressive
12
00:00:40,700 --> 00:00:41,766
Yeah, it’s like
13
00:00:41,766 --> 00:00:44,266
anyone can code now, so…
14
00:00:44,266 --> 00:00:47,166
I’m kind of guessing developers
might soon be out of a job
15
00:00:52,666 --> 00:00:55,500
Okay, so you want me… like,
16
00:00:55,500 --> 00:00:57,966
implement this, right now?
17
00:00:57,966 --> 00:01:00,066
More like copy paste it into our website
18
00:01:00,066 --> 00:01:01,500
It’s already working, look
19
00:01:01,500 --> 00:01:03,300
I can click here, and I can
slide this thing here
20
00:01:03,300 --> 00:01:06,566
And it’s updating the pricing there
21
00:01:06,966 --> 00:01:10,600
So, you know, I’ll have a fika, and you…
22
00:01:10,600 --> 00:01:12,666
Go ahead, take my place
23
00:01:13,100 --> 00:01:14,100
Kang
24
00:01:14,100 --> 00:01:15,866
It’s no longer a game of
“the more you know”
25
00:01:15,866 --> 00:01:17,866
It’s everyone’s game
26
00:01:18,133 --> 00:01:19,533
Future is looking bright
27
00:01:19,533 --> 00:01:20,933
See you
28
00:01:40,666 --> 00:01:43,766
[Chinese flute playing]
29
00:01:43,900 --> 00:01:45,466
Kung fu
30
00:01:45,466 --> 00:01:49,500
—a term often associated with Chinese
martial arts
31
00:01:49,500 --> 00:01:55,366
—traditionally refers to any discipline
achieved through dedication and time
32
00:01:55,900 --> 00:01:57,500
Tea making
33
00:01:57,500 --> 00:01:58,766
Cooking
34
00:01:58,766 --> 00:02:00,466
Calligraphy
35
00:02:00,466 --> 00:02:02,500
Even programming
36
00:02:02,800 --> 00:02:06,933
And after 30 years of straining my eyes
and fingers
37
00:02:07,033 --> 00:02:10,266
My level is well over 9000
38
00:02:10,266 --> 00:02:13,266
A master’s degree, literally
39
00:02:13,600 --> 00:02:16,100
But a new threat is rising
40
00:02:16,100 --> 00:02:17,900
In the face of AI
41
00:02:17,900 --> 00:02:22,700
My proud skills are but hands and swords
against guns and bombs
42
00:02:23,266 --> 00:02:27,000
What was once
the aura of the wise old master
43
00:02:27,366 --> 00:02:30,833
is now… just old age
44
00:02:41,233 --> 00:02:43,100
Welcome to the Secret Tech Club
45
00:02:43,100 --> 00:02:46,800
A video series where
we make tech fun and approachable
46
00:02:46,800 --> 00:02:51,666
I’m Kang, and today I have been joined
by our one and only Stellan
47
00:02:52,000 --> 00:02:52,766
Thank you, Kang
48
00:02:52,900 --> 00:02:57,633
And this probably means that I, for once,
don’t have to talk that much
49
00:02:57,900 --> 00:03:02,866
because you are much more knowledgeable
when it comes to AI, as proven before
50
00:03:03,433 --> 00:03:06,133
In our skit
51
00:03:06,133 --> 00:03:07,966
And as you might guess
52
00:03:07,966 --> 00:03:11,233
We are going to talk about
AI development today
53
00:03:11,966 --> 00:03:14,400
I don’t know about you,
Stellan, but from my perspective
54
00:03:14,400 --> 00:03:16,600
This topic is a bit dangerous
55
00:03:16,600 --> 00:03:18,600
Cuz whatever we might say today
56
00:03:18,600 --> 00:03:21,133
Might not be true in a few months
57
00:03:21,200 --> 00:03:24,266
Or weeks, or days, or seconds
58
00:03:24,266 --> 00:03:25,200
And secondly, the future is
looking really grim for us developers
59
00:03:25,200 --> 00:03:28,533
And secondly, the future is
looking really grim for us developers
60
00:03:28,533 --> 00:03:31,666
So, I think, I want
to record this with you
61
00:03:31,666 --> 00:03:33,466
And hopefully you will reassure (us)
62
00:03:33,466 --> 00:03:35,066
And hopefully you will reassure (us)
63
00:03:35,066 --> 00:03:36,733
That it’s okay
64
00:03:36,733 --> 00:03:40,733
Or we actually
might end up somewhere else
65
00:03:40,733 --> 00:03:43,400
Where everyone loses their job
66
00:03:43,400 --> 00:03:47,666
So what do you think
about the AI future, Stellan?
67
00:03:47,700 --> 00:03:51,566
I think it’s
kind of interesting in a way
68
00:03:51,566 --> 00:03:55,066
Cuz if you look at design,
for example
69
00:03:55,066 --> 00:03:58,500
At least in our trade, it’s always been
that everyone has an opinion
70
00:03:58,500 --> 00:04:02,666
about design,
and everyone, in some shape or form,
71
00:04:02,666 --> 00:04:07,166
think that they are a designer,
and have an opinion about design
72
00:04:07,166 --> 00:04:10,566
And so, now to a much larger
extent than before,
73
00:04:10,566 --> 00:04:15,300
you can actually have an opinion
about code and development
74
00:04:15,800 --> 00:04:18,800
Because if someone says it can’t be done
75
00:04:19,000 --> 00:04:21,766
You could potentially test it
76
00:04:21,766 --> 00:04:24,766
And it’s like “maybe it can”
77
00:04:25,233 --> 00:04:29,000
So that opens up interesting
doors, I think
78
00:04:29,900 --> 00:04:31,600
That’s one aspect of all the things
79
00:04:31,600 --> 00:04:34,700
Another aspect is this whole thing of
80
00:04:34,900 --> 00:04:36,966
I read on X that someone said
81
00:04:36,966 --> 00:04:40,566
Essentially, we’re moving
up another obfuscation layer
82
00:04:40,566 --> 00:04:46,200
So, you know, no one
is coding Basic anymore
83
00:04:46,200 --> 00:04:48,266
You know, so whatever
it took Bill Gates
84
00:04:48,266 --> 00:04:51,633
some six weeks
or whatever to do that in the 70s
85
00:04:51,933 --> 00:04:56,800
And we moved like a long way up
from the ship level, already
86
00:04:56,800 --> 00:04:59,633
And now we’re just moving up
one more level
87
00:04:59,633 --> 00:05:01,800
In terms of what we’re coding
88
00:05:01,966 --> 00:05:04,466
I always said that front-end development
89
00:05:04,466 --> 00:05:07,600
is harder than back-end development
90
00:05:07,600 --> 00:05:11,566
not because it is harder,
but because nobody has opinions
91
00:05:11,566 --> 00:05:14,666
on back-end development.
But front-end development, you see it!
92
00:05:14,700 --> 00:05:17,933
So you can have an opinion, like you said
93
00:05:17,933 --> 00:05:22,500
But now you can have even more opinions,
because you can try it out
94
00:05:22,500 --> 00:05:26,700
Look here! It’s already done
in a few seconds
95
00:05:27,833 --> 00:05:28,400
Exactly
96
00:05:28,400 --> 00:05:28,966
Yeah
97
00:05:28,966 --> 00:05:31,933
And then when it comes to like, okay,
more or less work, etc …
98
00:05:31,933 --> 00:05:36,133
I think that parallel can be drawn
from historical events like
99
00:05:36,633 --> 00:05:40,600
development has been
a scarce commodity in a way
100
00:05:40,600 --> 00:05:45,033
It’s like been way too few developers
101
00:05:45,033 --> 00:05:49,833
compared to what people has,
you know, demanded for a very long time
102
00:05:49,833 --> 00:05:53,766
And now all of a sudden,
we can sort of supercharge developers
103
00:05:53,766 --> 00:05:56,900
to maybe meet the
demand of development that’s out there
104
00:05:56,900 --> 00:05:59,900
That’s more
how I see it than anything else
105
00:05:59,900 --> 00:06:03,800
So, I think we
will see much more being done
106
00:06:03,800 --> 00:06:06,366
than what we have historically seen
107
00:06:06,366 --> 00:06:08,766
Which should
ultimately be good for humanity
108
00:06:08,766 --> 00:06:12,600
If we take it the full
step of what this will lead to
109
00:06:12,600 --> 00:06:13,233
Yeah
110
00:06:13,233 --> 00:06:19,466
And some have
compared AI to how we before,
111
00:06:19,466 --> 00:06:24,300
didn’t have calculators, and now
we have calculators, but we still do math
112
00:06:24,300 --> 00:06:25,200
Yeah
113
00:06:25,200 --> 00:06:30,100
But I can’t help but still
think that AI might be a threat
114
00:06:30,100 --> 00:06:31,500
regardless, you know
115
00:06:31,500 --> 00:06:34,266
But, if I compare it to art
116
00:06:34,266 --> 00:06:37,233
Throughout history,
for example, photography
117
00:06:37,233 --> 00:06:41,500
Instead of trying to replicate
the reality, they started to do more
118
00:06:41,500 --> 00:06:43,800
absurd stuff
119
00:06:43,800 --> 00:06:44,633
And we have always managed to
120
00:06:44,633 --> 00:06:47,000
And from Rembrant to Picasso
121
00:06:47,000 --> 00:06:48,500
Yeah
122
00:06:48,500 --> 00:06:53,433
And we have always managed
to innovate and find new expressions
123
00:06:53,433 --> 00:06:56,300
At least know I feel like
124
00:06:56,300 --> 00:06:59,733
the jobs that we
used to have that required a lot of
125
00:06:59,733 --> 00:07:02,433
basically just routines and repetition
126
00:07:02,433 --> 00:07:06,033
They will be eliminated first
127
00:07:06,033 --> 00:07:07,633
And that is a good thing
128
00:07:07,633 --> 00:07:10,400
But I’m also worried because
129
00:07:10,400 --> 00:07:12,766
a lot of entry level positions
in programming
130
00:07:12,766 --> 00:07:16,366
They rely on these type of
tasks to learn
131
00:07:16,366 --> 00:07:17,366
Right
132
00:07:17,366 --> 00:07:20,033
And now we automate it away, or
133
00:07:20,033 --> 00:07:23,500
let a senior developer do it with
the help of AI
134
00:07:23,500 --> 00:07:25,133
So we don’t need junior developers
135
00:07:25,133 --> 00:07:27,833
And that leads us to the next topic
136
00:07:27,833 --> 00:07:31,066
“Is AI good or bad
for developers, long term?”
137
00:07:31,066 --> 00:07:35,300
Because it seems like we don’t
need junior developers anymore
138
00:07:35,300 --> 00:07:38,700
When we’re moving
towards such a future
139
00:07:38,700 --> 00:07:42,000
I think essentially,
if you’re interested, this is a blessing
140
00:07:42,000 --> 00:07:45,600
because you can
now have your personal teacher
141
00:07:45,600 --> 00:07:48,033
that will teach you everything you
need to know much faster
142
00:07:48,033 --> 00:07:50,900
than you could learn by doing,
you know, trial and error before
143
00:07:50,900 --> 00:07:54,166
have a colleague try to explain
things and they explain it like,
144
00:07:54,166 --> 00:07:57,200
you know, Einstein level,
and you are not at Einstein level
145
00:07:57,200 --> 00:08:00,100
so you have no idea what they’re saying,
but, you know, you nod anyway, and go like
146
00:08:00,100 --> 00:08:04,166
“That sounds good” because you don’t want
to make anyone disappointed or whatever
147
00:08:05,200 --> 00:08:06,933
So I’m, you know
148
00:08:06,933 --> 00:08:10,700
with my kids, for example,
constantly using this. They’re
149
00:08:10,700 --> 00:08:14,333
in different subjects,
so I had, like, just last night
150
00:08:14,333 --> 00:08:18,300
my oldest was
preparing for a math test
151
00:08:18,300 --> 00:08:22,733
And is doing equations
and didn’t understand a certain thing
152
00:08:22,733 --> 00:08:26,300
I took a picture of it,
sent it to ChatGPT and said “explain this”
153
00:08:26,300 --> 00:08:30,166
for a 16 year old,
and they did in, like, 10s
154
00:08:30,166 --> 00:08:31,633
And I was like, oh, okay
155
00:08:31,633 --> 00:08:32,433
Yeah, that’s it
156
00:08:32,433 --> 00:08:35,200
Okay, now I remember from class or
whatever it was
157
00:08:35,200 --> 00:08:38,000
So, I think the same, like
158
00:08:38,000 --> 00:08:41,900
use it as a teacher and
it will sort of supercharge
159
00:08:41,900 --> 00:08:45,500
junior developers into becoming
senior developers faster
160
00:08:45,500 --> 00:08:48,066
Yeah,
I think this is a very interesting topic
161
00:08:48,066 --> 00:08:50,633
because I’ve always
tried to crack the code:
162
00:08:50,633 --> 00:08:54,633
How to teach junior developers,
and help them upskill, you know
163
00:08:54,633 --> 00:08:56,866
and I read theories about
164
00:08:56,866 --> 00:09:00,866
in larger companies, you need to have
junior developers to do the routines
165
00:09:00,866 --> 00:09:03,866
but they cannot directly
communicate with the senior
166
00:09:03,866 --> 00:09:07,133
because the level is, you know,
the gap is too big
167
00:09:07,133 --> 00:09:09,900
so we have to have mid-levels
168
00:09:09,900 --> 00:09:11,300
But with AI
169
00:09:11,300 --> 00:09:13,466
they can explain in any level
170
00:09:13,466 --> 00:09:18,000
So I think, AI for learning is so great
171
00:09:18,000 --> 00:09:20,266
I read an article, a blog post
172
00:09:20,266 --> 00:09:21,600
The blog post is called:
173
00:09:21,600 --> 00:09:24,633
“New junior developers can’t actually code”
174
00:09:24,633 --> 00:09:26,333
Kind of negative
175
00:09:26,333 --> 00:09:28,566
But I’m gonna summarize a little bit
176
00:09:28,566 --> 00:09:32,266
So, he had been observing that
177
00:09:32,266 --> 00:09:35,433
the junior developers today rely
heavily on AI tools
178
00:09:35,433 --> 00:09:37,733
Copilot and ChatGPT
179
00:09:37,733 --> 00:09:39,333
And his concern was that
180
00:09:39,333 --> 00:09:42,666
many can produce code but
don’t understand what they're doing
181
00:09:42,666 --> 00:09:46,233
When I started to learn code, it were books
182
00:09:46,233 --> 00:09:50,166
The rest you had to just guess
and like, find out yourself
183
00:09:50,166 --> 00:09:52,600
And then came Google,
and StackOverflow
184
00:09:52,600 --> 00:09:54,100
So the pre-AI era:
185
00:09:54,100 --> 00:09:54,766
You had a problem,
186
00:09:54,766 --> 00:09:56,833
Your searched StackOverflow,
187
00:09:56,833 --> 00:10:00,433
you read the docs, tried out
different solutions, trial and error
188
00:10:00,433 --> 00:10:02,033
Finally, you learnt something
189
00:10:02,033 --> 00:10:05,466
This whole process creates some kind of
deeper understanding and growth
190
00:10:05,466 --> 00:10:06,133
Yeah
191
00:10:06,133 --> 00:10:08,966
And now, AI just provides fast answers
192
00:10:08,966 --> 00:10:10,500
with minimal effort
193
00:10:10,500 --> 00:10:12,800
And there’s a temptation
to copy and paste
194
00:10:12,800 --> 00:10:15,566
without truly
learning the concepts
195
00:10:15,566 --> 00:10:18,466
We might develop a
generation that doesn’t have this
196
00:10:18,466 --> 00:10:19,833
habit of problem solving
197
00:10:19,833 --> 00:10:24,466
And then if the AI fails
for some reason or starts hallucinating,
198
00:10:24,466 --> 00:10:27,000
which it eventually
will do at some point,
199
00:10:27,000 --> 00:10:30,666
then they lack a
lot of knowledge to solve it
200
00:10:30,666 --> 00:10:33,166
Okay. So that was the general idea
201
00:10:33,166 --> 00:10:34,466
and how
202
00:10:34,466 --> 00:10:38,300
But I think the comparison
with StackOverflow is good, cuz
203
00:10:38,300 --> 00:10:41,233
you could potentially
just take that piece of code
204
00:10:41,233 --> 00:10:42,566
and you have no idea how it works
205
00:10:42,566 --> 00:10:45,233
paste it in, and it works.
It’s the same with AI
206
00:10:45,233 --> 00:10:47,500
You take that piece of code,
paste it in, it works
207
00:10:47,500 --> 00:10:48,466
You have no idea
208
00:10:48,466 --> 00:10:50,366
But if you’re interested
209
00:10:50,366 --> 00:10:51,633
then you learn something from it
210
00:10:51,633 --> 00:10:53,233
and you try
to understand how it works
211
00:10:53,233 --> 00:10:56,066
regardless if it’s AI or StackOverflow
212
00:10:56,066 --> 00:10:57,900
I think that it remains the same
213
00:10:57,900 --> 00:11:02,300
Yeah, it’s just that you have to piece
together so much in StackOverflow
214
00:11:02,300 --> 00:11:03,600
That's the great thing with AI
215
00:11:03,600 --> 00:11:07,133
that you can just ask it and say
“explain this piece of code for me”
216
00:11:07,133 --> 00:11:10,700
Yeah, and that
is a very good point, because
217
00:11:10,700 --> 00:11:14,466
I believe that AI is
super great for teaching
218
00:11:14,466 --> 00:11:15,566
Just like you said
219
00:11:15,566 --> 00:11:23,333
Because I view development
as a master-apprentice kind of profession
220
00:11:23,333 --> 00:11:27,866
where when you’re
new, you should just observe
221
00:11:27,866 --> 00:11:32,133
when the masters are working,
and then you learn from that
222
00:11:32,133 --> 00:11:34,200
And you need a many years of experience
223
00:11:34,200 --> 00:11:38,866
And you can ask the master
—nowadays you say mentorship
224
00:11:38,866 --> 00:11:42,633
You have to find a mentor to help you
225
00:11:42,633 --> 00:11:44,600
move right direction
226
00:11:44,600 --> 00:11:46,766
And that is not easy to find.
Like, who’s going to help you?
227
00:11:46,766 --> 00:11:48,633
And that is not easy to find.
Like, who’s going to help you?
228
00:11:48,633 --> 00:11:53,666
But with AI, every time
you ask it to generate something
229
00:11:53,666 --> 00:11:57,500
It will try to generate the
code with the latest best practices
230
00:11:57,500 --> 00:11:58,766
and also explain it
231
00:11:58,766 --> 00:12:00,966
And you can also bounce
ideas with it
232
00:12:00,966 --> 00:12:04,600
So it’s like having this mentor 24/7
233
00:12:04,600 --> 00:12:08,366
But the key is what you said,
you have to be willing to learn
234
00:12:08,366 --> 00:12:10,400
because it’s much
easier to just paste the code,
235
00:12:10,400 --> 00:12:14,300
just use the code, not read what
ChatGPT says, and not being curious
236
00:12:14,300 --> 00:12:20,100
We need people who can fill
in that gap and make the judgment
237
00:12:20,100 --> 00:12:23,233
Yeah, I can just compare
with myself as well
238
00:12:23,233 --> 00:12:26,133
It’s like I’m using
it a lot to create these
239
00:12:26,133 --> 00:12:29,433
what’s called a “custom coded
workflow action” for a certain tool
240
00:12:29,433 --> 00:12:33,066
that we’re working with called HubSpot
241
00:12:33,600 --> 00:12:37,666
And it’s like, I don’t know a
single line of Node (JavaScript backend)
242
00:12:37,666 --> 00:12:41,166
So Node.js, I don’t
know anything, honestly
243
00:12:41,166 --> 00:12:44,033
But it took me,
244
00:12:44,033 --> 00:12:47,366
with the working script
from the beginning
245
00:12:47,366 --> 00:12:50,200
which I could understand
246
00:12:50,200 --> 00:12:53,366
because all the comments to the code
247
00:12:53,366 --> 00:12:58,166
that, and in the beginning
ChatGPT and now later Grok
248
00:12:58,166 --> 00:13:04,266
took me, I guess, 2-3 months
to find a good way to prompt the AI
249
00:13:04,266 --> 00:13:09,233
to get what I wanted to do
in terms of results, to be able to do
250
00:13:09,233 --> 00:13:14,133
lots of things that would
require a proper developer early on
251
00:13:14,133 --> 00:13:19,000
Because it’s so good
at explaining what it’s doing
252
00:13:19,000 --> 00:13:24,666
If you can prompt it well,
it will also return a good answer
253
00:13:24,666 --> 00:13:28,100
And that leads us to the next topic
254
00:13:28,100 --> 00:13:31,666
that something has happened
over the last week(s)
255
00:13:31,666 --> 00:13:35,500
I feel like it’s happened
over the last week(s) that a new verb
256
00:13:35,500 --> 00:13:36,733
was invented
257
00:13:36,733 --> 00:13:38,300
And that is “Vibe coding”
258
00:13:38,300 --> 00:13:38,733
Yeah
259
00:13:38,733 --> 00:13:44,566
And you are
kind of an expert vibe coder
260
00:13:44,566 --> 00:13:47,800
So Stellan is vibe coding
during the evenings
261
00:13:47,800 --> 00:13:49,800
and sends all the
new tricks that he learnt
262
00:13:49,800 --> 00:13:53,066
so he’s much
more advanced than I am
263
00:13:53,066 --> 00:13:54,800
when it comes to vibe coding
264
00:13:54,800 --> 00:13:57,000
But what is five coding?
265
00:13:57,000 --> 00:14:00,433
So I don’t know who invented this
266
00:14:00,433 --> 00:14:05,066
this term, but basically it is
that you are not a developer
267
00:14:05,066 --> 00:14:07,933
with the developer background,
but you are able to code
268
00:14:07,933 --> 00:14:10,900
pretty advanced
things with these new AI tools
269
00:14:10,900 --> 00:14:14,133
So you’re kind of vibing your way
270
00:14:14,133 --> 00:14:16,700
So I guess that’s where it comes from
271
00:14:17,766 --> 00:14:21,000
And yeah,
people are doing amazing things
272
00:14:21,000 --> 00:14:26,400
especially in like games, people
are doing crazy things with games
273
00:14:26,533 --> 00:14:30,133
I also see
a whole industry popping up
274
00:14:30,133 --> 00:14:33,233
around tools to create websites,
for example
275
00:14:33,233 --> 00:14:37,100
The big issues that people have had
so far has been around security
276
00:14:37,266 --> 00:14:38,766
So people are hacking the shit
277
00:14:38,766 --> 00:14:40,433
out of these things, basically
278
00:14:40,433 --> 00:14:43,600
And some just laugh
at it, and say like, well
279
00:14:43,600 --> 00:14:49,800
“Someone put some crazy thing in
my game, and I’m trying to solve it”
280
00:14:49,800 --> 00:14:55,000
and other people go like,
“Okay, this is genuinely a problem”
281
00:14:55,000 --> 00:14:56,966
But I’m pretty convinced that
282
00:14:56,966 --> 00:15:01,166
this again goes back to like,
what kind of prompts
283
00:15:01,166 --> 00:15:04,066
are you giving to the system?
What are you asking it to do?
284
00:15:04,066 --> 00:15:06,666
Are you asking it to do
these kind of security checks,
285
00:15:06,666 --> 00:15:09,633
or are you asking it to do refactoring,
and things like that?
286
00:15:09,633 --> 00:15:12,333
What I found just from my
hobby vibe coding
287
00:15:12,333 --> 00:15:16,200
is that most people who do this well
288
00:15:16,200 --> 00:15:20,400
they have a process for this,
so it’s just like
289
00:15:20,400 --> 00:15:23,466
I would imagine
a bit of a development process
290
00:15:23,466 --> 00:15:26,900
where you have certain steps
that you do that you approach
291
00:15:26,900 --> 00:15:30,500
Whereas if you
just randomly prompt it
292
00:15:30,500 --> 00:15:34,366
it’s not going to end well,
but you can have a process for it
293
00:15:34,366 --> 00:15:38,933
And that process,
people use ChatGPT or Grok
294
00:15:38,933 --> 00:15:42,266
to do this kind of development,
as a step by step process
295
00:15:42,266 --> 00:15:43,933
And they ask you to create the process
296
00:15:43,933 --> 00:15:46,066
They ask you to create the documentation
297
00:15:46,066 --> 00:15:50,333
They ask you to create, you know, folder
structure, necessary checks, etc.
298
00:15:51,400 --> 00:15:53,066
and also create the prompts
299
00:15:53,066 --> 00:15:55,833
And then,
then they use that in whatever tool
300
00:15:55,833 --> 00:15:59,733
that they are working with, whether
it’s Cursor (AI code editor) or anything else
301
00:15:59,733 --> 00:16:04,600
But the thing you are talking about
though, I think it’s quite unfamiliar
302
00:16:04,600 --> 00:16:08,433
or even foreign for people who
don’t hang out on X (ex-Twitter)
303
00:16:08,433 --> 00:16:11,800
I don’t see these kind of things
on LinkedIn, for example
304
00:16:11,800 --> 00:16:13,800
And I’ve been hanging
out on Twitter
305
00:16:13,800 --> 00:16:17,966
or “X” now, for a long time,
and I know that
306
00:16:17,966 --> 00:16:21,200
people are much faster
at adopting a technology on X
307
00:16:21,200 --> 00:16:22,733
So many early adopters
308
00:16:22,733 --> 00:16:26,833
And the games that you’re talking about,
they’re often done by indie hackers
309
00:16:26,833 --> 00:16:29,933
And solo entrepreneurs
and they are usually so quick
310
00:16:29,933 --> 00:16:35,633
to jump on new technology
and try to earn money on them
311
00:16:36,033 --> 00:16:37,200
and for them, like,
312
00:16:37,200 --> 00:16:41,066
they don’t have to think about the quality
and code maintenance at all
313
00:16:41,066 --> 00:16:44,400
You know, one of the best indie hackers
314
00:16:44,400 --> 00:16:48,300
The ones who made the game
— the flying game
315
00:16:48,300 --> 00:16:52,266
He only codes with jQuery (JS library)
jQuery and PHP (backend)
316
00:16:52,266 --> 00:16:56,400
And he launched so
many projects over the years
317
00:16:56,566 --> 00:16:57,266
But…
318
00:16:57,966 --> 00:17:02,466
How about Zooma?
Will vibe coding have a place
319
00:17:02,600 --> 00:17:05,333
outside of your late evenings?
320
00:17:06,000 --> 00:17:06,733
I think one thing
321
00:17:06,733 --> 00:17:10,133
that we have already done
a bit of is actually prototyping
322
00:17:10,133 --> 00:17:10,833
You know
323
00:17:10,833 --> 00:17:13,466
You move faster from a wireframe
324
00:17:13,466 --> 00:17:16,933
to an actual working prototype
so you can test all the interactions
325
00:17:16,933 --> 00:17:20,133
and what you actually want to happen
when you click this button,
326
00:17:20,133 --> 00:17:22,733
what is actually going to happen,
and so on
327
00:17:22,733 --> 00:17:24,833
You can simulate that much faster
328
00:17:24,833 --> 00:17:27,833
actually, I think, with AI tools
329
00:17:27,833 --> 00:17:30,600
I think for me, the thing would be
330
00:17:30,600 --> 00:17:38,266
If designers could get into this workflow
so they can test out their ideas faster
331
00:17:38,266 --> 00:17:39,366
Yeah
332
00:17:39,366 --> 00:17:42,133
I don't think we
have necessarily cracked it
333
00:17:42,133 --> 00:17:48,233
So we move from Photoshop to Figma,
and then you do static prototypes
334
00:17:48,233 --> 00:17:50,233
You add more functionality to prototypes
335
00:17:50,233 --> 00:17:54,200
The more functionality you add,
the more complicated it becomes
336
00:17:54,200 --> 00:17:58,100
Certain designers
will just say “Stop! This is my line here”
337
00:17:58,100 --> 00:18:02,933
“I’m not gonna do more. I just
wanna draw my designs”
338
00:18:02,933 --> 00:18:06,433
So the dream would be
if they could actually test out
339
00:18:06,433 --> 00:18:10,333
their designs
without the help of developers
340
00:18:10,333 --> 00:18:11,900
Yeah
341
00:18:11,900 --> 00:18:13,000
Or what do you think?
342
00:18:13,000 --> 00:18:15,600
Yeah, I genuinely think so as well
343
00:18:15,600 --> 00:18:17,300
That it's like, what do we want?
344
00:18:17,300 --> 00:18:22,366
Well, we don’t want AI generated
websites and apps and things like that
345
00:18:22,366 --> 00:18:25,666
that look like…
everything looks exactly the same
346
00:18:25,666 --> 00:18:29,533
We want this
sort of creative element to it
347
00:18:29,533 --> 00:18:32,600
So far,
humans are better at that
348
00:18:32,600 --> 00:18:36,000
Eventually, you know,
AI might be really good at this as well
349
00:18:36,000 --> 00:18:37,833
with the right input
350
00:18:37,833 --> 00:18:39,833
I’m actually seeing AI more as of
351
00:18:39,833 --> 00:18:43,100
like another teammate in the room
352
00:18:43,100 --> 00:18:44,200
than anything else
353
00:18:44,200 --> 00:18:47,033
So if you just imagine an empty chair
354
00:18:47,033 --> 00:18:48,933
and there’s an AI in it
355
00:18:48,933 --> 00:18:50,433
And I think that
356
00:18:50,433 --> 00:18:53,800
eventually this
might lead us to have a more
357
00:18:53,800 --> 00:18:59,100
of a conversational relationship
with AI and that ultimately helps us
358
00:18:59,433 --> 00:19:03,133
you know, develop things
that are higher quality, faster
359
00:19:03,133 --> 00:19:06,966
As you mentioned,
there’s a process to prompting
360
00:19:06,966 --> 00:19:09,933
It’s kind of straightforward, you
use natural language, but
361
00:19:09,933 --> 00:19:12,066
but it’s still not as
362
00:19:12,066 --> 00:19:17,466
For example, for a designer,
it’s not as straightforward as drawing
363
00:19:17,466 --> 00:19:19,133
You have to describe your design
364
00:19:19,133 --> 00:19:24,366
But when it comes to
design, you want to create it visually
365
00:19:24,366 --> 00:19:26,800
The same with development
366
00:19:26,800 --> 00:19:30,700
So if you get more advanced,
now you just described a process
367
00:19:30,700 --> 00:19:33,466
of how to prompt, and the structure things
368
00:19:33,466 --> 00:19:35,700
I don’t think anybody do it
369
00:19:35,700 --> 00:19:37,166
There’s a certain line where
people will say
370
00:19:37,166 --> 00:19:38,800
“Nope, that’s not what I do”
371
00:19:38,800 --> 00:19:40,033
“That’s not how I work”
372
00:19:40,033 --> 00:19:43,000
We might be looking
at this specific person
373
00:19:43,000 --> 00:19:44,066
So
374
00:19:44,066 --> 00:19:48,066
Will we eventually hire someone
who is like prompt engineer
375
00:19:48,066 --> 00:19:51,233
Like ten times
better than the normal person
376
00:19:51,533 --> 00:19:54,366
Well, it’s hard to tell
377
00:19:54,366 --> 00:19:58,566
at the moment, it’s a lot of talk
about, like, a new profession
378
00:19:58,566 --> 00:20:00,233
to be able to do that
379
00:20:00,233 --> 00:20:01,500
But, yeah, definitely.
380
00:20:01,500 --> 00:20:05,766
I think there are some skills going on,
in relation to that
381
00:20:05,766 --> 00:20:07,633
and that it’s constantly evolving
382
00:20:07,633 --> 00:20:09,533
It’s like every day
there’s something new
383
00:20:09,533 --> 00:20:12,966
Just last week
I learned that you can use JSON
384
00:20:12,966 --> 00:20:16,600
to prompt ChatGPT
for image generation, for example
385
00:20:16,600 --> 00:20:19,733
So you basically, with
code, you tell it, you know
386
00:20:19,733 --> 00:20:21,866
what kind of copy you want in the image
387
00:20:21,866 --> 00:20:25,333
where you want it, where
you want to place certain items
388
00:20:25,333 --> 00:20:28,266
the type of person that
should be in the image, and so on
389
00:20:28,266 --> 00:20:31,133
And it has
a better understanding of that
390
00:20:31,133 --> 00:20:33,133
than if you just write it
391
00:20:33,133 --> 00:20:35,466
So if you just typed the description…
392
00:20:35,466 --> 00:20:36,600
Yeah, exactly
393
00:20:36,600 --> 00:20:39,166
It’s markup in a way
394
00:20:39,166 --> 00:20:44,133
Again, it’s like , for these guys
that are like coding games for example
395
00:20:44,133 --> 00:20:46,166
Most of them seem to use Grok
396
00:20:46,166 --> 00:20:49,200
to create the instructions
and the prompts
397
00:20:49,200 --> 00:20:51,633
for example, for Cursor (AI editor)
398
00:20:51,633 --> 00:20:53,700
So they don’t just use a single tool
399
00:20:53,700 --> 00:20:56,133
They use Grok as a foundation
400
00:20:56,133 --> 00:20:58,966
for the idea of the game,
and the prompts and the process
401
00:20:58,966 --> 00:21:02,200
and everything, and then they
use that in Cursor
402
00:21:02,200 --> 00:21:05,033
So I think that’s also, like,
to find your process
403
00:21:05,033 --> 00:21:07,300
is also part of
what tools are you are using
404
00:21:07,300 --> 00:21:08,700
and for what
405
00:21:08,700 --> 00:21:11,966
Cuz it’s a combination of things
at the moment
406
00:21:12,433 --> 00:21:14,933
When I think about it,
it’s kind of sad, because
407
00:21:14,933 --> 00:21:17,033
you described the
whole process of using Grok
408
00:21:17,033 --> 00:21:20,133
to bounce ideas,
and write design documents,
409
00:21:20,133 --> 00:21:22,100
to do design and programming
410
00:21:22,100 --> 00:21:23,600
It used to be
a whole team of people
411
00:21:23,600 --> 00:21:26,566
Where you bounce and talk,
and now it’s just one person
412
00:21:28,300 --> 00:21:33,266
I saw a list somewhere of the
startups over the last year or so
413
00:21:33,266 --> 00:21:35,833
and the size of the teams,
they’re really small
414
00:21:35,966 --> 00:21:37,900
It’s like 10 to 20 people
415
00:21:37,900 --> 00:21:41,300
in these teams that are developing
this sort of SaaS tools
416
00:21:41,600 --> 00:21:43,800
And I also saw that someone in
Y Combinator
417
00:21:43,800 --> 00:21:46,166
the incubator in Silicon Valley
418
00:21:46,166 --> 00:21:50,900
known for a lot of famous companies having
gone through that program
419
00:21:50,900 --> 00:21:55,466
They said that like a third
of all applicants for the last round,
420
00:21:55,466 --> 00:22:00,433
did something like
95% of their coding with AI tools
421
00:22:00,433 --> 00:22:03,366
Didn’t write more
than 5% of the code themselves
422
00:22:03,366 --> 00:22:04,733
So
423
00:22:04,733 --> 00:22:06,766
Yeah, that’s the
direction where we’re headed in
424
00:22:06,766 --> 00:22:10,000
You’re kind of turbo changing
what you able to do
425
00:22:10,000 --> 00:22:12,533
I don't think it's gonna
lead to mass unemployment
426
00:22:12,533 --> 00:22:15,000
I’m kind of optimistic about it,
I think we’re gonna see
427
00:22:15,000 --> 00:22:19,200
more tools and more new cool stuff
than we saw before
428
00:22:20,366 --> 00:22:21,400
As a manager,
429
00:22:21,400 --> 00:22:25,166
what do you think is a good advice
to future developers
430
00:22:25,166 --> 00:22:27,600
who are, “afraid”?
431
00:22:27,600 --> 00:22:30,033
because I myself,
I wouldn’t advice someone
432
00:22:30,033 --> 00:22:33,100
“Let’s study development”
Like “it’s a good!”
433
00:22:33,100 --> 00:22:35,333
I think if you’re interested
434
00:22:35,333 --> 00:22:38,600
as we said before,
this is not going to be a problem
435
00:22:38,600 --> 00:22:42,200
This is going to be a the door opener
to so many exciting
436
00:22:42,200 --> 00:22:45,466
new opportunities
that we have not seen before
437
00:22:45,466 --> 00:22:50,233
We’re sort of on the breakthrough
in a lot of areas related to AI
438
00:22:50,233 --> 00:22:52,233
Whether it’s self-driving cars
439
00:22:52,233 --> 00:22:54,600
or whether it’s in robotics
440
00:22:54,600 --> 00:22:56,633
or in science in general
441
00:22:56,633 --> 00:22:59,466
What we can develop vaccines for
442
00:22:59,466 --> 00:23:01,633
and find a cure for cancer
443
00:23:01,633 --> 00:23:03,100
But you say “interested”
444
00:23:03,100 --> 00:23:06,866
So if you just want to learn
web development, or development
445
00:23:06,866 --> 00:23:10,666
just because you want
to earn good money, and have a
446
00:23:10,666 --> 00:23:13,200
flexible work time… is that…
447
00:23:13,200 --> 00:23:17,000
I think it’s a bad idea in general
to choose your profession
448
00:23:17,000 --> 00:23:20,133
after the potential salary
you might have
449
00:23:20,133 --> 00:23:23,000
I think you need to do something
that you’re really interested in
450
00:23:23,000 --> 00:23:24,500
Cuz if you just do it for the money
451
00:23:24,500 --> 00:23:27,566
that it’s like impossible to say
what's going to be
452
00:23:27,566 --> 00:23:29,166
worth your while, so to speak
453
00:23:29,166 --> 00:23:30,533
but again, like, if we take
454
00:23:30,533 --> 00:23:33,100
ATMs, for example came around,
and everyone says that
455
00:23:33,100 --> 00:23:36,000
Everyone working
in banks will be unemployed”
456
00:23:36,033 --> 00:23:38,933
Well, in fact, that didn’t really happen
457
00:23:38,933 --> 00:23:40,600
It’s like more people employed
in banking
458
00:23:40,600 --> 00:23:43,600
than ever before,
but you’re just doing different things
459
00:23:43,600 --> 00:23:45,566
You’re not
servicing people coming in
460
00:23:45,566 --> 00:23:48,566
asking for cash over the counter
461
00:23:48,566 --> 00:23:51,166
ATMs solved that problem a long time ago
462
00:23:51,166 --> 00:23:54,433
But you moved on to do more advanced work
463
00:23:54,433 --> 00:23:57,133
And I think it’s similar here
464
00:23:57,133 --> 00:24:00,400
It’s also like the industralisation
465
00:24:00,400 --> 00:24:05,600
Going back to the 1800s
and 1700s in England, and so on
466
00:24:05,600 --> 00:24:08,933
It’s like people
moving from farming to industry
467
00:24:08,933 --> 00:24:12,233
and then eventually from the industry
into white collar and so on.
468
00:24:12,233 --> 00:24:15,233
It’s an always evolving landscape
469
00:24:15,233 --> 00:24:17,533
But if you’re be interested
in what you’re doing
470
00:24:17,533 --> 00:24:19,133
you’re hopefully really good at it
471
00:24:19,133 --> 00:24:21,933
and then you’ll be able to adapt
472
00:24:22,233 --> 00:24:25,466
So that’s our optimistic key takeaway
473
00:24:25,466 --> 00:24:27,933
You should do what you think is fun
474
00:24:27,933 --> 00:24:31,266
What you’re interested in,
a not go for tech because you…
475
00:24:32,200 --> 00:24:36,266
Because it sounds good!
No, that’s a bad idea!
476
00:24:36,300 --> 00:24:37,400
Okay
477
00:24:37,400 --> 00:24:40,333
Thank you for the
fantastic discussion, Stellan
478
00:24:40,333 --> 00:24:41,633
We need to do this again
479
00:24:41,733 --> 00:24:44,033
And thank you, viewers, for watching
480
00:24:44,033 --> 00:24:47,033
Until next time, keep learning
481
00:24:47,200 --> 00:24:50,166
Bye bye — See you
482
00:24:52,866 --> 00:24:54,533
We kept the time, right?
483
00:24:54,533 --> 00:24:57,100
Yeah. So smooth.
484
00:24:57,100 --> 00:24:58,133
So smooth.
485
00:24:58,133 --> 00:25:01,100
That needs to be our first episode
486
00:25:01,100 --> 00:25:02,533
as well.
487
00:25:02,533 --> 00:25:04,333
So it's very intense.
488
00:25:04,333 --> 00:25:06,733
We. Oh.
489
00:25:06,733 --> 00:25:08,033
Okay. All right.
