HomeVideos

What is happening at Meta?

Now Playing

What is happening at Meta?

Transcript

402 segments

0:00

The meta situation is actually somehow

0:02

worse. For those that don't know

0:03

anything about meta, Facebook, or the

0:05

Instagram situation, let me just give

0:07

you a quick uh speed run of its history.

0:10

Facebook went super hard in VR, then

0:12

found out, "Yo, VR is not actually where

0:14

it's at. It's all about AI." And then

0:17

decided that it's going to fire a

0:18

roughly 10% of its company or 8,000

0:21

employees and focus super hard on AI.

0:23

And a part of that focusing super hard

0:25

on AI, it's going to record everybody's

0:27

desktop and everything you do and then

0:29

feed that to the AI. Not only that, it's

0:31

going to create a new organization,

0:33

Applied AI, where it's going to put

0:35

approximately 10% of the company into

0:37

it, where they're going to help train

0:39

and make the next generation of AI. Of

0:42

course, during all this, is Zuckerberg

0:43

on a phone call called everybody super

0:45

highly intelligent, the most high-IQ

0:48

people in the valley. And then after

0:50

saying all of that, said that the reason

0:52

why they're recording your desktop and

0:53

not telling you why is because it's for

0:55

your own best interest.

0:58

>> Not strategically in your interest for

1:00

us to communicate everything like in all

1:02

the detail that we normally would on

1:04

this.

1:04

>> But now, the internals are coming out.

1:07

Employees are coming forward and

1:08

speaking about what's actually going on

1:11

inside of Facebook. They They does not

1:13

look good. It does not look good at all.

1:15

In fact, this is somehow the least bad

1:17

thing coming out of Facebook, which is

1:20

they just shipped JSON that's not even

1:23

possible to production. The golden age

1:25

of slop has arrived and it's happening

1:28

right now live at Facebook. But the

1:30

internal employee morale is even worse

1:33

than this. Like imagine working at a

1:35

place where they're like, "You ain't got

1:37

time to set up a CI/CD pipeline to make

1:39

sure JSON parses." That's where we're

1:42

at. But first, I'd like to say a quick

1:43

thank you to the sponsors.

1:45

>> Hey, is that HTTP? Get that out of here.

1:48

That's not how we order coffee. We order

1:50

coffee via SSH, terminal.shop. Yeah, you

1:53

want a real experience? You want real

1:55

coffee? You want awesome subscription so

1:57

you never have to remember again? Oh,

1:59

you want exclusive blends with exclusive

2:01

coffee and exclusive content? Then check

2:05

out Crown. You don't know what SSH is?

2:08

Well, maybe the coffee is not for you.

2:12

>> [music]

2:19

>> All right, thank you. The best way to

2:20

support the channel, check out the

2:21

sponsors. Links down in the description.

2:22

Okay, so welcome back. Let's talk about

2:25

this. We're going to look at this by

2:26

looking at what the employees are

2:28

saying. I am going to both laugh at the

2:30

situation and make fun of an employee or

2:32

two for this because they describe

2:33

things rather ridiculously. But then

2:35

we're going to look at how the company

2:38

is talking about it. How the management

2:40

is talking about it, which is also

2:41

somehow super ridiculous and is even

2:44

more nuts than how the employees look at

2:47

it. Now, a lot of this information comes

2:48

from the Wired article. You got to go

2:50

check it out. It's linked in the

2:51

description. But it starts off with the

2:52

title of tell him he's a piece of

2:55

Meta's new AI unit is a total mess. Yes,

2:58

it is absolutely a total mess. So now

3:00

that that little quote, that little line

3:02

comes from actually a live town hall in

3:05

which thousands of employees were all in

3:08

a single meeting video call with some

3:10

present presenters going over some sort

3:12

of information. Someone just joins,

3:15

flips out a mic, goes hot mic and is

3:17

just lets out a four-letter rant for the

3:20

ages finishing up with hey, you know the

3:22

meta AI executives? Yo, go tell Go tell

3:25

him he's a piece of Imagine being

3:27

in that call because the presenters, it

3:29

even says one of the presenters covered

3:31

their faces with their hands. A

3:33

according to a witness. Of course, you

3:35

know why they were covering their faces.

3:37

They were laughing, okay? You can't You

3:39

can't be showing all the people that

3:40

you're laughing at some profanity-laced

3:42

AI rant, especially when you're the one

3:44

presenting. You just got to You just got

3:45

to cover your face and pretend it

3:46

doesn't happen. So what's happening is

3:48

since April a new org has has formed in

3:50

which about 6,500 employees have joined.

3:53

Now, they have joined in waves, and it's

3:54

described as this. The organization has

3:56

grown in batches since early April. It's

3:59

crazy to watch people experience the

4:00

shock of it as each wave comes in. So,

4:04

this organization is not necessarily

4:06

known for being fantastic. So, what's

4:08

actually happening is that they're

4:10

taking all of their talented engineers,

4:12

right? These really top talented the

4:14

best, the most intelligent engineers

4:17

that are possible at Facebook, and

4:19

they're assembling them into this team,

4:21

and they're given about two tasks per

4:23

week to finish. And these tasks, they

4:25

involve generating complex software

4:27

coding problems to help AI scientists

4:29

better train and evaluate the

4:31

performance of the latest frontier

4:33

models. Meaning, they're not actually

4:35

solving or building anything real.

4:37

They're coming up with toy examples, toy

4:40

coding problems, and actually

4:43

like using their collective power of

4:45

thousands of engineers to create the

4:48

ideal coding examples to train the AI.

4:51

So, that means that employees, not only

4:54

are they getting their screens recorded,

4:56

which by the way, just in case you're

4:57

wondering, the screen recording, the

4:59

company has scaled the program back

5:01

slightly, allowing employees to pause

5:03

the data collection for up to 30 minutes

5:05

and request specific exemptions. So,

5:07

just in case they want to do a little

5:08

bit of Twitter following or whatever or

5:10

whatever goes on for 30 minutes on an

5:11

employee laptop that I do not care to

5:14

want to know, they're just asking for

5:16

some exemptions, and Facebook has

5:18

granted it in their infinite wisdom.

5:19

Facebook is not only recording the

5:21

desktops, they have thousands of people

5:24

handcrafting curated beautiful coding

5:27

challenges to make the greatest coding

5:29

robot of all time. Like they are

5:31

literally hiring and making it explicit.

5:34

They're trying to figure out a way to

5:36

replace employees, and all these

5:38

employees work on it. They're They're

5:39

They're actually trying their darndest.

5:42

But, how do the engineers feel about

5:44

this? Well, the engineers selected for

5:46

the unit have no choice to join or leave

5:48

the company. That's right. So, if you

5:50

were an engineer and you were working on

5:52

say some social media feature for the

5:54

billions of users of WhatsApp,

5:56

Instagram, or Facebook, they say, "Hey,

5:58

by the way,

5:59

you're in on the new program." And if

6:01

you say, "I don't want to be in on the

6:02

new program." They're like, "All right,

6:03

then leave the company, idiot." An

6:05

unusual requirement for a highly valued

6:07

technical employees in the Silicon

6:08

Valley. That's led some members of the

6:10

applied AI to describe themselves as

6:13

draftees. Yes, the employees are

6:16

actually comparing themselves like this

6:18

is their Vietnam. They're like, "Bro,

6:21

it's Vie-

6:22

>> [laughter]

6:23

>> I actually can't even say that with a

6:24

straight face. It's so ridiculous. Like

6:26

they're actually being forced into this

6:28

thing or being fired and they're calling

6:30

themselves draftees. It is

6:33

It is absolute cinema over there. This,

6:36

of course, led to one employee saying,

6:38

"It's literally the gulag.

6:40

You have zero purpose in life all of a

6:42

sudden. You barely interact with anyone.

6:44

You just have these tasks every week."

6:47

Now, I I do admit that does sound awful.

6:49

That does not sound like some kind of

6:50

fulfilling job where you're just simply

6:52

trying to develop whatever that coding

6:54

challenge is that you have to develop

6:56

and you're just kind of answering to

6:58

this amorphous blob of AI scientists in

7:01

which you have no light into. You have

7:02

no real meaning or impact at your job.

7:04

You have no idea if you're affecting

7:05

anybody or nobody. And you actually

7:07

don't even know if you're training your

7:09

replacement or not. Like I get that.

7:10

That would not be that great. I would

7:13

probably find myself pretty unhappy and

7:15

probably applying around at some

7:16

different jobs. But to describe it as a

7:19

gulag, the gulag, just in case anybody's

7:22

wondering, is best known for random

7:24

beatings, brutalized torture, high

7:27

amounts of starvation, absolute exposure

7:30

to the elements, and lots of diseases

7:32

and public defecation. Like

7:33

>> [laughter]

7:34

>> it is I hate to break this to you, but

7:36

your sushi lunches and like not

7:40

meaningful work is just not the same as

7:43

a gulag.

7:45

>> [laughter]

7:46

>> I can't I can't I can't even believe it.

7:48

I can't believe it. I can't believe I'm

7:49

defending

7:51

Facebook [laughter] in this situation.

7:53

Another employees describe some of these

7:54

tasks generating puzzles to test how

7:56

reliable AI models from Meta and other

7:58

companies can solve them. As easy as

8:00

compared to software development work

8:01

they had been doing previously, but the

8:03

new projects feel menial and almost all

8:06

employees seem unhappy, they say. Most

8:08

people find their work soul-crushing, a

8:10

third employee says. All right, I can't

8:12

actually make fun of the employees

8:13

anymore. Like I do understand that

8:15

boring work can be really, really

8:17

frustrating. I do remember sitting in my

8:18

car and actually being like, "God, help

8:20

me have the patience to do the work at

8:22

my job." Cuz I too worked at a

8:23

soul-crushing company for quite some

8:25

time and

8:26

like I In some sense, I do get that

8:29

feeling, but it does seem

8:31

it does seem a little overblown there.

8:32

But how is management responding to all

8:35

this? During a meeting this week to all

8:37

employees at Instagram, and Meta chief

8:39

product officer Chris Cox addressed the

8:41

difficult and brutal environment created

8:43

by the insanity of this company in the

8:45

past few months, according to the

8:46

recording heard by Wired. Cox applauded

8:49

the Instagram employees for launching

8:50

features and serving around 2 billion

8:52

users amid what he compared to running a

8:54

marathon in the middle of a hailstorm,

8:56

and then like your teammates get

8:58

replaced, and then we're recording you.

9:01

I mean, hey, at least at least he's

9:03

somewhat honest about the situation. But

9:05

then, you know what he says right

9:06

afterwards? He gives a fantastic summary

9:09

of AI. Cox said that he needed to reckon

9:11

with how he and other leaders could get

9:13

in touch with the company again and not

9:15

be over-earnest about the power of AI.

9:17

It is neither God nor it's the devil, he

9:19

said. It's nowhere near as good as you

9:22

think it is, and it's nowhere near as

9:23

bad as you think it is, and it changes

9:25

every day, and it doesn't know what day

9:27

of the week it is. Hey, not bad, Chris

9:29

Cox, good save. But what is the company

9:31

doing? Now, this is, I think, the

9:33

largest failure of this entire

9:34

situation. Okay, listen to this.

9:36

Zuckerberg reiterated a vow not to carry

9:39

out additional mass layoffs this year.

9:41

Okay,

9:42

so he's letting you know, you got 6

9:44

months of peace. By the way, just saying

9:45

it that way is so nuts. Like he's even

9:48

he's saying, "Hey, we're going to

9:49

probably lay off more people, just not

9:51

right now." He introduced a plan to

9:53

limit the number of employees per

9:55

manager, which some teams, such as

9:57

Applied AI, had deliberately ballooned

9:59

to a ratio of 50 to 1. That's That

10:02

number makes no sense. Like Jensen Huang

10:04

doesn't even have that many, and he's

10:06

considered the most unusual CEO of a

10:09

company of all time. Like how does this

10:11

even happen on a team? Like how do you

10:14

even organize or do anything with 50

10:17

direct reports? So, the big plan, of

10:19

course, is more money for team events,

10:22

which, of course, everybody knows team

10:24

events is really what makes you happy.

10:26

You know, like nothing's better than

10:28

spending a little extra time at the

10:30

office and getting something bougie out

10:32

of it, right? Like that's that's what

10:34

makes you a good employee. And the other

10:36

one, of course, is a hackathon designed

10:38

to bring the company back together

10:40

again. Now, let's not We I don't think

10:42

we have to think too hard about this

10:43

one, okay? I think we could both see the

10:46

problem with this.

10:47

Now, imagine for a second, you're at a

10:49

company in which you've been drafted

10:52

into a new organization, you're doing

10:55

menial work you're not happy about,

10:58

you're getting a lot of tasks you feel

10:59

like you have no insight into it, a

11:01

bunch of your, you know, co-employees

11:03

have all been fired. So, what does

11:05

management do? Hey, everybody, let's

11:07

recreate Navy SEALs Hell Week, except

11:10

instead of trouncing through the ocean

11:13

sand, we are instead going to be hacking

11:16

on the products you already work on

11:18

nonstop for a week. We're going to call

11:20

it a hackathon, and then everybody's

11:22

going to get pizza and an award at the

11:25

end of it. How

11:27

do you not see that that's not going

11:30

going a good idea? Okay, like I am I am

11:33

no Nostradamus,

11:34

but hackathons aren't great when people

11:38

feel burnt out, okay? This is not a way

11:41

to win the employees. Anyways, I just

11:43

thought this was so funny because we're

11:45

watching in real time one of the largest

11:47

companies on Earth slowly implode in on

11:50

itself. Now, will it actually implode in

11:52

on itself? I don't know. Right now, it's

11:55

set on pace to exceed uh Google for ad

11:58

revenue. It is by all accounts

12:00

absolutely just making hand over fist

12:03

money. So, will it actually implode? I

12:05

don't know. But, will it lose a lot of

12:07

the top-tier talent? Likely. And now,

12:10

will that lead it into such a technical

12:12

disaster that it has no other choice but

12:15

people leaving the platform for

12:16

somewhere else? Now, that is actually a

12:18

real situation. Again, hundreds of

12:20

thousands of Down Detector reports,

12:22

Facebook absolutely crashing, and JSON

12:25

lexical errors being brought to the

12:28

front end, letting you know that a JSON

12:31

cannot be parsed. This cannot be a

12:33

coincidence, okay? This cannot be a

12:35

coincidence. Deep employee morale hits,

12:40

plus increasing instability, plus

12:43

increasingly weird management decisions,

12:46

all are going to come together just to

12:48

create some of the worst products on

12:50

Earth. I just wanted to talk about this

12:51

because

12:52

it's just shocking to see. You know, as

12:54

somebody who worked in the Valley,

12:55

Facebook at one time was considered a

12:57

very prestigious place to work, right?

12:59

The best engineers got to go work there,

13:00

okay? You were going to get treated the

13:02

best, you were going to get paid the

13:03

most, and you're going to work on

13:04

problems that were likely some of the

13:05

hardest problems you've worked on,

13:06

right? It was supposed to be a great

13:08

place to go and work super hard if

13:11

that's what you were into. But now, look

13:13

at it. It's like

13:15

a disaster. It's like every single month

13:17

the reporting gets worse and worse

13:19

coming out of it. I just

13:23

It's It's crazy to see these things

13:25

happen to companies in which used to be

13:28

the tippity top. All right, anyhow, hey,

13:30

the name is

13:33

you know, the prime gen, okay, hey, hi.

13:36

It's nice to see you. I hope I hope

13:38

you're doing well. Are you doing well?

13:39

Like what's going on with you, you know?

13:41

Do you come here often?

13:42

I don't know, you know,

13:44

you guys

13:45

What are you guys doing? I'm going to

13:46

get a haircut. Did you know that?

13:48

Yeah, I'm going to get a haircut. It's

13:49

honestly, it's just too big like look at

13:51

this. Like

13:52

like 50% of my head size is devoted to

13:55

the hair. I think I just got to do

13:56

something about it, you know?

14:00

Oh, Jen.

Interactive Summary

The video analyzes the internal turmoil at Meta (formerly Facebook), focusing on the company's aggressive pivot toward AI. Employees are being forced into a new 'Applied AI' unit, where they perform repetitive, menial coding tasks. This environment has caused morale to plummet, leading to internal rants, poor product stability, and significant management blunders in an attempt to address the dissatisfaction.

Suggested questions

2 ready-made prompts