Latest / Elon Musk Thinking / Elon Musk Thinking About AI And The Tesla Campany

Transcript

Elon Musk Thinking About AI And The Tesla Campany · Elon Musk Thinking

0:00
M hm hm hmm. I think is that just taking me a little biting?
0:09
Let's see? This is this is cruzy? Yeah, okay, well
0:16
a lot of people don't, okay more than merry are. Oh yeah,
0:21
that'll be crazy. So it's just on speakers. That basically what we do
0:34
there, okay, and we'll just plug it in that your mics will beat
0:38
it once you plug it into that adapter. Oh great, okay, so
0:42
we'll just be a little bliger than, you know, than as more m
0:59
Let's see, can people hear me? I think they can hopefully, so
1:10
let's hopes yeah, m hm hm. Do you want to test the audio
1:33
with it bucked into the would you prefer to keep its Yeah? It should
1:38
work? Um, this is I think this is working? Um great,
1:49
yes, it's working. Okay, okay, all right, well we can
1:57
go. All right, you're getting ready. I want to have a frank
2:01
of water, take a deep breath a right, let's go. You're good
2:07
yeah? Sure, oh yeah, thanks, okay, all right, we're
2:12
ready to go here yep, okay, production lines moving. It's just such
2:23
a great thing to watch. Yeah, all right, we're gonna get started
2:29
in just a second, just a second. Is this is this live live?
2:32
Great? Yeah, okay, yeah, I wanted to do it live.
2:44
Yep. I'm ready. When they are, I'll tell them we're live
2:52
on Twitter spaces too, should absolutely okay, M yeah, Carl, thank
3:00
you. I am here with the Elon Musk. He's he's getting everything ready
3:04
on Twitter spaces, so join us there as well. But of course you
3:07
can just watch us live here at the gig of factory in Austin. Thank
3:10
you for allowing us to be here. You're welcome. H It's quite something
3:15
to just sit and watch for quite some time. UM. Yeah, this
3:17
this is actually where Um, I don't actually have an office here. I
3:22
either walk the floor or I hold meetings in this room, okay, And
3:25
this room allows me to see what's going on with production while I'm holding meetings.
3:30
Is there one particular production that you focus on when you're here or that
3:34
you sort of your eyes always go to, or you you actually walk over
3:38
to. Um, well, this is the end of line that we're seeing
3:40
here. So if if cars aren't moving, um, that means this there's
3:46
some blockage upstream, and then I can go and see what that blockage is.
3:50
UM, So I think we might be on a break right now.
3:53
But uh, this, if this, this is um, if this,
3:58
if vehicle end of line is not moving, that means there's something upstream that's
4:00
wrong. Got it. Well, it's been amazing just to walk around the
4:03
facility a bit and get a sense for the size. All Right, you
4:08
just finished your annual meeting. You took a lot of questions from a very
4:11
eager audience. I feel like most of it you've shared before, but there
4:15
was some news and I want to start there advertising. You got a big
4:19
applause. You didn't sound like you're going to advertise that much, but you
4:24
did at least seem to open the way to saying, okay, we might
4:27
consider doing some advertising. Why Well, I mean, I believe in listening
4:31
to listening to the shareholders, and I actually was surprised by the level of
4:35
enthusiasm for advertising, since we are not historically done that. But there perhaps
4:43
is some good logic to it in that if we're simply saying out information by
4:47
I say that the Tesla Twitter account of my Twitter account, we're somewhat preaching
4:50
to the converted and not reaching people that are not already convinced. Essentially,
4:58
so I think the I have a good point. Well, I mean I
5:00
think it's worth a try and we'll see, we'll see. How do you
5:04
have any idea how you'd like to try it or what that would involve?
5:09
Um? Well, I mean I have some general thoughts about advertising. Um
5:13
that that you know, if advertising is informative um and and entertaining, then
5:24
it is it can start to approach content. Um. So I think sometimes
5:28
advertising is perhaps a non informative or perhaps in some cases a bit misleading.
5:34
Um. And in fact, we've lost some advertisers on Twitter because community notes
5:40
applies to advertising too, and so if somebody advertises something that isn't perhaps a
5:47
bit inaccurate, then it gets community noted, and then they get mad and
5:51
stop advertising. But we're we're care enough about the truth that we're willing to
5:57
give out averting, give up advertising dollars on Twitter even you know, in
6:00
order to have the most the least inaccurate, right. And I want to
6:06
talk about that, but I want to obviously spend some time on Tesla again.
6:11
Back to adds though, I mean, am I going to start to
6:12
seek Tesla ads? During the NFL broadcast on a given Sunday Monday Thursday,
6:17
The NFL's always on. Well, bar in mind. I mean I only
6:19
just agree to it, so it's like I have a fully formed strategy.
6:23
Okay, Well, how much time does go into when you think about these
6:26
things? I mean, was it something you decide to do up on stage
6:29
or at least did you know coming in I may you know this comes up
6:32
that I may go there? No, I mean the questions are not scripted
6:36
in any way, right, So so in that moment, you just said,
6:41
all right, we'll do some ads. That's right. So We're only
6:44
like twenty minutes later and I'm asking you detailed questions. Yeah, have any
6:47
answers? I guess I'm not sure what the most spectrum thing is, except
6:50
that, like I said, I believe that advertising should be informative about a
6:55
product. It should be ideally esthetically pleasing. It should be you know,
6:59
beautiful, should have some autistic element to it, and it should be something
7:02
that you don't regret watching after it's done. And I think if advertising fits
7:10
those criteria, that it starts to approach content, and you want advertising that
7:15
is as close to content as possible, such that you don't regret the time
7:18
you spent watching it, but you did say something as well, which is
7:21
that you know you've been preaching to the converted. You mentioned a few statistics,
7:26
for example, about safety, about price during the meeting that may not
7:30
be fully understood by the general auto buying public at large. Yeah, so
7:35
a lot of people still think Teslas are super expensive because we did start out
7:40
with an expensive sports car, then slightly less expensive sedan and suv. But
7:45
now we're at the point where the starting price of a Tesla is actually below
7:48
the average selling price of a car in the United States. So it Teslas
7:53
are actually much more affordable than people realize, and so we should just make
7:59
sure people at least know that, right um, and that it tells is
8:01
also the safest cars in the road US in so many ways that people at
8:07
a lot of people with Tesla don't even know, um, you know,
8:11
sort of the cabin overheat protection for example, or the fact that we continuously
8:15
update the automatic emergency breaking software and the way the airbags deploy. So we
8:22
look, we look at an accident, we say, is there anything we
8:24
can do from a software standpoint to improve the cars that are already on the
8:28
road, And then what can we update in the design to improve safety?
8:31
So um, and I think that you know, the statistics speak for themselves.
8:39
This is not simply a matter of opinion. It's statistically it's safer than
8:41
anything else. Um. You mentioned affordability. It's something you discussed during your
8:46
meeting. You've also discussed it a bit on the call on the earnings call
8:50
recently as well, in relation to the Federal Reserve raising interest rates and obviously
8:54
the monthly cost for a typical autobuyer. You think it's going to be a
8:56
tough twelve months ahead. You've said that a few times times. Yes,
9:01
for everyone not just has understood. So it's um, I mean simply uh,
9:07
you know you can you can think of raising there fred rate as as
9:11
somewhat of a brake pedal and the economy frankly, so it's um. It
9:16
makes a lot of things more expensive, certainly things that are bought with credit,
9:20
but then it has downstream effects on on even things that are on board
9:22
with credit. So um, you know, if if the car payment or
9:26
your home mortgage payment is absorbing more of your monthly budget, then you have
9:31
less money to buy other things. So actually it affects everything, even those
9:35
that aren't things that aren't bortant or on credit so UM and my concern with
9:39
the with the way that the Federal Reserve is making decisions is that they they're
9:43
just operating with um too much latency. Basically, the data is somewhat stales,
9:48
so they So the Federal Reserve was was slow to raise interest rates um
9:54
and and now I think they are slow to they're going to be slow to
9:56
lower them. That's that's that appears to be the case. Yeah, that
10:00
may well be. We spent a lot of time talking about that, perhaps
10:03
too much. But when it comes to latency, that takes me to pricing,
10:07
because you've discussed the lack of latency in your own ability to understand exactly
10:11
what's going on in the market for those cars as opposed to yea many of
10:16
the legacy and other Yes, yes, we have real time information on demand
10:22
so UM. So we know how many people place an order for it has
10:26
yesterday, so the computer calculates that will UM and literally every day we get
10:33
a you know, an automated email to the exact staff that says how many
10:39
people place an order in which countries for which cars, So we know what
10:43
demand, we know, we know what the orders were yesterday. Um.
10:46
And you don't want to overreact to these things because sometimes you get like little
10:50
dips for you know, reasons that are hard to explain. Do you look
10:54
at them every day? Yeah? Yeah. Um but like I said,
10:58
you don't want to over react to to you know, if like a week
11:01
is slow or something, you don't overreact to that. Um. But uh,
11:05
if but if you look at the trend, say for over two weeks
11:11
span or something like that, you can see that, Okay, there's this
11:13
for some reason the demand is is less than it was or it's higher.
11:18
Um. Just pricing Yeah yeah that I mean is that are are you almost
11:24
like an airline at this point in terms of kind of dynamic pricing model?
11:30
Yes? So, Um, we we're basically adjust pricing to match demand.
11:37
Um and um we obviously did a big price drop in Q one. But
11:43
you no, no, January is it's usually a terrible type of car binding.
11:46
Um. So there's there's seasonality to car purchases. Um with January.
11:50
January, January is often the worst month. So um, so we did
11:56
we did a big price drop and then and then then recently we did a
12:00
price increase. So, as I mentioned to the audience. The reality is
12:07
that all companies do significant All car companies make significant adjustments to price because you've
12:11
got the MSRP number, and then if demand is high, a dealist will
12:16
charge some premium over MSRP. If demand is lower, they'll they will have
12:20
manufacturing setups so that you can actually see a very big difference over the course
12:26
of say six months, between the peak to trough of all cars. It's
12:31
just that Tesla is so immediate and obviously transparent. It's not a question of
12:35
MSRP and then mock ups or discounts. Right, But after your last earnings,
12:41
a lot of investors came away wanting to talk about your margins and wanting
12:45
to talk about your pricing in particular. I think you know, you talked
12:48
about vehicle cost reduction. You sort of even said it's hard to explain the
12:54
profundity of technically selling cars now for no profit and still yielding tremendous economics over
13:00
time. Yes, I'd like you to explain a little bit more of what
13:03
that means. You did a bit in the meeting as well. You're talking
13:07
about sort of the move I think from manual to autonomous and the value add
13:11
that will come along with that, But explain to people why that's profound and
13:16
why conceivably not that you want to You could sell automobiles now for no profit
13:20
and still make enormous profit in the future. Yeah, so Tesla is the
13:26
only car company selling cars that where we believe the car is capable of achieving
13:31
full autonomy with a software update. So the value of a folios or folio
13:35
autonomous car is we think perhaps five times more valuable than an autonomous car.
13:45
Why, well, the utility of a car, typically a passenger car is
13:48
going to be maybe ten ten hours a week, maybe twelve if you say,
13:54
like somebody's going to drive an hour and a half a day on averaging,
13:56
so maybe maybe in our commute per day and then an occasional long trip.
14:01
But figure it's like ten twelve hours a week is typical for um,
14:07
a passenger vehicle. UM. And then, uh, you also have a
14:13
lot of costs associated with parking. You need a garage, or you've better
14:15
buy a parking space, or you've got to get you know, get a
14:18
parking ticket at them at the mall. It's there's there's a lot of costs
14:20
associated with cars. UM and uh, you know if you've got a car
14:26
that's autonomous, UM, that can go around and essentially be like an autonomous
14:31
uber. UM. The utility I think is going to be what's gonna be
14:35
much higher? Perhaps you know? And then this again there was there's so
14:39
much speculative I understand we're talking about robot taxis here or at least what people
14:43
have called and you have called. Yeah, autonomous uber is what you think
14:46
about. So perhaps the utility that would be on the order of fifty hours
14:50
a week. It's just a guess, say there's one hundred and sixty eight
14:54
hours a week, UM, and probably as rough guests an autonomous car is
14:58
we'll be able to be active instead of four or four ten hours a week,
15:03
probably in argue for about fifty But it's the same car, so the
15:07
and it costs the same to build. Um, You're I want to understand
15:13
the business model a little bit because I'm buying the car and instead of it
15:16
parking at my lot while I'm working, it goes off and picks people up
15:20
and drops them off. Yes, who's making the money from that? I
15:24
assume that that's the value add you're talking about. Is it a revenue share?
15:28
Do you have this model sort of planned out specifically is how it would
15:31
look. Um, yeah, it's it's it's been tens of terms and conditions
15:35
for for quite a long time. It has. Yeah, So the the
15:39
owner of the car would make I don't know some amount, who knows what
15:43
would be, but perhaps, uh, you know, it could be a
15:48
fifty fifty split or seventy thirty I don't know. But um, the cars
15:54
are are if you buy it tells a car it can only be used in
15:56
the tells a network, I cannot be used in someone else's at work.
16:00
So um that that means that if the car is able to be used five
16:07
times as much and and Tesla is likely to make basically two or three times
16:14
the original value or sale value of the car um in in robotaxi revenue.
16:19
Right. This this is this is gigantic. Um as it said, it'll
16:25
be like selling cars for software margins because in fact it is software. Right.
16:30
So um, instead of effectively having say twenty five cent margins, it
16:37
might be seventy seventy percent or more. Um and uh, I mean the
16:47
pre cash flow associated with that it is actually truly a staggering amount. The
16:52
best analysis that I've seen thus far about This is from Kathy Woods from ARC
16:56
invest Yeah. She's great. Ye, I would assume you like her obviously,
17:00
she's been a used proponent of your stock for many years. Was right,
17:03
by the way, very early on she was right. Yeah. Um,
17:07
so I thought she was being optimistic, but I she turned out to
17:12
be in fact almost spot on. Yeah. Um, and I am familiar
17:18
somewhat with her analysis. What a lot of investors come back to where everybody
17:22
is your promises of full autonomy. Um, and you seem to be saying
17:26
again it's close. Yeah, you said it before, you know that?
17:32
Yeah, So, I mean you said nineteen, I think about twenty and
17:36
I've certainly been Um. You know if I guess I have somewhat of a
17:41
pathological optimism. Um, well, it works, by the way, because
17:44
you don't end up with this unless you're optimistic. But yeah, exactly,
17:47
I mean, who would try to do this and this and rockets if they
17:51
weren't pathologically opted? No, you'd have to be. Yeah, so there's
17:55
a you know, there's a double edged sword, I guess, um.
17:56
But but you know, do you ever say to yourself, are I'm getting
18:00
a little ahead of myself here? I mean when you said again, we're
18:03
gonna have full autonomy. No, I didn't say, well, I said,
18:04
I think you think, my opinion, we probably will. Okay,
18:08
right, there's a number of qualifiers there. I think my opinion, we
18:11
probably will. So what do you think in your opinion you probably will win?
18:17
I mean, it does look like it's going to happen this year.
18:18
Why, well, we're not at the point where the car can drive on
18:25
highways and in cities with and where a human intervention is extremely rare. So
18:36
I mean, just I was able to drive for several days just dropping a
18:41
navigation pen in random locations in the Greater Wilson area with no interveentions on the
18:45
same in San Francisco, which is a very dificult place to drive. So
18:48
I mean it's you got bus lanes, one way streets. You know,
18:55
it's quite quite a challenging a homeless situation. You were driving recently in San
18:59
Francisco. The car was I have been doing doing that quite a lot because
19:02
of the Twitter's headquartered quarter. Maybe less, maybe a little less soon.
19:08
Yeah, well we'll get to that. But finished your point though, in
19:11
terms of why you feel more why you're optimistic. I'm optimist for that reason,
19:18
which is that the I've actually never, um, the first time I've
19:23
had a situation where there's been several days in a row going picking random destinations
19:32
and having no safety related intervention, So the first time ever. UM.
19:37
I want to get to a lot of other topics, but you know,
19:41
I am curious on Tesla in particular. UM. One thing I haven't heard
19:45
you discuss is the impact on your business of the Inflation Reduction Act. At
19:49
least I haven't heard it. Maybe you've discussed it. UM. You know
19:52
Adam Jonas, who follows your company, Morgan Stanley said, we believe the
19:56
IRA is so advantageous to Tesla in terms of absolute dollar versus the legacy OAMS
20:00
it could equal as much as forty five percent of current earnings per share.
20:06
Is the IRA going to be usually beneficial to your company? I'm not sure
20:10
it'd be it's gonna be that beneficial. Uh, there's UM, there are
20:17
there are significant advantages for having domestic production of batteries, which I think is
20:23
important for strategic reasons. You know, I think as we moved to an
20:32
electric transport economy, I think it's important for the United states to UM have
20:36
some independence with respect to battery production, including all of the precursors necessary for
20:41
a battery. So the the IRA strongly incents that, and I think that
20:47
will prove to be something of important of national strategic importance in the future.
20:55
So I'm not sure I would say it's as good as a forty five percent
20:59
increase and only share or it's it's a it's helpful. Yeah, yeah,
21:04
I would say it's maybe it's a half that or something, okay, But
21:08
I mean it's quite significant. Have you mapped that at all? Do you
21:11
have a sense. I mean, obviously you talk about storage, for example,
21:15
which is so important. You spend time on that as well with megapac.
21:18
Is it beneficial in that particular area beyond obviously the the credits, the
21:22
tax credits that certain people get for buying a Teslam. Yeah, there's two.
21:26
There's really two posts to the IRA. There's the incentive for battery manufacturing
21:34
UM and it's really quite a detail. It's's actually a very well written
21:37
uh well in that it's it really makes sure that you can't game it,
21:42
you know, so you actually have to build the batteries in the US,
21:45
and you actually have to build the precursis to the batteries. Um. But
21:48
if you do it is um Uh, I believe thirty dollars at the at
21:55
the cell level and fifteen dollars I believe partly so that that is very significant
22:03
batteries. And then you've got the consumer tax credit for evs provided they are
22:08
built in the US. Uh. And and that the package built in the
22:12
US. So UM, no, this this this is the really the first
22:18
time that we've we've actually had the you know, the first time in many
22:22
years that Tesla has actually had the consumer tax credit. Yes, because we
22:26
actually exceeded the limit for the consumer tax credit several years ago. Um.
22:30
And so up until December, until January this year, I should say,
22:36
there was no consumer tax in center to buy a Tesla, but there was
22:40
for the other car companies because they had not reached They've not made that many
22:44
electric cards, right, they didn't reach their moment. Yeah. Um.
22:47
You know, before I wrap on Tesla, I would like to get to
22:49
China, which is obviously an important market for you. Um. You know,
22:52
I noticed that chart you put up by D's the only company that seems
22:56
to be making money so far. Um, are you concerned at all about
23:00
UM, you know, ruthless pricing sort of at the low end and the
23:03
Chinese market and what that will mean to your market share though well, currently
23:08
a limitation UM with it is that is the is the production rate of our
23:15
Shanghai factory. It's not demand so UM. I mean, so that's there's
23:25
some constraints on our ability to expand in China and UM and so it's we're
23:30
making as many conis we can't it's not a demand issue. And you can
23:34
be making batteries soon as well. Yes, yes, UM for the Mega
23:40
pac UM for for mostly for non US markets. So are you concerned at
23:45
all about the growing belligerence between China and the US? UM. I think
23:49
that should be a concern for everyone. I think you're right. I think
23:55
it is shared by many people who run large organizations and smaller ones. Do
24:00
you think, for example, China will will make a move to take control
24:03
of Taiwan? The official officially, the official policy of China is that Taiwan
24:14
should be integrated. One does not need to read between the lines. One
24:18
continually read the lines, do you think so? I think there's a certain
24:22
there's some inevitability to the situation that would not be good for Tesla conceivably or
24:30
for any any company in the world. Frankly, yes, for any company
24:34
in the world. I think most almost no one really realizes that the Chinese
24:41
economy and the global the rest of the global economy are like conjoin twins.
24:47
It would be like trying to separate conjoin twins. That that's the severity of
24:51
the situation. Um. And it's actually worse for for a lot of other
24:56
companies than it is for for I mean, I'm not sure, I'm sure
25:00
where you're gonna get an iPhone for example. UM, and Apple's recently started
25:04
doing some some small amount of production in India, but it's tiny compared tiny,
25:08
not to mention an advanced semiconductor chip if they take over Taiwan Semi.
25:12
Correct, you design your own chips, but you manufacture them a Taiwan Semi
25:17
too, right, we do some we do. We do. We use
25:21
Samsung an TSMC um. But uh, you seem to think it's it's likely
25:26
to happen. I'm so only saying that that is their policy, and I
25:33
think you should take their word seriously. They mean it. Um, you
25:40
know your time. I think you've said in fact, some of the brief
25:42
conversations we've had, you know, it's one of the most valuable things in
25:45
your own control. And I am curious now. And Tesla investors sort of
25:49
seemed to rejoice at the announcement of a new CEO of Twitter, in part
25:52
because they thought, well, he's gonna have more time for Tesla. That's
25:56
true. Is it true? How do you see sort of allocate your time
26:00
now that you will suit no longer be the CEO of Twitter? Um?
26:03
Yeah, so, um, you know it was. It was really a
26:08
very very challenging um situation at the point in which the Twitter position closed because
26:14
um, you know, in rough terms, uh, there was a billion
26:19
and a half dollars of debt sipcing was added. Um, while at the
26:22
same time there was a massive side exactly you've said, you were four months
26:26
away from bankruptcy. It was three billion dollars in a year, conceivably negative
26:30
three billion dollar cashload, and had a billion billion dollars in the bank.
26:33
Yeah. Um, no, that's not good. No, anybody who knows
26:36
anything about corporate finance knows it's bad. No. I mean, I say,
26:40
like the analogy as using was, it's like being teleported into a plane
26:42
that's in a nose dive, headed to the ground, with the engines on
26:47
the fire and the controls don't work. Right, you were the guy that
26:51
you chose to get into the plane. Well, well, I did try
26:55
to exit the deal and it wouldn't let me. I mean, it was
26:57
it was a funny situation where when when I first first proposed the acquisition,
27:02
they said, hell no, they adopted a poison pill, basically saying they'd
27:07
rather die than be then be acquired, like they'd rather chew on cyanide.
27:14
Then a few months later that they're like, basically have a gun to my
27:17
head saying no, you must acquire us. I mean, that's quite a
27:19
change. It is quite a change. But you seem to think I can
27:23
figure out a way out of this, until finally, I guess you realized
27:26
that I can't. It was made clear to me that the decision would not
27:29
come down in my favor legally. Yeah, all right, Well, now
27:33
you have a new CEO. Why I think little Linda Yaqoan is gonna be
27:38
great? Why? Um, well, Twitter is it's very much an advertising
27:44
dependent business. Linda's obviously incredble with that, and she's just a great executive
27:48
in general. Um so, the and you know, my skills and interests
27:56
are in technology. Um so, you know I will continue to play a
28:00
role advancing the software. Um and you know, you know, getting the
28:07
futures and product stuff basically. M So. I mean the general idea is
28:15
Lenda will operate the company and I will build products. Yeah, and that's
28:18
I think that's that's that's a good. That's a good, and that will
28:21
take less of your time. Yeah, and I have I have that situation
28:25
at SpaceX. Um you know, Gwyn Shotwell's amazing um uh, you know
28:32
operates the company and I and I booken the sort of advanced technology, right,
28:36
both sort of designing in your rockets and that kind of thing. And
28:38
then that's uh, you know, every once you play to their strengths.
28:41
And that's that's that's mine. That's although I wonder you mean in twenty twenty
28:45
one you said you spend most of your time on the development of Starship.
28:48
I think you've also been quoted, I've read I think I heard you say
28:51
actually most of your mental energy at certain points has been on Starship. Yes,
28:55
So I wonder have you been diverted from Starship a little bit and we'll
28:57
return to it as the main focus. Or is it Tesla you know again,
29:03
your time, I know, having tried to schedule this interview with time.
29:07
Time treach is a real thing. Yeah. In fact, I i'd
29:12
say that the only real currency is time. Time is the only true currency.
29:17
So where you're going to spend the currency now that you don't have to
29:18
spend as much on Twitter, Well, I'm going to be devoting a lot
29:22
more time to to Tesla UM and especially on the AI development UM and your
29:33
product development, the and and then there's and I'll also be allocating some more
29:38
time to getting Starship to orbit. UM. In the case of SpaceX UM,
29:47
what I was able to do there was as as the Starlink constellation I
29:52
started working, I was able to move some of the best people from the
29:56
Starlink program to the Starship program UM, so they were you know, the
30:03
timing kind of worked out well that that UM. I mean, we have
30:07
really some of the world's best engineers it SpaceX and Tesla really some incredible human
30:12
beings. There was some statistic you shared. I think three point six million
30:17
people applied for a job at Tesla. I think I didn't I wrote it
30:19
down. I didn't write it down here since member, but that was that
30:22
the number. Yeah, it's crazy. There's a lot of demand. One
30:27
hundred and twenty seven almost hundred twenty thousand employees, but three point six million
30:32
people applying for a job. Yes, and I mean there's you know,
30:37
we would only add like to twenty or thirty thousand jobs. So it's it's
30:41
it's sort of the acceptance rate for for Tesla is it's much more difficult to
30:48
get into Tesla or SpaceX than Harvard, right, the acceptance rate is even
30:52
lower one for two. Well, yeah, it's it's. It's it's the
30:56
excerptance rate is lower than than the most demanding universities in the world. It's
31:02
insane. Speaking of employment, though, you had seventy eight hundred people at
31:07
Twitter when the plane was nose diving, I think you're at fifteen hundred.
31:10
Now we're roughly sixty three hundred people. Were they all superfluous? No?
31:15
Not all. Do you figure out which ones aren't or is it a little
31:19
late? You know, sometimes it gets a little late. I mean,
31:22
look, desperate time was called for desperate measures. So um, there's no
31:26
question that some of the people who are let go probably shouldn't have been let
31:30
go because we we simply did not have the time to figure out we had
31:37
to be you know, make make widespread cuts to get the run rate under
31:41
control. So um you know. And uh so this is not to say
31:47
that hey, everyone who is like go from Twitter is like somehow terrible or
31:49
something. It's just we have to with very little information. You get get
31:56
the the the headcount expenses and that the non posonal expenses down to where we're
32:02
at least break even. And we're not quite a break even yet, but
32:06
we're close. We need to do it fast and if you if you,
32:09
if you do it fast, unfortunately, there are gonna be some babies thrown
32:14
out with the death orders. So um so, I definitely, I definitely
32:17
would not want to say, uh, you know, disparage you know,
32:22
any I understand. But it's funny, you know, I hear from other
32:24
I've heard from other tech CEOs quietly they look at what you did at Twitter,
32:30
and they sort of they won't admit it publicly, but you know,
32:34
they said, wow, that was something I mean to cut that deeply and
32:37
still be able to shroun the company. People will argue about how well or
32:40
things are running. Have you heard from any of the CEOs and said,
32:44
thank you for doing that giving us sort of runway or leeway to at least
32:47
make some cuts of our own. Um I have. I've heard that from
32:51
a few people, and and I've heard that through the grapevine. Yeah.
32:53
Um, but um, like I said, if desperate times call for desperate
33:00
measures, unfortunately that those were desperate times. So do you see starting to
33:02
I mean, we'll lend up hire people. Do you think are you reading
33:06
from people to be hired a twitter? You know, I think we absolutely,
33:08
um, I need to need to hire people, and and and if
33:13
they're not too mad at us, probably re hire some of the people.
33:15
But they were like go um, and they're gonna have to come into the
33:21
office. Look, I I'm a big believer that that that that people need
33:25
to are more productive when they're in person, um, and and and and
33:31
and really man and who the whole sort of work from home thing, it's
33:36
like, I think it's if there are some exceptions, but I kind of
33:44
think that that the whole notion of work from home is a bit like the
33:46
you know, the the fake Marie Antoinette quote let them eat cake. It's
33:51
like it's like, it's like, really, you're gonna work from home,
33:53
and you're gonna make everyone else who made your car come work to the fact
33:57
work in the factory. You're gonna make the people who your food that gets
34:00
to live it that they can't work from home. That you know, the
34:04
people that come fix your house, they can't work from home but you can.
34:07
Does that seem morally right? That's messed up? You see there's a
34:12
moral issue. Yes, I mean I see it more. It's a productivity
34:15
issue, but issue. People should get off the moral high horse with the
34:21
work from home bullshit. Um because they're asking everyone else to not work from
34:24
home while they do, and they're still pushback. By the way, it's
34:28
still going on. This battle is still happening. I mean leaders of organizations
34:31
and I speak to plenty of them. I want people back. I want
34:34
people back three days a week. They're still battling. It's not clear that
34:38
it's ever going to change to people are not people the laptop classes living in
34:45
Lala land. Okay, As I said, you can't but look at the
34:50
cars or people working from from home here? Of course not um so so
34:54
the people were, you know, building the cars, servicing the cars,
35:00
holding houses, fixing houses, making the food, um, making all the
35:04
things that that people consume. It's it's messed up to assume that that this
35:09
they have to go to work, but you don't. How is that that
35:15
is? It's not just a productivity thing. I think it's Marley Marley wrong.
35:20
Although productivity is definitely impacted too, and the ability people learn, I
35:24
mean on and on list. So I mean, look, you know people
35:29
will disagree with me about this, but I you know, it's like,
35:31
um, so if you want to work at Tesla, you want to work
35:35
at SpaceX, you want to work at Twitter, You've got to come into
35:37
the office every day. Yes, I mean you know, Like I'm not
35:42
saying, um, I'm saying, like, look, we put forty hours
35:46
in you know, and it frankly, it doesn't even even need to be
35:49
like, you know, Monday through Friday. You know, you work Monday
35:52
through Thursday. And also not saying no one should take I mean I think
35:57
people should take vacations, Like I work seven days a week, but I'm
36:00
not expecting others to do that. How much sleep do you get, by
36:04
the way, about six hours you do six hours typically. Yeah, no,
36:08
that's not bad. I thought it would be less. I've tried less,
36:10
but my productivity even though I'm awake more hours, I get lest one,
36:15
okay, and the brain pain level is bad if I get lessen in
36:17
six hours. But you work seven days a week? Yes, yeah,
36:22
I actually only um in terms of actually you say, like, how many
36:29
days in a year do I not put in some meaniqual amount of work?
36:32
It's probably about two or three, right, two or three days a year?
36:37
Yeah? A year? Yeah? Yes, um. I want to
36:42
get to some more specific questions about Twitter that's sort of have a more global
36:46
aspect, and I still want to get to AI. You're gonna give us
36:49
some time here, can we still? Um? Yeah? I think about
36:51
a board meeting coming up, but I'm happy to give as much time as
36:52
reasonally possible. I think it's maybe it's six or something. Okay, Well,
36:55
let's our time, isn't now? I love another twenty years? Another?
37:00
All right? At least twenty great? Um, let's talk about free
37:04
speech a bit. You know, you call yourself a free speech absolutist.
37:07
You want Twitter, and this isn't aspirationally. Aspirationally you want Twitter to be
37:09
as truthful as possible, most accurate source of information about the world. Um,
37:15
So what does that mean for how you police lies on the platform?
37:20
You mentioned community notes? Is that I think community Yeah? I mean,
37:24
I'd say so. My overall kind of vision for actual Twitter is to be
37:31
a cybernetic collective mind for humanity. This is going to sound quite esoteric and
37:38
sci fi bit um. So the if if you know, in pursuit of
37:44
that objective, you want to have information move quickly, have that information be
37:50
accurate, and you want to have error correction on that information. So you
37:52
can think of community notes as like an error correction on information in the network.
37:58
Um. And the effective community notes is actually bigger than it would seem.
38:01
It's bigger than the number of notes because if somebody knows that they're going
38:05
to get noted, they are less likely to say something that is false because
38:09
it's embarrassing to get community noted. Okay, And and that applies even to
38:15
advertisers. Right, We've lost that's so far forty million dollars in advertising because
38:22
it was misleading or because because the community notes get to two pretty big advertisers
38:29
got community noted. Um and I yes, UM, I think on balance
38:37
the community notes who are correct? And I did I did say to those
38:42
advertisers, look provide just go on Twitter and provide some facts that contradict the
38:51
community note. That's the way to deal with the community note is to say,
38:54
is that the community note is saying that the ad is misleading for the
38:59
following reasons. You've got the information that rebuts that note, then just add
39:05
that to the ad. Okay, well, we're coming up on an election.
39:07
I mean it's a ways a way, but it's going to all start.
39:10
Um. Yeah, President Trump is allowed back on the platform. He
39:15
hasn't actually come back, right, but one would imagine if and when he
39:17
does, or there are others who will say twenty twenty election was rigged?
39:22
Is that something I assume that saw something you believe? I well, I
39:28
think the answer the answer is is nuanced, Like do I believe Biden?
39:35
One? Yes, I believe you want and you voted for I did?
39:37
Actually do you regret that? I mean, man, I wish we could
39:49
have just a normal human being as president. That's what I wanted. I
39:52
think if if if we could, you know, there's that that old saying
39:55
of like we're better we're better off being run by um, by people picked
40:00
at random from the phone book then the faculty of Harvard. Yeah. I
40:04
don't know who said that, but someone very wise. UM. And and
40:07
I would I would say if we could do that for the president, that
40:12
would be So you're not obviously you're not happy with don't we all just want
40:16
an all human beings whatever? What I mean? Like you know, just
40:23
I don't know. So you want somebody's competent it's helpful, Yes, I
40:28
think definitely. Um, somebody's executive ability is underrated. Since the president is
40:32
effectively the chief executive officer of the country. UM, it actually matters if
40:37
they are a good executive officer. Yeah, it's it's not simply a matter
40:39
of do they share your beliefs and you know, um, but but are
40:44
they good at getting things done? There's a lot of decisions that need to
40:47
be made every day. UM, many of them are unrelated to moral beliefs,
40:53
you know. And and you just want a good executive because their CEO
41:00
of America they are they are. We want a good CEO of America who's
41:05
going to do being effective? Fortunately, we live in highly partisan times where
41:10
there is war about everything. Including ideas, including the truth, which gets
41:15
back to it's not true that the election in twenty and twenty was rigged.
41:20
It wasn't stolen. And I wonder on the platform when you see that,
41:22
does that end up in a community note or is that something you take more
41:25
action on? And obviously the place is so many I mean, to be
41:29
clear, I don't think it was there was a stolen election, um um.
41:35
But by the same token, if somebody's going to say that there's there's
41:37
never any election fraud anywhere, this is obviously also false. Yeah, if
41:43
you if you if one hundred million people vote, the probability that the fraud
41:46
is zero is zero. Of course, there's always going to be something.
41:51
I mean, perhaps this election was audited. It was so many judges.
41:57
I mean, it went on and on and on, and there was no
41:59
nothing whatsoever that. I don't want to debate this with you. My question
42:02
is more about what I think it's important to say, like that in any
42:06
given election, even if you try your hardest, if you've got one hundred
42:08
million votes, there's gonna be some some amount of fraud that is not zero,
42:12
and that that it's important to acknowledge that without saying that that the fraud
42:16
was of sufficient magnitude to change the outcome. So, uh so my opinion
42:22
would be that there was some there was some small amount of fraud, but
42:24
it was not enough to change the right and by the way, it might
42:27
have been either way. I mean, you know, yeah, this is
42:29
probably a little bit. But again it's gonna be You're gonna let people say
42:31
that though on on Twitter, and then you're gonna hope that they're corrected,
42:35
and they will be corrected. They will be Oh, let's talk a bit
42:39
about your tweets, um, because it comes up a lot um even today,
42:45
it came up in you know, anticipation of this. I mean,
42:49
um, you know, you do some tweets that seem to be or at
42:52
least give support to some who would call others conspiracy theories. Well yes,
43:00
but I mean honestly, you know, some of these conspiracy theories, uh
43:07
I've turned out to be true. Which ones, well, like the hunter
43:10
Biden laptop that's true. Yeah yeah, so, uh you know that that
43:15
that that was a pretty big deal. There was Twitter and and and and
43:20
others engaged in active suppression of information that was relevant to the public. M
43:23
that's that's a that's a terrible thing that happened. That's like your interference.
43:27
But how do you make a choice you don't see I mean in terms of
43:31
when you're going to engage. I mean, for example, even today,
43:35
elon you you you tweeted this thing about George Soros. Well, I'm looking
43:38
for because I want to make sure I quote it properly. But I mean,
43:42
you know what you wrote, but you basically this is like, you
43:45
know, calm down, people, this is not like you said, he
43:51
wants to rot the very fabric of civilization and so hates humanity. Like when
43:54
you do something like that. Yeah, I think that's true, that's okay,
43:59
But why share it? Why share it? Especially because I mean,
44:01
why share it when people who buy Teslas may not agree with you, advertisers
44:07
on Twitter may not agree with you. Why not just say hey, I
44:10
think this. You can tell me, we can talk about it over there,
44:13
you can tell your friends, But why share it widely? I mean,
44:19
this is a freedom of speech. I'm allowed to say whatever you absolutely
44:22
are, But I'm trying to understand why you do, because you have to
44:23
know it's gotta there. It puts you in a in the middle of a
44:28
the partisan divide in the country. It makes you a lightning rod for criticism.
44:31
I mean, do you like that. You know people today saying he's
44:36
an anti Semite. I don't think you are. No, I'm definitely.
44:37
I'm like, I'm like a pro Semite. If anything, I believe that
44:43
probably is the case. Why would you even introduce the idea of it that
44:46
that would be the case. I mean, it looks we don't want to
44:50
make this a George Stars interview. No, God, no, I don't.
44:52
I don't want it at all. But I'm what I'm trying even came
44:54
up then when the annual meeting. I mean, you know, do your
44:58
tweets hurt the company? Are there test loan ers to say I don't agree
45:00
with his political position because and I know it because he shares so much of
45:05
it? Or are there advertisers on Twitter that Linda Yak Greena will come and
45:07
say you gotta stop man, Or you know, I can't get these ads
45:12
because of some of the things you tweet. You know, I'm reminded of
45:28
as a scene in The Princess Bride, Great Movie, Great Um where he
45:39
confronts the person who killed his father. He says, offer me money,
45:52
offer me power. I don't care. So you just don't care. You
46:02
want to share what you have to say, I'll say I don't want to
46:05
say. And if, if, if the consequence of that is losing money,
46:08
so be it. Okay. But I mean when you when you when
46:15
you link to somebody who's talking about the guy who killed children in a mall
46:21
in Allen, Texas, you say something like it might be a bad siop.
46:25
I'm not quite sure what you meant in that particular case. Uh,
46:34
there was a somehow that that's that's not not not that the that the people
46:45
people were killed, but the it was I think incorrectly ascribed to be a
46:49
white supremacist action. Um. And the evidence for that was some obscure Russian
46:57
website that no one's everihood of that had no followers, UM and the like.
47:02
The company that that found this is belling Cat. Right. Did you
47:07
know what belling cat does? Psy ops? Right? I couldn't really even
47:10
follow exactly what it was you were trying to express there, So that's part
47:15
why I was curious. But I'm saying I thought that the ascribing a two
47:17
white supremacy you is bullshit okay, and and uh and and and that the
47:22
information for that came from an obscure Russian website and was somehow magically found by
47:29
Belling Cat, which is a company that does psy ops. And there's no
47:34
proof, by the way that he was not. There's no I would say
47:37
that there's no proof that he is. And that's a debate you want to
47:42
get into on Twitter. Yes, because we should not be ascribing things to
47:46
why supremacy? If if there if it's false. Okay, Um, can
47:52
we talk about AI now? Actually I want to talk about AI. Well,
47:54
let me end with Twitter in the sense of Sam Altman was on the
47:59
Hill today and he said AI's ability to manipulate interactive disinformation is a significant area
48:05
of concern. It's moving so quick, it is. And how do you
48:10
know, I'm curious to whether you agree with that? How you see that
48:14
even playing out on Twitter with people who know somebody could you know look like
48:17
you and or me and use our voice. I don't know what it could
48:22
be. Oh that that that that that is happening in big time. Um.
48:25
So the so the reason that I'm uh asking people to be verified on
48:34
Twitter and and and and that we're saying okay. Verification means you've got a
48:38
phone number from a reputable carrier, which means that you've at least passed through
48:44
whatever they're uh security mechanisms are that you have a credit card, which so
48:50
you now have passed through whatever security mechanisms the credit card company has. And
48:54
that there's some small amount of money paid for a month. Um uh.
49:00
That that set of actions um uh significantly increases the cost of of fake accounts.
49:07
Um and the with with modern with with with the latest AI um it
49:14
can bypass basically every test for are you a human? So so then how
49:19
do you know that a million accounts were created? How do you know that
49:23
those are people? I don't know. How do you exactly you have to
49:29
do? You have to do account verification? Um and and the thing that
49:32
makes the like I sort of put myself in a position of like, if
49:37
it was my goal to manipulate public opinion and create millions of accounts and make
49:42
it seem as though a topic was trending, um and and that and that
49:45
this is actually what public beliefs um and But in order to do so,
49:50
I had to get a million phone numbers, a million credit card numbers and
49:54
pay uh you know, eight bucks a month, um and and and have
50:00
that old not be traceable and clustered. I'd say it's impossible, impossible,
50:06
So that the goal of the uh of the sort of Twitter verification is fundamentally
50:13
to prevent AI manipulation of the system. UM. Final question on Twitter,
50:19
Walter Isaacson, your biographer, He said your big goal for Twitter is disrupting
50:23
the banking industry. Oh. Um, I'd say that's that's that's a that's
50:30
a look. First of all, I don't want to disrupt something with the
50:34
sake of disrupting it. It's more like, if there is a better product,
50:37
that's great. But I'm not output disruption for thesatic disruption. I'm like,
50:42
if if you're gonna make a product that improves a quality of life for
50:45
people that they find more useful, that that's great. UM and Uh,
50:51
what what people see in PayPal is sort of like sort of a halfway.
50:55
It's sort of a sort of it's frankly a half baked version of what it
51:00
could be. UM and um and so I think there's potential to create,
51:06
um a more efficient financial system. UM and here we can get again quite
51:14
quite esoteric sort of into sort of information theory. But um, the actual
51:19
financial system today is a heterogeneous set of databases running on mainframes and cobal um
51:23
that still engage in batch processing. It's really quite uh, very inefficient UM
51:30
so UM things are still not real time UM and and so it's possible to
51:35
have um a much more efficient homogeneous real time data system. Money is just
51:40
information um and and so the so. But that's not like the only reason.
51:49
It's it's just a thing that that would be I think poetic to fulfill
51:53
ultimately the vision that I had for X over twenty three years ago, and
52:01
and and actually see that count tuition would be nice. But but there are
52:06
many other things for Twitter, it's not just financials. Well, you talk
52:09
about enhancing humanity. You know, I'm curious than a bad AI, which
52:14
many people say will lead to great productivity gains. You showed those robots.
52:20
I mean, I can imagine what they conceivably could do empowered by AI.
52:22
But I'm also curious, because you've certainly been concerned, what percentage do you
52:27
give the chance that it will destroy humanity? Well, the advent of artificial
52:36
general intelligence is cool to seeingularity because it is so hard to predict what will
52:40
happen after that. I think it's it's it's very much a double double edged
52:47
sword. I think it's there's there's a there's a strong probability that it will
52:51
make life much better and that will have an age of abundance, and there's
52:58
some chance that it goes wrong um and UH destroys humanity. Hopefully that chance
53:04
is small, but it's not zero um. And so I think we want
53:06
to take whatever actions we can think of to minimize the probability that AI grows.
53:10
You've called for a pause, along with the number of other people.
53:13
Yes, I look when I call for the A friend of mine, Max
53:15
Tegmark, as a business said at M I t um, you know,
53:21
I wanted me to sign onto the letter, and it's it's like I knew
53:22
it would be futile. It is futile. I knew it would be futile.
53:27
I just want to could call it like it's one of those things.
53:29
Well, for the record, I recommended that we pause. Uh. Did
53:32
I think we would there would be a pause? Absolutely not. So you're
53:37
starting uh x AI, I think that's what you're calling it or some new
53:44
AI effort. UM. How is it going to be different than Open AI
53:49
or alphabet this is this is not the time to to this. I we
53:57
don't have enough time and and order. Is this the moment to really talk
53:59
at it? Um, we will have a launch event and will explore the
54:05
issues in more detail. But UM, I should say that that, and
54:12
I mentioned this at the shareholder meeting, that UM Tesla has actually um tremendous
54:20
capability in real world AI. Yes, in fact, it is very far
54:23
ahead of of anyone. I know. People actually on Twitter prior to our
54:28
interview, we're saying, you know, he never gets asked about how advanced
54:30
his AI is A Tesla. You always talk about the other names. TESLAI
54:37
is like said by It's like there's not I'm not even sure who's second.
54:40
Frankly, Um, why is that? What? Then? What? What
54:44
are people not understanding about what you have? And why are we talking so
54:45
much about chat GPT in generative AI, that open AI and what Microsoft's cann
54:51
be able to do with it and not about Tesla. I don't know.
54:53
I mean people do talk about it online. UM. I think I think
55:00
Tesla will have sort of a chat GPT moment maybe that If not this year,
55:04
i'd say no. Later the next year, Um, wait, say
55:07
that again. I'm sorry, You're gonna have a sort of chat GPT moment.
55:10
Oh you will. In terms of suddenly it will Yeah, suddenly,
55:14
three million cars we will drive themselves with no one. Right, it goes
55:15
back to that, right, yeah, and then five million cars and then
55:19
ten million cars. So um, the and and and and and I would
55:24
I would also say that if if positions were reversed, um and say um,
55:30
Well, in fact, the positions are are reversed. For example,
55:32
Google has way Moo, which is, you know, sort of attempting self
55:37
driving, and they are able to make self driving work in a very limited
55:42
geography with with very tightly mapped streets. But as soon as any as soon
55:45
as that anything goes wrong with those streets, like there's an accident or a
55:49
parade um or road construction, uh, it stops working. Basically, Google
55:55
is unable to produce a generalized alution to self driving that works anywhere. They've
56:01
been trying to that for a long time they have been unsuccessful. Tesla basically
56:06
has that and is far more advanced that than Google. Yea, and so
56:12
if the positions were reversed and you said, okay, um, Tesla's got
56:16
to produce a large language model that has output equal to a greater than Chat,
56:22
GBT or Microsoft. Opening Eye has to do self driving, and we
56:27
just we flipped the tasks. Tesla would win, you'd win. Yes,
56:30
you have a computing power and everything else. You have to do it absolutely.
56:35
Um I'm being told we don't have that much time. You can you
56:37
give me another five minutes or yeah, I think so, I do have
56:42
a board meeting. I know I five five minutes is probably fine. Sure,
56:45
okay, thank you, um um open Ai. I mean you seem
56:52
somewhat frustrated with them. You were one of the big contributors early on the
56:57
reason I am the reason Opening Eye exists. How much money did you give
57:00
them? Um so. Uh, I'm not sure the exact number, but
57:07
it's some some number on the order of fifty million dollars. So the man
57:14
fate loves irony next level, um So. I used to be close friends
57:20
with our page, and I would say at his house and we'd have these
57:22
conversations long long into the evening about AI and I would I would be constantly
57:28
urging him to be careful about the danger of AI, and and and and
57:32
he just he was really not concerned about the danger of AI, and was
57:37
quite cavalier about it. Um and and and and At the time, Google,
57:45
especially after their acquisition of deep Mind, had three quarters of the world's
57:49
AI talent. They had obviously a lot of computers and a lot of money.
57:52
Um, so it was a unipolar world for four AI and and we've
57:55
got a unipolar world. But but the the person who who controls that does
58:00
not, or at least do not seem to be concerned about AI safety.
58:05
That that sound that sounded like a real problem. So um. And then
58:09
the final straw was Larry calling me a speciist for being um pro human consciousness
58:16
instead of machine consciousness. And I'm like, well, yes, I guess
58:20
I am. I. I am a specist. And so right, so
58:23
you helped to the creation of open AI. You put as much as fifty
58:27
dollars more than help. It wouldn't exist with that. It wouldn't exist with
58:30
that. I came up with a name. The name opening Eye refers to
58:34
open source. So it's the intent was what's the okay? So what was
58:37
the opposite? What's the opposite of of Google? Would would be an open
58:43
source nonprofit because Google is closed sourced, poor profit um and that profit motivation
58:49
can be potentially dangerous. Um. So uh, should you've gotten governance for
58:54
that money? Should you have gotten some level of control perhaps in retrospect?
59:02
Yeah, I fully admit to being a huge ilient here um so um anyway,
59:10
So, so Opening I was meant to be. Opening I opened as
59:14
an open source. It was created as a buy one C three um and
59:19
but but so so part of it is also in the beginning, I thought,
59:22
look, this is this is probably a hopeless endeavor. How could we
59:25
possibly compete with How could Opening I possibly compete with with Google Deep Mind?
59:30
I mean this, this seemed like an ant against an elephant, you know,
59:34
which is not a contest um and um. And I was also,
59:45
I mean I I was instrumental in recruiting that the key scientists and engineers,
59:53
most specifically, most notably Elias skier Um. In fact, Elia went back
1:00:00
and forth several times because he would say he's going to join an Opening Eye,
1:00:05
then Demis would convince him not to. Then I would convince him to
1:00:07
do so, and then and this went back and forth several times, and
1:00:10
ultimately decided to join an Opening Eye, and really Eliot joining was the was
1:00:15
the lynchman for Opening Eye being ultimately successful. You're very disappointed in what's happened
1:00:21
there. In terms of it becoming a for profit. I do think that
1:00:30
there's some Look, it does seem weird that something can be um a nonprofit
1:00:39
open source and somehow transform itself into a for profit closed source. UM.
1:00:46
I mean this would be like like, let's say you funded an organization to
1:00:50
save the Amazon rain forest and instead they became a lumber company and chopped down
1:00:54
the forest and sold it for money. And you'd be therefore like, well,
1:00:59
wait a second, that's the exact opposite of what I gave the money
1:01:01
for. Is that legal? That doesn't seem legal? And if it is,
1:01:08
and in general, if it is legal to start a company as a
1:01:14
nonprofit and then take the IP and transfer it to a for profit that then
1:01:19
makes tons of money, shouldn't everyone start? Shouldn't that be the default?
1:01:25
And and and then I also think it's important to understand the like like when
1:01:31
push comes to shove, let's say they do create some some digital superintelligence,
1:01:36
almost godlike intelligence, Well, who's in control and what exactly is the relationship
1:01:42
between Open and I and Microsoft, and I do worry that Microsoft actually may
1:01:46
be more in control than say, the leadership team and Opening Eye realizes.
1:01:51
I mean Microsoft had as part of the Microsoft investment, they have rights to
1:01:54
all of the software, all of the model weights, and everything necessary each
1:02:00
run the inference system, so they essentially have a great deal of control.
1:02:05
At any point, Microsoft could cut off Open the Eye. Elan, I'm
1:02:09
being told we have to wrap up. Your board has been very satient.
1:02:14
I want to end on one AI question. You have a lot of kids.
1:02:17
I have some kids. I have one who's actually soon to go into
1:02:22
the workforce. I struggle with how to advise him about a career when this
1:02:27
technology exists and will only improve. I'm just curious when you think about advising
1:02:31
your children on a career with so much that is changing, what do you
1:02:35
tell them there's going to be a value. Well, that's a tough question
1:02:52
to answer. I guess UM. I would just say, you know,
1:03:02
to to sort of follow their heart in terms of what they find interesting to
1:03:07
do or fulfilling to do. UM and UM, you know, try to
1:03:13
be as useful as possible to the rest of society. UM. You know,
1:03:20
if we do get to the sort of like magic Genie situation where um,
1:03:23
you can ask the A for for anything. Um. And let's say
1:03:29
it's even the benign scenario. Let's say it's a benign scenario. How do
1:03:31
we actually find fulfillment? You know, it's a how do we find meaning
1:03:37
in life? If the air I could do your job better than you can.
1:03:40
Um. I mean, if I think about it too hot it correctly,
1:03:47
it can be uh just dispiriting and demotivating. Um, Because I mean
1:03:54
I I go through I mean I I've put a lot of blood, swen
1:04:00
and tears into building companies. And then and then I'm like, wait,
1:04:03
well should I be doing this? Because if I'm sacrificing time with friends and
1:04:09
family that I would prefer to do. But then ultimately the AI can do
1:04:14
all these things. Does that make sense? I don't know. To some
1:04:19
extent, I have to have deliberate suspension of disbelief in order to remain motivated.
1:04:26
So I guess I would say, just you know, welcome things that
1:04:32
you find interesting, fulfilling and that contribute some good to the rest of society.
1:04:39
Well, that's a great place to end. There's so much we didn't
1:04:41
get to. I hope you'll give me another chance to sit down with you.
1:04:44
But Elan, thank you for being so generous with your time, Thanks
1:04:46
to your board for waiting as well, and thanks for having us here at
1:04:49
this incredible facility. Appreciate it. Thank you all right, pleasure Carl.
1:04:56
I'm going to send it back to you numberststst