शैडोइंग अभ्यास: Maybe we were wrong - वीडियो के साथ अंग्रेजी बोलना सीखें

The last couple years of being a developer has been,

you know, I guess kind of interesting, right?

Fellas, I actually have some good news here, okay?

Things have been changing.

Some whispers have been halving and new news is hitting the front lines.

It turns out being a software developer might actually be pretty dang awesome.

Now, if you haven't been in the software development realm for the last three years,

let me do a quick recap for you.

First off, you shouldn't teach your kids to code because honestly,

you should let AI do

that says ai salesman number one you've said ai could wipe out half of all entry-level white-collar jobs

and spike unemployment to 10 to 20 percent in the next one to five years yes

and of course dario here always just giving it to you

straight by the way look at how happy he is he

is answering this question with such alacrity he has never been

more excited to tell you you are about to lose your

job than i've seen any man on earth this guy loves

firing people he loves firing people not even from his own company and

if that's not enough for you of course last year we

had sam altman sam gyppity altman just dropping the death star

the day before big product announcements just letting everybody know hey what we're about to do it's pretty big

and of course this is a very confusing message is he saying he's the bad guys is

that what he's saying is does he release the death star

i'm still confused to this like this post actually eats like

it's an earworm okay it eats me from the inside like

what the hell happened here why why why would you post this

but the thing i wanted to to talk about this this

turning point that's happening is this right here look at this

just in sam altman says ai probably won't trigger the job apocalypse he once predicted not only

that but uber coo now says hey

when you're spending a whole bunch of tokens honestly can't really tie it to features

so there's something just kind of happening in our world there's

some maybe some news realizing how important it is to have competent

and good developers it's like the timeline is almost kind of oddly healing in a way.

So we, of course, we absolutely have to go over this.

I got a lot of yeah.

Okay.

The vindication gap is going to land on this video,

but more importantly, I want to do a quick thank you to the sponsors.

Hey, is that HTTP?

Get that out of here.

That's not how we order coffee.

We order coffee via SSH terminal.shop.

Yeah.

You want a real experience.

You want real coffee.

You want awesome subscriptions.

So you never have to remember again.

Oh, you want exclusive blends with exclusive coffee and exclusive content then check out cron you don't know what ssh is

well maybe the coffee is not for you

okay first let's talk about sammy jippity altman right here saying

hey maybe he was wrong here's kind of exactly what he

said this of course is from a commonwealth bank of australia

conference in sydney just last tuesday it says this i don't think we're gonna have the kind of jobs apocalypse

that some of the companies in our space advocate

or talk about by the way i love the fact

that he says advocate like there's actual companies in his space

advocating for everybody to lose their jobs you know who we're

talking about okay you know who sammy boy's talking about he's

not talking about you know minstrel ai the bro's talking about anthropic he's just like yo anthropic

they really want you to lose the jobs here it could

also be to be fair this could also be some sort

of jab at musk as well i can't really tell

but either way very hilarious i thought there was going to

be more impact on entry-level white-collar jobs being eliminated by now

than has actually happened i now think i understand more about why it hasn't

and i'm obviously grateful but

that is an area where my intuitions were just off we

really do care about our interactions with people he added

which he said for better or for worse updated me to thinking

that the jobs picture is likely going to be very different

than we thought okay whoopsie poopsie turns out maybe people like to talk to people

and interact with people

and building great products actually turns out maybe just might take people might take people to do

that huh crazy i i just thought we were gonna get

metered intelligence i thought i thought you stole the internet took

everything compressed it down into the sweet next token prediction

and then sell us back to us and then we pay for it

and it's just like that

and that's how the world works of course sam's actually not

the only one goldman sachs ceo david m solomon says

that ai won't eliminate 25 of the jobs he's just saying

hey it turns out people will spend their time in more productive ways

and he makes some cases down here saying hey did like any other technology ever do this

and what we're observing is not true in the data

so this is actually kind of a positive development i'm actually

pretty excited about this i would love to see things actually

healing from this craze where you have all these ceos

that have been told for the last two years of these

companies like no bro bro i'm just letting you know you

gotta just like dude just like fire everybody you don't need them

because you got ass dude you got open claw open claw

is gonna do everything you just say yo million dollar sass no mistakes

and bam in just a few minutes crunching some clod four points actually 4.6 4.7 is kind of

you're gonna have yourself a million dollar sass.

I am so happy that this is starting to crumble

because this is something that has needed to crumble now for a while because it is ridiculous.

There is so many people right now that at their jobs are being forced to like a certain level of AI output.

And if they don't make it,

100

they're like heads on the chopping block,

101

even if it's not a good idea.

102

Like they don't even get the option of going,

103

hey, is this good or bad?

104

It's like, no, use it or that's bad.

105

Like if you don't use it, that's bad.

106

I was like, well, maybe it's not always the best thing.

107

And where this is really starting to show is this right here,

108

this Uber COO saying

109

that heavy AI spend is actually getting harder to justify as higher token usage fails to show clear payoff in consumer features.

110

Now, this of course is a headlining item.

111

So let's actually watch the video as opposed to,

112

you know, seeing what Twitter has to say.

113

Let's look at the actual video.

114

Let's hear what he has to say.

115

I know, novel concept, right?

116

You make your head explode, right?

117

When you hear companies talking about,

118

hey, 25% of code commits over the last quarter were AI driven.

119

Or, you know, our token usage went from X to Y or percentage of employees,

120

you know, all these sort of numbers. And it's amazing.

121

And I think it's like this massive transformation of society.

122

But then you sometimes go and you talk to your senior engineering leaders and you're saying,

123

okay, How many projects that were on the cutting room floor got moved above the line because of the,

124

you know, productivity gains because 25% of our code commits were via cloud code last, last quarter.

125

That link is not there yet.

126

Right?

127

Like you're not, I mean,

128

I think maybe implicitly there there's more that is getting shipped,

129

but it's, it's, it's very hard to draw a line between one of those stats and okay,

130

now we're actually producing like 24,

131

5% more useful consumer features.

132

And this is the COO of Uber.

133

Like this guy has fully committed to this AI kind of adventure.

134

In fact, they spent within the first four months,

135

their operating year budget for AI.

136

Like these people were committed.

137

And now after all this,

138

they're like, okay, wait, wait, wait, wait, wait.

139

Maybe not everything that glitters is gold.

140

Perhaps there are some things that we don't really see.

141

These implicit features that he was talking about.

142

Maybe there's more bug fixes,

143

which I'm a little bit dubious on that one.

144

But nonetheless, like maybe things have been changing a little bit and now we are shipping a bit faster,

145

but there's no material impact.

146

What we thought would be these huge,

147

gigantic windfalls are, in fact,

148

just not showing up in the actual data.

149

And this, of course, gets me to the linear CEO post,

150

which I really think just nails this whole thing,

151

which is we keep hearing about the 10x or the 100x productivity gains in engineering and knowledge work.

152

But outside of Model Labs,

153

I haven't seen the corresponding 10x to 100x revenue growth across the market or increase in quality.

154

Where is the productivity going?

155

This is so good because you know where the productivity is going.

156

All these people talking about this just incredible increase in productivity,

157

what they're really saying?

158

They're like, dude, bro, I've just set up the greatest harness ever.

159

Oh my gosh, you will not believe how many different versions of Cloud Code I have running.

160

Like I got G stack on top of the P stack.

161

I got all the stacks going so hard.

162

I got TRT coming in hot.

163

Like, we're doing everything.

164

I'm gonna make the most crazy commits ever.

165

Like, I've just spent the last three months sharpening the sword.

166

Honestly, it reminds me a lot of NeoVim config enthusiasts.

167

Now, remember, back in the day,

168

maybe like a year ago to five years ago,

169

maybe this guy got made fun of a lot.

170

You know why he got made fun of?

171

Because, oh, I was always playing around with my NeoVim config.

172

I wanted the perfect experience,

173

and everyone was like, how could you do that?

174

Now, what's, it's, I swear today is just the same story again,

175

which is, oh, you know what we really need?

176

We need to make the greatest agent harness.

177

Like that's what we're actually missing, bro.

178

Bro, it's just one more harness and then we'll actually be able to write infinite,

179

free, high quality code forever.

180

It'll be so fast.

181

It'll be so amazing.

182

Just one more harness, bro, bro.

183

It's just like, I'm almost there.

184

My system prompt is like 98%.

185

It's just so close.

186

Also, I do love the,

187

like this line right here,

188

because if you really think about what he's saying is,

189

hey, it's all the labs.

190

They're making like 10 to 100 X more money.

191

Where's everybody else's money from all these AI productivity gains, huh?

192

Because if everybody really is 10 to 100x more productive,

193

wouldn't that show up in a bunch more revenue?

194

Kind of crazy that the only people making money are the people selling shovels, huh?

195

I mean, they are seeing 10 to 100x revenue.

196

Jeez, that seems a little strange, isn't it?

197

See, the thing that I think people are missing about this whole AI craze and all that is that,

198

sure, there's a bunch of people that go zero to one with AI,

199

and they're like, whoa, bro,

200

this is the coolest thing ever.

201

But there's a bunch of us who want to go,

202

say, from 30 to 60, right?

203

We're actually in the further part of the project.

204

And so what I end up personally doing,

205

which is something I do right here,

206

is I go off and I talk about everything I want to build to the AI.

207

And then I have it implement a bunch of different versions of it.

208

And I go, okay, well,

209

what if we change this?

210

How about we put all the responsibility over here?

211

And I can see like five different versions of the exact same thing go, huh?

212

Okay, I like this.

213

Yeah, I like that.

214

Okay, I like this, like,

215

it is demonstrably helping me see a bunch of different versions to really understand what makes a good interface,

216

but it's not this like 100x improvement.

217

I'm not even sure if I'm actually improving the speed in which I'm actually making these features.

218

Maybe my understanding is a bit faster.

219

Maybe the ideas I'm coming up with are a little bit more solidified.

220

But maybe also, just sitting down with a pen

221

and piece of paper and drawing out my ideas would have made the exact same impact.

222

See, the thing that I think most people are missing people that actually instead building say the zero to one projects, right?

223

There's a whole bunch of like flashy kind of shallow projects that people make.

224

And they're like, see, yo,

225

bro, this is so fantastic.

226

No, like trying to make deep,

227

integrated products that actually take a long time to think about

228

and actually take a long time to come up with good ideas.

229

When you use AI, it increasingly makes things more brittle and more difficult.

230

No matter what you're doing,

231

you have to take a step back and actually work through the problems yourself.

232

And this is one thing that I've really appreciated about AI,

233

I can come up with an idea,

234

I can kind of type out what it should be.

235

And then I can say,

236

hey, go build me a mock version,

237

I can look at the mock version,

238

go, okay, yeah, I hate all these parts.

239

This is just going to be a nightmare to maintain,

240

this is going to be a source of bugs, let's change things up.

241

How about you do it this way?

242

Then I can look at it again,

243

okay, I actually really like how this is looking,

244

but I would really like this thing moved over here,

245

I'd like this, and then look at it again.

246

And so you can see this right here.

247

This is me working through how would animations look in an immediate mode UI?

248

Well, now I've had three swings at it and I go,

249

okay, I actually really like this.

250

This actually looks pretty good.

251

Now I get to implement the thing.

252

And this is what I think we're seeing is that

253

when you really use this as a means to actually help you build real software,

254

there's still just so much decision making and time spent thinking about how things should actually look that yeah,

255

exploration is cheaper than it's ever been before,

256

but actual implementation still feels really heavy.

257

And if anything, making good decisions now almost feels harder,

258

because of how fast you can move.

259

Anywho, I just wanted to talk about this for a little bit,

260

because I feel like there's just there's so much in our kind of world right now that's it's doom and gloom.

261

Or it's just absolute hyping.

262

It's like no AI is the worst thing.

263

No, that's clear.

264

Clearly, that's not real.

265

Or B, it's like AI is the greatest thing that has ever existed.

266

Also clearly not real.

267

Like yes, it is absolutely magic.

268

You can describe with practically broken English and have something come out that approximately looks like what you're describing.

269

Like honestly, truly, that is a magical human feat.

270

I hope that I never stopped being amazed by that because honestly, that's wild.

271

You could never have convinced me six years ago,

272

like that was going to happen in my immediate future.

273

It's just one of the yap about something maybe a bit more positive that,

274

hey, actually, I think a lot of good things are still ahead of us.

275

I still think we got a lot of good time.

276

And I still think competency is actually extremely important in learning how to program and being a good programmer is still,

277

in fact, really important.

278

Hey, the name is the Primogen.

शैडोइंग तकनीक क्या है?

शैडोइंग (Shadowing) एक विज्ञान-समर्थित भाषा सीखने की तकनीक है जो मूल रूप से पेशेवर दुभाषिया प्रशिक्षण के लिए विकसित की गई थी। विधि सरल लेकिन शक्तिशाली है: आप मूल अंग्रेज़ी ऑडियो सुनते हैं और तुरंत इसे ज़ोर से दोहराते हैं — जैसे वक्ता की छाया 1-2 सेकंड की देरी से। शोध से पता चलता है कि यह उच्चारण सटीकता, स्वर, लय, जुड़ी हुई ध्वनियाँ, सुनने की समझ और बोलने की प्रवाहशीलता में काफ़ी सुधार करता है।

शैडोइंग अभ्यास: Maybe we were wrong - वीडियो के साथ अंग्रेजी बोलना सीखें

इस पाठ के बारे में

मुख्य शब्दावली और वाक्यांश

अभ्यास के सुझाव

शैडोइंग तकनीक क्या है?