Luyện nói tiếng Anh bằng Shadowing qua video: Mark Zuckerberg on AI Glasses, Superintelligence, Neural Control, and More

Chọn chế độ luyện tập

Mỗi chế độ mở một phòng luyện tập toàn màn hình.

663 câu

Nếu các câu quá ngắn hoặc quá dài, hãy bấm Edit để chỉnh sửa.

Somewhere between one and two billion people in the world who have glasses for vision correction today.

Within five to seven years, is there like any world where those aren't all replaced with AI glasses?

The neural band is going to be personalized to you over time.

It's not actually picking up motion.

It's picking up your muscles firing.

A lot of people who are designing technology, they make something that's kind of bulky, and they add a bunch of cool functionality to it, and they're like, why don't people want to wear this?

It's kind of a mantra internally that we have, which is this idea that AI is going to be the most important technology in our lives.

We want to make sure that AI is in service of people, not just like something that sits in data centers and automates stuff.

There were rumors that you were like personally going founder mode, calling researchers.

What drove the decision for this?

LAMA 4 was not on the trajectory that I thought it needed to be on.

Like why am I going out and meeting all of the top AI researchers?

It's like, well, like I want to know who the top AI researchers are

and I want to like personally have relationships with them.

I moved everyone around me who sat with me and now the lab is there.

So, I mean, I'm pretty hands on.

Every time I think of what a milestone would be in AI, they all seem to get achieved sooner than we think.

Thanks so much for being here.

Yeah, good to see you.

Thanks for doing this.

So today we're talking everything MetaConnect 2025.

Can you give us the rundown of kind of everything you're announcing and what you're personally most excited for?

Yeah, so the main things that we announced at Connect were our fall 2025 line of glasses.

and um you have them here so um so we can just start

and go through them i mean the first one is the next generation of ray-ban meta

so that's sort of the classic ai glasses that we've shipped

so far like they're some of the most uh you know

fastest growing consumer electronics of all time we're very happy with

that um and a lot of the improvements

that we have for these are you know we doubled the

battery life we we have 3k video resolution in them now for um for capture

and we're introducing new ai features like this this thing conversation focus

which um allows you

if you're in a loud place um to basically turn up the volume on friends who you're talking to

so if you're in a restaurant or something like that then that i think is gonna be really neat um

then we've got this guy which is the oakley meta vanguards this is our second collab um with oakley

it's um it's more kind of a performance glasses

and um these i think are really cool for a number of things

and they're designed for for a number of things around sports um you've got the camera that's centered

so that's great for alignment it's wider field of view louder

speakers um they're water resistant um they can you know just

just a bunch of they can connect with your garmin watch So you can be running a marathon and basically,

you know, like you can tell it every mile, capture a video.

And then at the end, it'll produce a video for you where it stitches them all together and puts your stats from Garmin on top of it.

So I think that's a pretty neat thing.

But I think the most interesting thing by far that we announced is this one, which we call Ray-Ban, Meta Ray-Ban Display.

and that is because it is the first AI glasses that we have shipped that have a high resolution display in them.

And the other big breakthrough is that they pair with this guy,

the meta neural band, which is the first mainstream neural interface that we're shipping as the way to control them.

And I know that's pretty neat.

I mean, basically every computing platform has its own new input method, right?

So when you went from computers with the keyboard and mouse to phones with touchscreens, you kind of got a completely new input method.

And the same, I think, is going to be true for glasses.

And having a neural interface where you can just send signals from your brain with micro muscle movements like this.

I mean, this is basically about as much movement as you need to make.

And you can, you know, enter text neurally.

just all the fun stuff that you got a chance to try out.

But I think that this is a pretty big breakthrough.

So I'm very excited about this.

But overall, I mean, the whole lineup is good.

So it was a fun connect.

So you got the Ray-Ban metas for kind of everyday, the Oakleys for athletes, and then the displays now for power users.

How do all these glasses kind of tie into that personal super intelligence vision?

Yeah, so, I mean, our theory is

that glasses are the ideal form factor for personal super intelligence

because it's the only real device that you can have that can see what you see, can hear what you hear, can talk to you throughout the day,

and can generate a UI in your vision in real time.

And so there are other devices that people have.

I mean, obviously you can have some AI on your phone, you can do some AI on your watch, um you can do some of it

if you just have air like airpods type type thing

but um but i think glasses are going to be the only thing

that can that can do all of the pieces

that i just said kind of visual and audio in

and out um and that

that i think is going to be is going to be a really big deal

so i got to try it before this interview yeah

and i have to say there were way more many use cases than I initially thought they were going to be.

Like the AirPods, for example, have the live translation, but this is live translation, the GPS, the captions, there's so much more.

Yeah, the subtitles are awesome.

It's awesome.

So I'm curious, like what are your favorite kind of use cases that you've been using them for?

I mean, the thing that I really had in mind when we were designing them for is just sending messages, right?

I mean, this is the thing that I think we do the most frequently on our phones.

I mean, I would guess that we all are sending dozens of messages a day.

I don't really need to guess.

And we run a lot of the messaging apps that people use.

And so I think we know that people send dozens of messages a day.

And we wanted to make it so that the experience is just really great.

So now with the glasses, a friend texts you, shows up in the corner of your eye for a few seconds.

It's off center.

So it doesn't block your view.

It goes away quickly.

It's not distracting.

But if you want, you can just respond as easy as, you know, moving your finger like this.

And, you know, I can write, I mean, I'm up to like 30 words a minute typing with the neural text.

So I think this is, I think this is great.

One of my kind of mental models for this is that in-person conversations are awesome.

The one thing that like about Zoom is that you can multitask, right?

It's like if we're having a conversation and you say something, I want to get some context on what you're saying, or if you remind me that I should go do something or,

100

oh, you bring up a person and it's like, oh yeah, I've been meaning to reach out to that person.

101

Like if we're having a live conversation, I'm not just going to do that right then.

102

And then the chances that I'm probably actually going to forget by the end of the the conversation, not follow up on most of those things.

103

On Zoom, I kind of feel like you can just kind of text someone off to the side and that's fine.

104

And this, I think, is going to bring that to physical conversations too, where it's like, you don't have to break your presence.

105

You don't have to like, you know, it's not rude, right?

106

You're not like breaking out a phone.

107

You're not really, I mean, you can continue to really pay attention to the person.

108

It's just like a very quick gesture with your wrist.

109

And I keep on having this phenomenon where there's some information I want to have in the middle of a conversation.

110

And normally, otherwise, I would have had to wait till the conversation

111

and text or call the person to get the information and go back to the person I was having the conversation.

112

Now I just get the information in the middle of the call.

113

I just text the person, information comes in, and then it's like, okay, you can have a much better and more informed conversation.

114

So texting, I think, is really going to be the big thing.

115

So how does a texting feature work?

116

I know there's like you can talk to MetAI, but you're saying you can actually write it down and it automatically goes to the phone. Is that?

117

Yeah, to the glasses.

118

Okay, to the glasses.

119

Yeah, with the neural band.

120

So like you just really just handwriting.

121

Yeah, I mean, it's you can kind of think about it like as

122

if you're holding a small pencil and then what like what motion you would make to write out certain letters.

123

but then what's going to happen is

124

that the neural band is going to be personalized to you over time

125

so it's going to learn okay how do you make a w

126

or a j differently from someone else and it will learn

127

and you will kind of co-evolve with it to like it'll

128

learn what the most minimal version of you making a j is

129

or no or whatever and and then the feedback loop around

130

that well you'll just be able to make like increasingly subtle

131

movements over time where it's not actually picking up motion it's picking up um your muscles firing

132

so over time i think you're basically going to be able

133

to just move your muscles in opposition to each other um not even really move your hand

134

and and be able to send messages

135

or control a ui with your hand at your side in a jacket pocket behind your back it's not

136

and this isn't um it's not hand recognition it's not uh

137

it's not the cameras on the glasses looking at what your

138

hand is doing it's this band picking up micro signals

139

that your brain is sending to your muscles before they even move

140

so you don't actually have to move you just need to

141

to be able to send the right send the signals to it

142

and it picks it up and yeah that's so the wristband reads the electrical signals before you even move Yeah.

143

Why is that important over just like adding buttons to the glasses or like improving the voice commands?

144

Well, I mean, okay, for voice, a lot of the time you're around other people, right?

145

So, like, I just think that you want a way to control your computing device that is private and discreet and subtle.

146

And one of the things that we thought about was, okay, can we get whispering to work?

147

because by the way voice does work too right you can

148

control the stuff with voice you can do voice dictation you talk to med ai with voice

149

but we just i just thought that there needed to be a way

150

on top of that that wasn't voice um for for

151

when you want to do it more subtly

152

and privately we could be having this conversation i could you know have the glasses on

153

and i get a message it comes in and okay we're talking

154

and i'm like writing it and i'm i've sent it's done i just i mean in

155

that time I could have sent a 10-word message and

156

that would be fine that's

157

so okay then it's a guard now I got my response back in

158

and okay here like maybe I'm writing another response

159

and whatever yeah okay done so I think

160

that that's gonna be I think it's gonna be awesome

161

but in terms of buttons I mean yeah

162

that works too I mean you can you can take a

163

photo by tapping the button on any of these you can

164

hold it down to take a video you can you know skip to the next track with the swipe

165

or you can start or stop music by tapping on it

166

but um but i think the point is you don't want to you don't necessarily want

167

that to be the only way

168

that you could do it like now with the neural band

169

i can also just be like i don't know start

170

and stop music with it with my hand on my side

171

it's like all right like you like it's very subtle gesture

172

like you can't even tell right yeah we were joking in the demo room

173

that it's almost like right now

174

when you go go to classes all the teachers are like oh put your phones down they can see you

175

and like students are like just checking their phones distracted

176

but now it's going to be like hey like keep your thumbs out almost

177

because it's like i think it'll be take your glasses off

178

or take your glasses off and show your thumbs maybe both yeah

179

but uh yeah so looking ahead i guess um what you're saying is like typing

180

and speaking might not be enough to give enough data to ai um what do you mean typing

181

and speaking might not be enough to give well now we

182

have electrical signals uh we can also give to to the class.

183

Yeah, for the UI.

184

Yeah, for controlling the UI.

185

Do you think the neural band is almost like a new interface to interact with AI?

186

Yeah, I mean I think it's AI and it's the glasses.

187

I mean there's some parts of the glasses that are like, okay, you just have a menu of your apps

188

and you basically can just swipe through it very subtly with your hand and it's like, okay, now I've selected, okay, now I've picked up that.

189

This is basically what the interaction looks like.

190

Effectively, it's like, okay, swipe a couple of times.

191

I've brought up the camera.

192

Oh, now I have a viewfinder, so I can see exactly what the image is.

193

I wanna zoom in.

194

I just turn my finger a little bit.

195

Actually, that's one of the coolest, is when you're playing music,

196

the way that you change the volume is you just pretend that there's a volume dial and you just turn your wrist.

197

And it's like, it just feels like magic.

198

It's like really fun.

199

The way you can bring up Meta AI anywhere, you just tap twice.

200

It's like, but it's like, but I mean, tiny, right?

201

It's like, it could be your side, like no one would even notice.

202

So yeah, no, I think it's, I think this is just gonna be by far the most intuitive way.

203

But if you don't have it, yeah, you can control them with voice, you can control them with the buttons.

204

I mean, different ways to do it.

205

Yeah, so it's gonna be both the AI and the UI of glasses, and potentially over time, other things beyond glasses, I think that you'll be able to control with the neural band.

206

That's really fascinating.

207

Yeah.

208

How long did it take you to pick up the controls?

209

For the neural band?

210

Yeah, for the neural band.

211

Not that long.

212

I mean, it's pretty intuitive.

213

I mean, the way that you kind of think about it is navigating.

214

It's kind of like a D-pad.

215

You kind of swipe your thumb right, left, up, down.

216

You bring up the menu by kind of, you can tap twice.

217

You bring up MetAI by tapping twice you type as if you have a, or enter text as if you have a pencil.

218

The trick is that when you're learning up front, the motions are a little bit bigger.

219

And then very quickly, within a few days of using it, you get to like very subtle movements.

220

And the reality is, is that a lot of the optimization isn't going to be on our side or your side.

221

It's going to be the machine learning system over time getting personalized to you and getting better.

222

and you know it's the the times

223

when the the neural text entry is slow it's

224

because I make a mistake and the autocorrect doesn't fix it

225

so now I have to go back manually and re-enter the text

226

but you know I mean this is just like standard machine learning stuff as this just gets better and better

227

and learns your patterns better it just it's going to be able to autocomplete more of what you're doing

228

so you won't even have to make all the motions so you've been calling glasses kind of the next platform for years.

229

Yeah.

230

But the original Ready by Metas really took off with just audio and AI.

231

What did you guys stumble on there?

232

So we have a few principles in building glasses.

233

And the three principles that we talk about the most are, they have to be great glasses first.

234

Right?

235

So you have to think about, okay, for any of the, before you even get to any of the technology, if you're going to wear these on your face all the time,

236

they need to be good looking, they need to be light, and these are.

237

I think a lot of people who are designing technology, they make something that's kind of bulky and they add a bunch of cool functionality to it and they're like, why don't people want to wear this?

238

It's like, well, most of the day, people aren't taking photos or something.

239

I mean, most of the day, it's just there.

240

And the magic of these is that they're good looking, comfortable glasses that you want to wear and then you have them on your face.

241

So when you do want to listen to music or call AI or whatever, it's available to do that. So good-looking glasses first.

242

The other big principle is that the technology needs to get out of the way, right?

243

So I think a lot of people, when they're designing technology, they have this impulse to kind of add a lot of flourishes

244

and make it so that the thing that they've built is kind of center stage.

245

And one of the things that I keep pushing back on with our team is, no, actually, look, especially if you're building glasses,

246

it is extremely important that whatever technology you have, it is there when you want to use it.

247

And then it gets out of the way.

248

You don't want random stuff in your vision.

249

It's like, I don't care how cool of an animation you made on the design team.

250

It's like people, we want this to be super minimal.

251

If a message comes in, you see the message, it goes away when you don't want it.

252

That, I think, the classic design just distilled those principles.

253

So, yeah, I think it was kind of a hit for those reasons.

254

The third principle, for what it's worth, is take superintelligence seriously.

255

So it's kind of a mantra internally that we have, which is this idea that AI is going to be the most important technology in our lives.

256

and we want to make sure

257

that ai is in service of people not just like something

258

that sits in data centers and like automates stuff and um

259

so we designed these even the early version the first version

260

of this um to be able to just have easy software updates to have improvements like the conversation focus thing

261

that we talked about so

262

that way you can get like okay now you have this

263

great accessibility feature you're in a loud place you can turn

264

up the volume on a person right it's like kind of crazy

265

but it's like okay so you build in the sensors first

266

that you think that the ai is going to need over the next couple of years

267

and then you make it

268

so as you get the software working you can just update it so

269

that we can bring the latest ai technology to two people

270

make them smarter with the software update well i'm a big

271

fan um personal question yeah how far away do you think we are from these glasses being

272

so good you just replace your phone So I think about it a little bit differently.

273

I think about what is your main computing device.

274

Right now, phones are your main computing device.

275

It's my main computing device for sure.

276

But I didn't get rid of my computer, right?

277

I just, I have my computer, but I just use it less.

278

Even a lot of the time, I'm even at my desk and my computer's right there.

279

But if I want to go do something, I just take out my phone because that's like my primary thing that I do.

280

So what I think is going to happen here is

281

that I don't think we're going to get rid of our phones anytime soon.

282

I just think our phone is going to stay in our pocket more or in our bag.

283

And even very subtle things, like I don't know how many times a day I used to look at my phone to see what time it was.

284

Now I don't do that when I have the glasses.

285

You just tap quickly, the clock comes up, done.

286

It's like, okay, I don't need to take out my phone to look at the time.

287

Okay, now I don't need to take out my phone to look at messages.

288

It's like, okay, now with these, the camera's great.

289

But now with the viewfinder, I know exactly what I'm capturing.

290

So I think you just go kind of use case by use case.

291

And all these things that I would have used to have taken out my phone for, I now don't.

292

I think that's sort of the way it's going to go.

293

But I would be surprised if we kind of get rid of our phones anytime in the next five years.

294

I just think it's going to be less and less use.

295

Slowly using your phone less.

296

Yeah, I mean, the GPS feature is another one, right?

297

Instead of having to look at my phone constantly, I'm going to use this like all the time.

298

Just have it in my glasses and navigating me where I'm wearing to go.

299

The lines translation even was useful.

300

There's a lot.

301

I'm curious to see how other people are going to be using it.

302

But I think another thing people are going to be really curious about is the metaverse.

303

how much of the vision is still distilled in these products?

304

Well, I think we're getting closer to it, right?

305

So the metaverse is obviously, it's a very visual experience of presence.

306

There's some parts of it like stereo audio, which I mean, you can 3D audio, which you can get, that is great.

307

But, you know, we're starting to get the display.

308

And then, you know, the Orion prototype that we showed last year, it's a wider field of view.

309

it'll be a more expensive product when it's when it's ready

310

but i think you're basically going to get to you know

311

this is one where you can you can have a hologram

312

you can you can kind of have context it's it's enough to kind of watch a video

313

or have a message thread but this product um

314

especially because it's monocular um it's not putting 3d objects in the world right

315

so there there still are things around delivering this sense of presence like you're there with another person

316

that I think are going to be really magical.

317

And when we get the consumer version of Orion, you'll really be able to start doing that.

318

So I think we're kind of inching our way there.

319

But this is a big breakthrough.

320

Having the first kind of high resolution display with the neural band.

321

And it's going to be a good step for learning how that goes.

322

But yeah, no, I think the vision is

323

that all of the kind of immersive software around presence that we have for VR, we would like a version of that to run on glasses too.

324

I want to switch gears to the vanguards.

325

Yeah, they're fun.

326

Really stylish.

327

I love the design.

328

I recreationally go on runs and cycle, so I'm using them as well.

329

How did you build these with athletes in mind?

330

Well, I think a lot of people on the team are pretty intense athletes.

331

And some of it is, you know, we just use the Ray-Bans.

332

and you know i can't tell you how many pairs of ray-bans i've fried taking them surfing um

333

but it was worth it but they're not water resistant

334

so i mean it's pretty obvious like you're like okay we

335

want something that's water resistant um the extra um sound the

336

the kind of power in the speakers these are six decibels louder is really valuable

337

and helpful when you're doing loud things So if you're cycling at 30 miles an hour and it's windy.

338

The other day, a few weeks back, I was taking a call on a jet ski

339

and I could hear the person completely fine over the engine of the jet ski.

340

And we have this advanced noise reduction in the background.

341

So we've designed these.

342

It's got the microphone in the nose pad, which basically means that you could be in a wind tunnel, and the other person on the other end of a phone

343

call would not be able to hear any of the background noise.

344

You just come in crystal clear.

345

So I think stuff like that is awesome.

346

Then pairing it with, and then there's All this other stuff that I think you want for sports.

347

The video stabilization is good.

348

The pairing with Garmin and Strava, I think is good.

349

We put an LED in the corner, so you can have it light up when you're,

350

to give you a reminder to make sure you stay on pace that you want or in your target heart rate zone.

351

So yeah, I mean, there's a lot of things like that that I think are gonna be pretty neat for athletes.

352

Yeah, it's fascinating.

353

Why do you think it was important to design so many different glasses for different lifestyles.

354

And how much broader do you really expect to go from here?

355

Well, I mean, I think people have different styles, right?

356

So this isn't like a phone where everyone's going to be okay having the same thing and just like, okay, maybe I have a slightly different color case

357

or I like put a different sticker on my case and then I'm good.

358

Like I think people have different face shapes, they have different aesthetics.

359

Like it's, I mean, people wear different kinds of clothes.

360

I think this is like an important part of our identity and sense of style.

361

And some of it is functional too.

362

I mean some of it is a little more lifestyle some of it is more active and

363

You know part of this is

364

Some people prefer thinner frames some people prefer thicker frames you

365

will always be able to fit more technology in thicker frames So as we get to more display

366

like more advanced holograms

367

it will always be a little bit thicker of a frame than what you can do for the thinner ones.

368

So I think that there's a choice that people will make around, okay, like, do I prefer the aesthetics and the kind of the simplicity of the thinner ones,

369

or do I like the aesthetics of the thick ones?

370

It's, I mean, fortunately, kind of big glasses are in style, so that kind of works.

371

So there's the whole spectrum.

372

And then there's also the price point and affordability.

373

And the less technology you're always going to be able to offer for a more affordable price.

374

And that's going to be great, too, because we want to make it so that there's more than a billion people.

375

It's somewhere between one and two billion people in the world who have glasses for vision correction today.

376

And I don't know.

377

I think within whatever it is, five to seven years, is there any world where those aren't all replaced with AI glasses?

378

And it feels to me like it's kind of like the iPhone came out

379

and everyone had flip phones and it's like just a matter of time before they're all smartphones, right?

380

So I think that that's going to be the case with glasses too, but people have different glasses and they want different glasses and they want different glasses at different price points and styles.

381

So our goal is to just work with a lot of iconic brands to do some great work.

382

So on the vanguards, you mentioned the garment integration.

383

I thought this was fascinating because this basically allows AI to kind of see your heart rate, pace, location, everything, kind of all hands-free while you're running

384

or cycling how long do you think it will be before

385

we have like a personalized coach that's there like proactively in your ear yeah i mean it can do

386

that a bit so i think um for running for example

387

if you want to ask it what your heart rate is

388

or what your pace is it can answer now i mean i don't know how hard you run

389

but when i'm running like for performance i don't really want to talk to something, right?

390

It's like there's, you're probably above that heart rate zone.

391

Um, but there are times when I'm like skiing really fast

392

and I want to know how fast I'm going because that's kind of fun.

393

And, um, you know, you're not necessarily out of breath doing that. So yeah, I mean, part of it is being able to talk to it and it can give you the stats in real time.

394

That's one of the reasons why we built the led in is, um, we want to make it so that, uh, you just have a very simple visual indicator of,

395

okay, am I on my, my kind of pace target?

396

Am I on my heart rate target?

397

Um, but I don't know, you can kind of get a sense of like over time, it's probably gonna be pretty good to have a display in those two.

398

Yeah.

399

I mean the real time coaching as well.

400

So like, I guess for an example, uh, if I'm running and I want to stay in zone two, uh, which is below a certain heart rate, but I don't want to go too, too hard because I want to stay in this zone.

401

Um, I guess my question is, will I be able to have a coach that's telling me, oh, hey, slow down a little bit.

402

You're a little too fast.

403

Yeah.

404

That's the goal of the LED to start.

405

And if you're trying to stay in zone two, you probably can talk to it.

406

Right?

407

Because I mean, that's kind of the definition of zone two.

408

Right?

409

It's like you should be able to.

410

Right?

411

So I don't know if it's the definition, but it's a property.

412

It's the sound two.

413

Yeah.

414

So you can.

415

You can basically ask it different stats, and it knows if you're connected to Garmin.

416

Yeah.

417

So I know you've been training a lot.

418

Have you found any good use of it when you're training as almost like a coach or?

419

Well, I take them surfing and they don't break, which is nice.

420

There we go.

421

Yeah, so, because the Ray-Bans are good, but like I said before, I've fried a few pairs of those.

422

Okay, and they capture moments in real time too, right?

423

Yeah, I mean, one of the nice things about this is because the camera is centered, it's really perfectly aligned.

424

I mean, the image really in the video feels like it's coming exactly from your perspective.

425

It's nice even when the camera's off to the side, but it's like these things about delivering a sense of presence,

426

they kind of have this feeling where it's like when you get it to be perfect, it really feels like it clicks.

427

And if you're like two degrees away, it's still pretty valuable, but it's not.

428

But there's something about having it in the middle that I think is really quite compelling.

429

Okay.

430

And wide field of view.

431

I'm excited to try them out.

432

Yeah.

433

So you're raising three kids while building superintelligence.

434

I'm curious what conversations you're having right now about what's about to grow up in.

435

Well, I mean, it comes up in a bunch of different things, but in our family, we build a lot of robots.

436

I'm really into 3D printing.

437

I think that's like a really fun DIY hobby.

438

My daughters are really into it.

439

They like making stuff.

440

Kind of got to the point where I'm like, all right, you need to design your own stuff.

441

I'm not just going to print stuff that we find on the Internet.

442

So now they're using all the different tools to do that.

443

And I work with them on that.

444

It's like a fun hobby.

445

But in a bunch of ways, I kind of think of robotics as like the thing after you get AI.

446

I mean, obviously, they'll kind of develop contemporaneously to some degree.

447

but um but i think we're gonna have really smart um kind of just kind of intellectual ai

448

and then you get the robotics

449

which is going to be the kind of fully embodied version of it

450

so um so yeah i mean there's there's not that much that you know the kids can do with

451

you know developing ai at their age but you can actually um they can

452

and they they just enjoy making stuff

453

so i mean it's like kids love 3d printing

454

and it's like not very complicated and it's just a fun hobby and

455

so yes we have this project going on right now first

456

i just ordered a bunch of robots from the internet

457

that you could assemble um that because there's like a bunch of starter packs

458

and things like that and i love that

459

and then um the next one is basically designing our own

460

and and basically just like yeah you 3d print the shell you have the the uh you know whatever the

461

you could probably do a version with the Raspberry Pi or the Jetson, you know, and just like get that to work pretty well for running language models and doing interesting stuff with it.

462

Yeah, that's a step up over Legos.

463

That's for sure.

464

Yeah, no, that's true.

465

It's nice because Legos are, you run out of blocks.

466

Yeah, that's true.

467

but with like the internet you don't you don't run out are there any like specific i guess traits

468

or skills you're trying to teach your kids so for example demis sabis he said

469

learning how to learn is a really important trait is there

470

anything you're thinking of teaching um i mean i mostly try

471

to teach our kids to be good people i mean i guess there's like the there's all intellectual stuff

472

that you're talking about it's like what's interesting robots whatever

473

but um i don't know i mean i think like being caring

474

and kind and things like that i think are really important um

475

so just a a bunch of values type things like

476

that um but yeah i mean i i agree agree with Demis' answer too.

477

It's less about like learning a specific thing and more about kind of learning how to go deep on a thing.

478

So I'm, I'm a big believer that, so a lot of people say some variant of like,

479

oh yeah, you want to learn how to learn, but okay, but how do you do that?

480

Right.

481

And I guess my theory on how you do that is basically a very depth first approach.

482

So rather than just like trying to have some conceptual framework, you complete a project, right?

483

So like, you will learn a lot by building a robot,

484

like whether you actually care about robotics after that is completely incidental to like,

485

you took a problem, we decomposed into all these things, we needed to figure out like the design for the 3D printing

486

and she needed to get the printers to work

487

and then like um and then we needed to like get the raspberry pi to work

488

and it's like okay we're like learning a lot about wheels

489

right now right it's like okay we're gonna do treads are

490

we gonna do wheels like how many wheels like how do you want you know

491

so um and

492

that stuff i think you basically it's like you're learning how

493

to decompose problems how to debug things um i think

494

that that's like you just learn by doing so that's That's kind of my philosophy on it.

495

Yeah, I like that.

496

And I think just doing stuff where you have common interests.

497

So, you know, it's like sometimes we build robots, sometimes we watch K-pop demon hunters.

498

You know, it's like, you gotta have fun.

499

Okay.

500

Okay, so when we do achieve super intelligence, these AI glasses are gonna see everything we see, hear everything we hear, and now even some of our muscle impulses.

501

What human traits should we fight to preserve, and what should we just let go?

502

Well, I'm not sure I agree that they're gonna see everything, or maybe there will be like some sensor that does,

503

but they're not necessarily gonna retain it.

504

I'm not sure that you want, I mean, that you want them to, but that's a whole separate direction that we could go into,

505

but I think just kind of like being able to pick out the salient bits are important

506

and giving people control over that is important.

507

But I don't know.

508

That's not really the core of what you're asking.

509

So the, I think like to me, creativity is very important.

510

And like having a sense of here's like what you want to make in the world.

511

and like having an idea of how you I mean it's it's some of the themes

512

that we just talked about like how do you decompose

513

that what are the steps to do that what are the tools

514

that are going to be most helpful in doing

515

that I think is part of the job of a creator

516

or a builder is to like be a kind of master of the tools

517

that are available to them right if you're not then you're not going to kind be at the edge of your work, you know, it's a competitive world.

518

So, yeah, I kind of think,

519

I don't know, I think all of the AI progress, I think, has been very fascinating because people have sort of conflated intelligence with,

520

like, an intent or, like, a desire to do something.

521

And I think part of what we've seen

522

so far from the AI systems is that you can actually separate those two things.

523

The AI system has no impulse or desire to create something.

524

It just sits there.

525

It waits for you to give it directions.

526

Then it can go off and do a bunch of work.

527

And so I think the human piece at the end is going to be, well, what do we want to do to each make the world better?

528

and I think some of

529

that will be around like a personal kind of creative manifestation of what you want to build

530

but I do think some of it

531

and I think in our industry would probably underplay this a bit is just like caring about other people

532

and you know taking care of people and spreading kindness

533

and like so I think that that stuff is really important too and I think AI will help with that as well.

534

So this conversation wouldn't be without obviously the super intelligence labs

535

that you've been on a spree over the last couple past months and all over the news.

536

Yeah, it's been busy.

537

Yeah.

538

When did you decide to kind of make that change?

539

From the outside, there were rumors that you were like personally going founder mode, calling researchers, emailing, all that stuff.

540

What drove the decision for this?

541

Yeah, I think, I mean, it was largely LLAMA 4 was not on the trajectory

542

that I thought it needed to be on right i mean

543

llama four was in many ways a big improvement over llama three

544

but like we weren't trying to be a bit better than

545

llama three right we were basically trying we want to build like we're a frontier lab right

546

that wants to be doing leading work right that um so i kind of felt like

547

I'd learned enough about how to set up a lab

548

that I wanted to kind of reformulate the work that we're doing with a few basic principles.

549

So one is huge focus on talent density, right?

550

So you don't need many, many hundreds of people.

551

You need, you know, 50 to 100 people.

552

It's a group science project who can basically keep the whole

553

thing in their head at once um like seats on the boat

554

or precious right it's like

555

if someone is not is not um isn't pulling their weight on

556

that then that's like it has this huge um negative effect in the way

557

that it doesn't for a lot of other parts of the company um

558

so i i think like just this absolute focus on like having the highest talent density in the industry,

559

drove a lot of what you heard about, okay, well, like, why am I going out and meeting all of the top AI researchers?

560

It's like, well, like, I want to know who the top AI researchers are, and I want to, like, personally have relationships with them,

561

and I want to build the strongest team that we can.

562

So that was a big part of the focus.

563

There are other parts, too.

564

Like, I think when you're doing long-term research, one of the principles that we have for the lab is we actually don't have deadlines that are top down, right?

565

It's like, it's research, right?

566

You don't know how long the thing is going to take.

567

All the researchers, you know, everyone's competitive.

568

They all want to kind of be at the frontier and doing leading work.

569

So, you know, me setting a deadline for them isn't going to, you know, help.

570

It doesn't help them at all, right?

571

It's like they're moving as fast as they can to try to understand the kind of, what we need to solve the problem.

572

So, but that's kind of, yeah.

573

So we organized the lab to be very flat.

574

We don't want layers of management that are not technical.

575

The issue is people start off technical, and then they go into management, and then six months or a year later, you still think you're technical, but you actually haven't been doing the stuff for a while,

576

so the knowledge just kind of slowly decays or quickly decays in an environment that's moving as quickly as this.

577

So anyway, those are the basic principles.

578

but I think that overall taking a step back I think

579

that AI is going to be the most important technology in

580

our lives in our lifetimes building the capacity to build absolutely

581

leading models I think is going to be a really critical

582

and very valuable thing for for unlocking creativity building a lot of awesome things

583

so yeah I'm absolutely just going to focus and do the things

584

that we need to do in order to make sure that we can continue to do that.

585

How hands-on are you still with the lab?

586

Well, I mean, the lab, I moved everyone around me who sat with me, and now the lab is there.

587

So, I mean, I'm pretty hands-on.

588

It's like, I mean, Shengji, the chief scientist, you know, sits right next to me.

589

So, and, you know, a lot of the other teams sit within 15 feet of me.

590

So, I mean, what would I say?

591

I mean, look, I'm not an AI researcher.

592

All right.

593

So it's not like I'm like in there, like telling them what research ideas to do.

594

The things that I can do is as CEO of the company are I can make sure

595

that we have the very best people and talent density.

596

I can make sure that we have by far the highest compute per researcher

597

and that we do whatever it takes to go build out all that capacity.

598

I mean, you probably saw the announcement that we had.

599

We're building this Prometheus cluster, which is, I think it's going to be the first kind of gigawatt plus single cluster.

600

That's like a contiguous cluster for training coming online next year.

601

And we're building a cluster in Louisiana that is going to be five gigawatts

602

and we're building several more clusters like that that are that are just going to be

603

multiple gigawatts so yeah so i mean and that's

604

that obviously takes some conviction i mean we're talking about many hundreds of billions of dollars of capital

605

so you both need to have like a good business model

606

that can support it and you need to believe in it

607

so i i um that's that's one thing

608

and then like there's all these other parts of running a

609

company where i don't know i mean like big companies have positives

610

and negatives and it's sort of my job to make sure

611

that we channel all the positives that um

612

that we can towards this effort and

613

that i can clear as many of the kind of things

614

that would otherwise be annoying about running a company out of the way

615

so you know when we build something awesome we'll get in our apps

616

and get it to billions of people quickly that's a good thing

617

that you can do inside a big company

618

and then like whatever the other kind of things

619

that would be more annoying it's sort of my job to clear them out of the way

620

so how do i do that well i want to make sure

621

that i know the researchers well right and that they you know

622

that like when they have an issue that they feel comfortable texting me

623

or just like walking over to my desk and um

624

so i think that that's actually part of running the whole thing well is like is

625

that they kind of like i have that rapport with them

626

and um in that way i can be in the information flow to understand what the issues are

627

that they're having so i can go solve them for them

628

so it's like building the culture of everything goes open everything moves fast yeah

629

and you're personally talking with all these researchers yeah i mean i'm doing other stuff for the company too obviously

630

but yeah and but yes i'm just trying to make sure that we move as quickly as as as um

631

as we can to get the researchers what they need to do the best research in the industry

632

last question if we do achieve super intelligence and whenever

633

that is what are the plans to integrate into meta's existing products yeah i mean i i think

634

when you have super intelligence the nature of what a product is will change pretty fundamentally.

635

If you think about our products today, a lot of the big ones, right?

636

So if you look at Instagram or Facebook or our business model around ads,

637

they're basically already these very large transformer systems, right?

638

They're recommendation systems, not language models today.

639

But they're basically already these big, massive scale machine learning systems.

640

So, I mean, in the future, do you think we'll get to a point where, you know, instead of having it be a recommendation system that,

641

you know, has some kind of basic understanding of what you might care about

642

and therefore can kind of spit out an ID for a story and then you have the app that renders it?

643

I just think you'll have models that you'll interact with directly that it'll recommend content,

644

it'll generate content, you'll be able to talk to it, right?

645

It's like just, so I think this stuff will all be much more integrated.

646

And then I think

647

that sort of culminates in the Glass's vision where I think

648

what you're going to get is eventually this always on experience

649

that I mean you can control when it's on and off

650

but it can be always on

651

if you want where it can you can just let it

652

see what you see hear what you hear it can go off

653

and think about the context of your conversations

654

and come back with with more context or knowledge

655

that it thinks you should have when you need an app it can just generate the ui from scratch for you

656

in your vision I I'm not sure how long it's going to take to get to

657

that I don't think this is like

658

five years I think it's gonna be quicker

659

so two three it's hard to exactly know but

660

but I don't know I would guess it's every time I

661

think of what a milestone would be an AI they all seem to get achieved sooner than we think

662

so I think like my optimism about AI has generally only

663

increased as time has gone on in terms of both the timeline for achieving it and how awesome it's going to be.

Tải Ứng Dụng

Mọi thứ bạn cần để nói trôi chảy

Chấm điểm AIChấm từng câu nói

Luyện IPAThành thạo mọi âm

Từ vựngMở rộng vốn từ

Game Từ vựngHọc qua chơi

Quét QR

Download on theApp Store Get it onGoogle Play

FOR YOU

Video gợi ý

54 câu

Từ vựng và ghi chú luyện nói cho bài học này

Bài luyện nói trình độ B2 này dựa trên video “Mark Zuckerberg on AI Glasses, Superintelligence, Neural Control, and More”. Video này có 663 câu và 8580 từ để luyện shadowing. Phần lời nói dài 45:23. Người nói nói nhanh, khoảng 189 từ mỗi phút, nên sẽ có nhiều chỗ nối âm và nuốt âm. 85% số từ nằm trong 3.000 từ tiếng Anh thông dụng nhất; phần còn lại nên xem trước khi luyện.

Từ vựng quan trọng trong video

15 từ khó nhất trong video, kèm phiên âm và nghĩa:

Từ	Phiên âm	Nghĩa
researcher danh từ	/ˈɹiˌsɝ.t͡ʃɚ/	nhà nghiên cứu
llama danh từ	/ˈlɑː.mə/	lạc đà không bướu
athlete danh từ	/ˈæθ.liːt/	vận động viên
vanguard danh từ	/ˈvanˌɡɑːd/	tiền phong, tiên phong
aesthetics danh từ	/ɛsˈθɛt.ɪks/	mỹ học
hobby danh từ	/ˈhɑ.bi/	sở thích riêng, sở thích
annoy động từ	/əˈnoɪ̯/	làm phiền, quấy rầy
distil động từ	/dəˈstɪl/	chưng cất
metaverse danh từ	/ˈmɛt.ə.vɝs/	siêu vũ trụ ảo, siêu vũ trụ số
mantra danh từ	/ˈmɑn.tɹə/	mạn-đát-la, man-tra
trajectory danh từ	/tɹəˈd͡ʒɛktəɹi/	quỹ đạo
trait danh từ	/tɹeɪt/	đặc điểm, đặc tính
milestone danh từ	/ˈmaɪlˌstoʊn/	cột mốc, cây số
rumor danh từ	/ˈɹumɚ/	tin đồn, lời đồn
microwave danh từ	/ˈmaɪkɹoʊˌweɪv/	vi ba, vi sóng

Cụm động từ bạn sẽ nghe

Từ	Nghĩa
pick up động từ	nhặt, lượm
bring up động từ	nuôi nấng, nuôi
go out động từ	ra, chết
go back động từ	trở lại
run out động từ	hết

Những câu đáng luyện lại

Các câu ngắn, trọn ý trong video mà bạn có thể dùng lại khi giao tiếp hằng ngày:

It's picking up your muscles firing.
What did you guys stumble on there?
How did you build these with athletes in mind?
Am I on my heart rate target?
How hands-on are you still with the lab?

Ngữ pháp trong video

Những cấu trúc người nói dùng nhiều nhất, kèm đúng cụm từ trong video:

Cấu trúc	Trong video
Thì hiện tại hoàn thành tiếp diễn have/has been + -ing — việc bắt đầu từ trước và vẫn đang tiếp tục	you've been using · you've been calling · you've been training
Thì hiện tại hoàn thành have/has + quá khứ phân từ — việc đã xảy ra nhưng còn liên quan đến hiện tại	we've shipped · have shipped · i've sent
Mệnh đề quan hệ who / which + mệnh đề — thêm thông tin về người hoặc vật	people who are · have, which is · friends who you
Câu bị động be + quá khứ phân từ — nhấn vào việc xảy ra, không phải người làm	be personalized · is still distilled · is centered

Phát âm cần chú ý

Người nói dùng 229 dạng rút gọn, ví dụ you're, don't, I'm. Hãy nói theo dạng ngắn đúng như bạn nghe.

Âm “th”: athlete /ˈæθ.liːt/, aesthetics /ɛsˈθɛt.ɪks/, breakthrough /ˈbɹeɪk.θɹuː/, thumb /ˈθʌm/, marathon /ˈmæɹəθən/
Âm “sh” và “zh”: correction /kəˈɹɛkʃən/, recommendation /ˌɹɛkəmɛnˈdeɪʃən/, animation /ˌæn.ɪˈmeɪ.ʃən/
Từ dài — đặt trọng âm cho đúng: personalize /ˈpɜː.sə.nə.laɪz/, intuitive /ɪnˈtjuːɪtɪv/, trajectory /tɹəˈd͡ʒɛktəɹi/, internally /ɪnˈtɜːnəli/, creativity /kri.eɪˈtɪv.ᵻ.ti/

Những âm người Việt thường gặp khó:

Cụm phụ âm cuối — đọc đủ từng âm, đừng nuốt âm cuối: robotics /ɹoʊˈbɑ.tɪks/, aesthetics /ɛsˈθɛt.ɪks/, resistant /ɹəˈzɪstənt/, impulse /ˈɪmpʌls/, wrist /ɹɪst/
Âm /tʃ/, /dʒ/ và /ʒ/ — tiếng Việt không có, đừng đọc thành “ch” hay “gi”: researcher /ˈɹiˌsɝ.t͡ʃɚ/, trajectory /tɹəˈd͡ʒɛktəɹi/, gesture /ˈd͡ʒɛs.t͡ʃɚ/, scratch /skɹæt͡ʃ/
Âm /æ/ — mở miệng rộng hơn âm “e”: fascinate /ˈfæsɪneɪt/, athlete /ˈæθ.liːt/, navigate /ˈnæv.ɪ.ɡeɪt/, interact /ɪn.təˈɹækt/, scratch /skɹæt͡ʃ/

Cách luyện với video này

Nghe hết video một lần, chưa cần nói, và ghi lại những từ bạn chưa biết.
Bắt đầu ở tốc độ 0,75×, nói đuổi từng câu, rồi quay lại tốc độ bình thường khi đã quen.
Ghi âm giọng mình rồi so với bản gốc, chú ý các từ như researcher, llama, athlete.

Phương Pháp Shadowing Là Gì?

Shadowing là kỹ thuật học ngôn ngữ có cơ sở khoa học, ban đầu được phát triển cho chương trình đào tạo phiên dịch viên chuyên nghiệp và được phổ biến rộng rãi bởi nhà đa ngôn ngữ học Dr. Alexander Arguelles. Nguyên lý cốt lõi đơn giản nhưng cực kỳ hiệu quả: bạn nghe tiếng Anh của người bản xứ và lặp lại to ngay lập tức — như một "cái bóng" (shadow) đuổi theo người nói với độ trễ chỉ 1–2 giây. Khác với luyện ngữ pháp hay học từ vựng bị động, Shadowing buộc não bộ và cơ miệng phải đồng thời xử lý và tái tạo ngôn ngữ thực tế. Các nghiên cứu khoa học xác nhận phương pháp này cải thiện đáng kể phát âm, ngữ điệu, nhịp điệu, nối âm, kỹ năng nghe và độ lưu loát khi nói — đặc biệt hiệu quả cho người luyện IELTS Speaking và muốn giao tiếp tiếng Anh tự nhiên như người bản ngữ.

Phương pháp shadowing: đọc hướng dẫn từng bước đầy đủ →

Luyện nói tiếng Anh bằng Shadowing qua video: Mark Zuckerberg on AI Glasses, Superintelligence, Neural Control, and More

Chọn chế độ luyện tập

Tải Ứng Dụng

Meta is pivoting again... everything you missed from Connect 2026

Top five moments from CNBC's interview with OpenAI CEO Sam Altman

The perfect technology for creeps?

Did someone say... SecondBrain?

The Future Mark Zuckerberg Is Trying To Build

半导体女王 AMD CEO苏姿丰Lisa Su:AI未来10年新发现将超过去30年总和！如何创造属于自己的运气？

think like a business analyst in tech (3 core skills + solving case study)

Jensen Huang & Satya Nadella on unmetered intelligence | Microsoft Build 2026

OpenAI’s new browser feels familiar…

Communicating the Future: Defining Where We Want AI To Take Us

WWDC26: Run local agentic AI on the Mac using MLX | Apple

What Happens If China Invades Taiwan? Elon Musk's AI Warning

Từ vựng và ghi chú luyện nói cho bài học này

Từ vựng quan trọng trong video

Cụm động từ bạn sẽ nghe

Những câu đáng luyện lại

Ngữ pháp trong video

Phát âm cần chú ý

Cách luyện với video này

Phương Pháp Shadowing Là Gì?