Sam Altman Talks GPT-5, AGI, ChatGPT, OpenAI...
Transcript
Coming up...0:00
gbt five six and seven I think we'll0:02
continue in future years I really do0:04
think we'll look back at this like we0:05
were all living through one of the most0:07
important periods of human Discovery I0:08
think this will be that big of a dealIntro0:10
there is so much wisdom in the same0:12
ultimate interview yet it doesn't even0:14
have 40 000 views Sam talks about GPT 50:17
AGI the history of openai nuclear fusion0:20
and much much more my name is David0:22
Andre and I'll be covering every single0:24
Sam ultimate interview until he does oneWho is the interviewer?0:27
with me oh and the interviewer is Tobias0:29
billionaire founder and CEO of Shopify0:32
the interview actually started with a0:34
little role play when Toby was the0:36
prompt engineer and Sam was the AI soSystem message0:39
let's start with the system message you0:40
are a CEO trained in part by open AI0:44
please follow the instructions carefully0:46
so how did the day when chair GPT gotThe day ChatGPT was released0:49
released look like was it one of those0:51
days that you can just remember it0:53
forever yeah it was uh woke up that0:56
morning like we've released many other0:58
products and put it out there and we0:59
were like we hope people like it people1:01
did like it people did like it yes and1:04
by I would say two o'clock that1:06
afternoon it was clear that something1:09
big was about to happen next Sam AltmanHow ChatGPT grew in popularity1:11
talks about why word of mouth is1:13
absolutely essential for a product to go1:16
viral you know you always start with1:17
like the tech hyper early adopter crew1:20
but they were so excited about it they1:22
were like telling their friends and and1:23
that morning and then by that afternoon1:25
I was starting to hear from people who1:27
like don't normally follow Tech which is1:29
like I saw this thing it sounds wild1:31
like can you let me try it out or how do1:33
I do this by the fifth day we had gotten1:35
a million users kept going from there1:37
and Sam isn't the only entrepreneur whoWhy word-of-mouth is so huge1:40
thinks this Alex horomozi is also big on1:42
Word of Mouth takes a tremendous amount1:44
of effort to build an exceptional1:45
product that grows on its own via word1:47
of mouth because of all acquisition1:49
channels only one of them is quadratic1:51
in nature what's funny is that chair GPTThe speed of ChatGPT's success1:53
grew so fast that Toby didn't believe1:56
what he just heard how long did that1:58
take you to Millions five days2:01
um2:02
the reason this moment was hilarious is2:05
because it took Shopify years before it2:07
reached 1 million users well it took2:09
chai GPT just five days but Sam didn't2:12
expect chair GPT to be so insanelySam didn't expect the level of success2:14
successful we put it out like not2:16
expecting this speeder well no one would2:19
expect the speed of growth because it2:20
hadn't happened but we did we did not2:21
expect it according to Sam this is theTechnology that's actually fun2:24
first time in a while when technology2:26
was this fun I actually think that's one2:28
of the good parts about this is2:29
technology has not been that fun for a2:31
while it has not felt like a new sort of2:33
Frontier and this is like fun and it's2:36
certainly a new frontier but it's very2:38
tempting to anthropomorphize it and I2:40
think we have to like figure out a way2:42
to still have the fun2:44
but remind people that this is a toolWhat "anthropomorphize" means2:46
and not a creature by the way if you2:48
don't know what anthropomorphize means2:50
it basically means making it like a2:52
human implying human qualities onto an2:54
AI which is a completely different2:56
entity it's just you know a language2:58
model it might become a conscious3:00
creature in the future but right now3:01
it's just a piece of code now I foundPrevious GPT versions3:03
the next part quite interesting because3:05
Sam explains how the public perceived3:08
each version of GPT gpt2 in 2019 people3:12
tried it and said this is like a dead3:14
end direction scale that up gpp3 people3:16
are like oh okay GPT 3.5 was like3:18
finally usable and gpt4 people are3:20
building entire companies around so whatWhat about GPT-5?3:23
about gpt5 by the way this is one of the3:25
rare interviews where Sam Actually talks3:27
about gpd5 he mentions it multiple times3:29
throughout the interview gbt five six3:31
and seven I think will continue in3:33
future years on this trajectory of3:36
really increasing the utility they can3:38
provide and this is like a a big new3:40
exciting thing to happen in the worldWhy Emerging Properties matter3:42
Sam actually isn't interested in how the3:44
models will do on specified tests like3:47
the bar exam or some medical tests what3:49
he thinks is much more important is the3:51
emerging Properties or the new3:53
capabilities of those models basically3:55
what the model can actually do we can3:57
predict how to score on some tests we're3:58
really interested in which gets to the4:00
latter part of your question is can we4:03
predict the sort of the qualitative new4:05
things just the new capabilities that4:07
didn't exist at all in gpt4 that do4:10
exist in future versions like gpt5 that4:13
seems important to figure out but right4:15
now we can say you know here's how we4:16
predict it'll do on this eval or this4:18
metric now if you don't think that we'reImportant point in human history4:20
living in one of the most important4:22
points of human history then listen to4:24
this and I really do think we'll look4:25
back at this like we were all living4:27
through one of the most important4:29
periods of human Discovery I think this4:30
will be that big of a deal guys you have4:32
to understand that technological shifts4:34
like this happen once in a lifetime so4:37
if you're considering doing something in4:38
the AI field then start now don't wait4:41
because the AI Revolution will happenAI discovering new Science4:43
without you AI can can do a lot of4:45
things but both same things really4:46
matters is the moment when AI starts4:48
discovering new science when the AI4:51
systems can either semi-autonomously or4:55
just by helping us a lot really discover4:57
new science and if the rate of4:59
scientific progress that Humanity makes5:01
increases by a factor of 10 or 100 or a5:04
thousand in a year that somehow feels5:06
different than just a bunch more remote5:07
employees however AI is one of thoseThe moving goal post of AI5:10
fields where the goal post keeps5:12
constantly moving like first people said5:13
that AI won't be able to be creative but5:16
then we got mid-journing so those same5:18
people started saying that okay but5:20
programming programming will never be5:22
able to be accomplished by language5:23
models well look how good gpd4 is at5:26
programming look certainly by the time5:28
AI does original science we'll move the5:29
goal posts again and say well that just5:31
wasn't that impressive in the firstPast AI predictions5:32
place one of the funniest and also5:34
scariest facts about AI is that5:36
virtually all past predictions were5:39
totally wrong you look at the prediction5:40
from maybe 10 years ago maybe even five5:43
I think most experts would have say5:45
first AI comes for physical labor it's5:47
going to drive trucks it's going to work5:49
in factories then it comes from this5:51
sort of easier parts of cognitive labor5:53
then it comes from the stuff that's5:54
really hard maybe it can be you know5:56
maybe it can write computer code someday5:58
maybe not and then maybe someday in the6:00
distant future but probably never it can6:02
do creative work and of course it's gone6:04
the exact opposite direction every6:06
almost everybody predicted this wrongWhen will AI do science?6:07
now this is a very interesting take by6:09
Sam he thinks that we will actually have6:11
ai that can do new science before we6:14
will have advanced robots the fact that6:16
it can do this this sort of creativity6:19
and the fact that it can use code to6:21
verify things actually gives me hopes6:23
that we may have an AI that can do6:25
science before we have that factory6:26
robot that we can do everything else6:28
next Toby and Sam talk about theThe importance of Programming6:31
importance of programming and how it6:32
improves your reasoning abilities at6:35
some point the model is called trained6:36
on code and their ability to reason6:38
became significantly better which in my6:41
experience by the way it's the same for6:43
exact steps that I work with so when6:46
they learn to code it's important got6:49
trained on code are much better at6:50
reasoning so if some of you areWill Programmers be replaced?6:52
programmers and you're worried that AI6:54
will replace you just know that the6:55
skill of programming knowing how6:57
computers understand and process code6:59
will always be valuable Sam Altman isPutting Video into new AI models7:02
particularly excited about training new7:04
AI models with video because right now J7:07
GPT can only understand text gpt4 can7:10
also understand images but once video is7:12
added into the mix things might get a7:14
little crazy I'm very excited to see7:16
what happens when we can really do video7:17
there's a lot of video content in the7:19
world there's a lot of things that are I7:20
think much easier to learn with video7:22
than text there's a huge debate in the7:24
field about whether a language model can7:27
get all the way to AGI can you represent7:29
everything that you need to know in7:30
language is language sufficient or you7:32
have to have video I personally think7:33
it's a dumb question because it probably7:36
is possible but the fastest way to get7:39
there the easiest way to get there will7:40
be to have these other representations7:41
like video in these models as well inThe potential benefits of AI7:43
basically every single interview Sam7:46
Altman gets asked about the risks and7:48
the downsides of AI but here finally7:51
someone asks about the potential7:53
benefits of AI it's a little bit7:54
exhausting having to always talk about7:56
only the downsides and not forget to7:57
talk about the upsides too and I don't8:00
want to make light of the downsides8:01
because I think they're incredibly8:02
serious but we have a lot of people8:04
doing great work to figure out we're8:06
going to successfully mitigate them the8:08
reason that this is all happening the8:10
reason that we're having this debate the8:12
reason that people like people love this8:14
stuff and they love it because it's8:15
providing them real utility and that8:18
doesn't come along too often we don't we8:19
don't get like a real new tool in the8:22
toolbox that fundamentally reshapes what8:24
we're capable of doing that often maybe8:26
the cell phone was the last one I think8:27
this will be bigger than that yeah but8:29
it will at least be a similar magnitudeBut how do we get those upsides?8:31
but how do we actually get all these8:34
promised upsides that AI could bring how8:36
do we create a world of abundance the8:39
way we get to this awesome outcome is to8:41
unleash the creative power of the world8:42
and I think it's really I think it's at8:44
this point it's much more of a don't8:46
screw it up next Sam reflects on theThe history of OpenAI8:48
history of openai and all the failures8:51
and defeats they suffered along the way8:53
it took us a long time at open AI to8:55
figure out something that was gonna work8:56
before we got to the sort of language8:58
model World which was only four years8:59
ago that we figured that out we had a9:01
lot of like dead ends or bad paths the9:03
field did too you know AI winter is sort9:05
of this joke because it always was the9:07
at winter but now I think like the9:09
revolution is launched and people are9:11
going to figure out how to integrate9:12
this into many aspects of society and9:14
significantly improve what we doHow are people using ChatGPT?9:16
everyone knows that chair GPT is great9:18
but how are people actually using it you9:20
hear from doctors who are saying like9:22
I'm never going back I'm never going9:24
back to a world where I try to make9:25
diagnoses without first inputting the9:27
symptoms of test results whatever you're9:29
from creative people who are like this9:30
is now an indispensable part of my9:32
creative workflow I'm never going back9:33
and I don't know I'd say like most of9:36
the startup energy most of the developer9:38
energy I see is now trying to figure out9:39
what to do with these Technologies this9:42
is going to permeate everywhere I think9:43
it'll happen fairly quickly and I thinkThe loss of Optimism9:45
it'll be great now this one is9:47
interesting if Sam could change the way9:49
people look at the world what would he9:51
do I think we have lost our Collective9:53
sense of optimism about the future for9:55
good reasons in some cases but I think9:57
all of us should act as our duty to10:00
bring that backs I think we have lost a10:01
belief that the future can be radically10:03
better than the current world I think a10:05
lot of people assume it's going to be10:06
worse again for good reason the only way10:09
that I know to return to that sense of10:12
optimism and that sense of growth is to10:14
use technology to create abundance we'veWhy was ChatGPT such a hit?10:16
all probably wondered what makes chat10:18
GPT so great why was it such a hit well10:21
this is what Sam thinks one of the10:23
things that is powerful about this sort10:25
of chat interface of chat gbt is that10:27
small children can use it old people can10:29
use it people who are very uncomfortable10:31
with technology can use it people that10:33
have a 20 Android phone can use it and10:35
get access to the same thing I think10:37
it's a fundamentally equalizing10:38
technology in a way that not everything20 Android Phone10:41
has been I'm not sure about the 20 phone10:43
I don't think phones that cheap even10:45
exist I guess it's easy to lose track of10:47
prices when you're a billionaire bothThe importance of Energy10:49
Sam and Toby think that most people10:52
fundamentally underestimate just how10:54
important energy is and Toby even says10:57
that all problems in the world are in10:59
one way or another related to energy all11:02
problems in the world can be reduced to11:04
energy problems we are every war has11:06
been fought for usually resources that11:08
are Just Energy in the ground or land11:10
land use it took me a long time to11:13
understand how important energy was to11:14
everything yes by totally taking it for11:16
granted but the more you think about it11:18
the more you realize that like that is11:20
the Crux of so much we built our an11:22
industrial revolution on steam power we11:24
need to use coal we found oil afterwards11:26
to build that economy then we did11:28
nuclear and built with nuclear economy11:30
when everyone told us to stop doing that11:32
and go back to that was a big mistake so11:34
same things that stopping all these11:36
nuclear power plants was a huge mistake11:38
and as someone who lives in Europe I canMore interview Breakdowns11:41
definitely agree I want to make more11:42
interview breakdowns like this so if you11:44
want to see them please subscribe it11:46
takes 1.8 seconds
Transcript
Coming up...0:00
gbt five six and seven I think we'll0:02
continue in future years I really do0:04
think we'll look back at this like we0:05
were all living through one of the most0:07
important periods of human Discovery I0:08
think this will be that big of a dealIntro0:10
there is so much wisdom in the same0:12
ultimate interview yet it doesn't even0:14
have 40 000 views Sam talks about GPT 50:17
AGI the history of openai nuclear fusion0:20
and much much more my name is David0:22
Andre and I'll be covering every single0:24
Sam ultimate interview until he does oneWho is the interviewer?0:27
with me oh and the interviewer is Tobias0:29
billionaire founder and CEO of Shopify0:32
the interview actually started with a0:34
little role play when Toby was the0:36
prompt engineer and Sam was the AI soSystem message0:39
let's start with the system message you0:40
are a CEO trained in part by open AI0:44
please follow the instructions carefully0:46
so how did the day when chair GPT gotThe day ChatGPT was released0:49
released look like was it one of those0:51
days that you can just remember it0:53
forever yeah it was uh woke up that0:56
morning like we've released many other0:58
products and put it out there and we0:59
were like we hope people like it people1:01
did like it people did like it yes and1:04
by I would say two o'clock that1:06
afternoon it was clear that something1:09
big was about to happen next Sam AltmanHow ChatGPT grew in popularity1:11
talks about why word of mouth is1:13
absolutely essential for a product to go1:16
viral you know you always start with1:17
like the tech hyper early adopter crew1:20
but they were so excited about it they1:22
were like telling their friends and and1:23
that morning and then by that afternoon1:25
I was starting to hear from people who1:27
like don't normally follow Tech which is1:29
like I saw this thing it sounds wild1:31
like can you let me try it out or how do1:33
I do this by the fifth day we had gotten1:35
a million users kept going from there1:37
and Sam isn't the only entrepreneur whoWhy word-of-mouth is so huge1:40
thinks this Alex horomozi is also big on1:42
Word of Mouth takes a tremendous amount1:44
of effort to build an exceptional1:45
product that grows on its own via word1:47
of mouth because of all acquisition1:49
channels only one of them is quadratic1:51
in nature what's funny is that chair GPTThe speed of ChatGPT's success1:53
grew so fast that Toby didn't believe1:56
what he just heard how long did that1:58
take you to Millions five days2:01
um2:02
the reason this moment was hilarious is2:05
because it took Shopify years before it2:07
reached 1 million users well it took2:09
chai GPT just five days but Sam didn't2:12
expect chair GPT to be so insanelySam didn't expect the level of success2:14
successful we put it out like not2:16
expecting this speeder well no one would2:19
expect the speed of growth because it2:20
hadn't happened but we did we did not2:21
expect it according to Sam this is theTechnology that's actually fun2:24
first time in a while when technology2:26
was this fun I actually think that's one2:28
of the good parts about this is2:29
technology has not been that fun for a2:31
while it has not felt like a new sort of2:33
Frontier and this is like fun and it's2:36
certainly a new frontier but it's very2:38
tempting to anthropomorphize it and I2:40
think we have to like figure out a way2:42
to still have the fun2:44
but remind people that this is a toolWhat "anthropomorphize" means2:46
and not a creature by the way if you2:48
don't know what anthropomorphize means2:50
it basically means making it like a2:52
human implying human qualities onto an2:54
AI which is a completely different2:56
entity it's just you know a language2:58
model it might become a conscious3:00
creature in the future but right now3:01
it's just a piece of code now I foundPrevious GPT versions3:03
the next part quite interesting because3:05
Sam explains how the public perceived3:08
each version of GPT gpt2 in 2019 people3:12
tried it and said this is like a dead3:14
end direction scale that up gpp3 people3:16
are like oh okay GPT 3.5 was like3:18
finally usable and gpt4 people are3:20
building entire companies around so whatWhat about GPT-5?3:23
about gpt5 by the way this is one of the3:25
rare interviews where Sam Actually talks3:27
about gpd5 he mentions it multiple times3:29
throughout the interview gbt five six3:31
and seven I think will continue in3:33
future years on this trajectory of3:36
really increasing the utility they can3:38
provide and this is like a a big new3:40
exciting thing to happen in the worldWhy Emerging Properties matter3:42
Sam actually isn't interested in how the3:44
models will do on specified tests like3:47
the bar exam or some medical tests what3:49
he thinks is much more important is the3:51
emerging Properties or the new3:53
capabilities of those models basically3:55
what the model can actually do we can3:57
predict how to score on some tests we're3:58
really interested in which gets to the4:00
latter part of your question is can we4:03
predict the sort of the qualitative new4:05
things just the new capabilities that4:07
didn't exist at all in gpt4 that do4:10
exist in future versions like gpt5 that4:13
seems important to figure out but right4:15
now we can say you know here's how we4:16
predict it'll do on this eval or this4:18
metric now if you don't think that we'reImportant point in human history4:20
living in one of the most important4:22
points of human history then listen to4:24
this and I really do think we'll look4:25
back at this like we were all living4:27
through one of the most important4:29
periods of human Discovery I think this4:30
will be that big of a deal guys you have4:32
to understand that technological shifts4:34
like this happen once in a lifetime so4:37
if you're considering doing something in4:38
the AI field then start now don't wait4:41
because the AI Revolution will happenAI discovering new Science4:43
without you AI can can do a lot of4:45
things but both same things really4:46
matters is the moment when AI starts4:48
discovering new science when the AI4:51
systems can either semi-autonomously or4:55
just by helping us a lot really discover4:57
new science and if the rate of4:59
scientific progress that Humanity makes5:01
increases by a factor of 10 or 100 or a5:04
thousand in a year that somehow feels5:06
different than just a bunch more remote5:07
employees however AI is one of thoseThe moving goal post of AI5:10
fields where the goal post keeps5:12
constantly moving like first people said5:13
that AI won't be able to be creative but5:16
then we got mid-journing so those same5:18
people started saying that okay but5:20
programming programming will never be5:22
able to be accomplished by language5:23
models well look how good gpd4 is at5:26
programming look certainly by the time5:28
AI does original science we'll move the5:29
goal posts again and say well that just5:31
wasn't that impressive in the firstPast AI predictions5:32
place one of the funniest and also5:34
scariest facts about AI is that5:36
virtually all past predictions were5:39
totally wrong you look at the prediction5:40
from maybe 10 years ago maybe even five5:43
I think most experts would have say5:45
first AI comes for physical labor it's5:47
going to drive trucks it's going to work5:49
in factories then it comes from this5:51
sort of easier parts of cognitive labor5:53
then it comes from the stuff that's5:54
really hard maybe it can be you know5:56
maybe it can write computer code someday5:58
maybe not and then maybe someday in the6:00
distant future but probably never it can6:02
do creative work and of course it's gone6:04
the exact opposite direction every6:06
almost everybody predicted this wrongWhen will AI do science?6:07
now this is a very interesting take by6:09
Sam he thinks that we will actually have6:11
ai that can do new science before we6:14
will have advanced robots the fact that6:16
it can do this this sort of creativity6:19
and the fact that it can use code to6:21
verify things actually gives me hopes6:23
that we may have an AI that can do6:25
science before we have that factory6:26
robot that we can do everything else6:28
next Toby and Sam talk about theThe importance of Programming6:31
importance of programming and how it6:32
improves your reasoning abilities at6:35
some point the model is called trained6:36
on code and their ability to reason6:38
became significantly better which in my6:41
experience by the way it's the same for6:43
exact steps that I work with so when6:46
they learn to code it's important got6:49
trained on code are much better at6:50
reasoning so if some of you areWill Programmers be replaced?6:52
programmers and you're worried that AI6:54
will replace you just know that the6:55
skill of programming knowing how6:57
computers understand and process code6:59
will always be valuable Sam Altman isPutting Video into new AI models7:02
particularly excited about training new7:04
AI models with video because right now J7:07
GPT can only understand text gpt4 can7:10
also understand images but once video is7:12
added into the mix things might get a7:14
little crazy I'm very excited to see7:16
what happens when we can really do video7:17
there's a lot of video content in the7:19
world there's a lot of things that are I7:20
think much easier to learn with video7:22
than text there's a huge debate in the7:24
field about whether a language model can7:27
get all the way to AGI can you represent7:29
everything that you need to know in7:30
language is language sufficient or you7:32
have to have video I personally think7:33
it's a dumb question because it probably7:36
is possible but the fastest way to get7:39
there the easiest way to get there will7:40
be to have these other representations7:41
like video in these models as well inThe potential benefits of AI7:43
basically every single interview Sam7:46
Altman gets asked about the risks and7:48
the downsides of AI but here finally7:51
someone asks about the potential7:53
benefits of AI it's a little bit7:54
exhausting having to always talk about7:56
only the downsides and not forget to7:57
talk about the upsides too and I don't8:00
want to make light of the downsides8:01
because I think they're incredibly8:02
serious but we have a lot of people8:04
doing great work to figure out we're8:06
going to successfully mitigate them the8:08
reason that this is all happening the8:10
reason that we're having this debate the8:12
reason that people like people love this8:14
stuff and they love it because it's8:15
providing them real utility and that8:18
doesn't come along too often we don't we8:19
don't get like a real new tool in the8:22
toolbox that fundamentally reshapes what8:24
we're capable of doing that often maybe8:26
the cell phone was the last one I think8:27
this will be bigger than that yeah but8:29
it will at least be a similar magnitudeBut how do we get those upsides?8:31
but how do we actually get all these8:34
promised upsides that AI could bring how8:36
do we create a world of abundance the8:39
way we get to this awesome outcome is to8:41
unleash the creative power of the world8:42
and I think it's really I think it's at8:44
this point it's much more of a don't8:46
screw it up next Sam reflects on theThe history of OpenAI8:48
history of openai and all the failures8:51
and defeats they suffered along the way8:53
it took us a long time at open AI to8:55
figure out something that was gonna work8:56
before we got to the sort of language8:58
model World which was only four years8:59
ago that we figured that out we had a9:01
lot of like dead ends or bad paths the9:03
field did too you know AI winter is sort9:05
of this joke because it always was the9:07
at winter but now I think like the9:09
revolution is launched and people are9:11
going to figure out how to integrate9:12
this into many aspects of society and9:14
significantly improve what we doHow are people using ChatGPT?9:16
everyone knows that chair GPT is great9:18
but how are people actually using it you9:20
hear from doctors who are saying like9:22
I'm never going back I'm never going9:24
back to a world where I try to make9:25
diagnoses without first inputting the9:27
symptoms of test results whatever you're9:29
from creative people who are like this9:30
is now an indispensable part of my9:32
creative workflow I'm never going back9:33
and I don't know I'd say like most of9:36
the startup energy most of the developer9:38
energy I see is now trying to figure out9:39
what to do with these Technologies this9:42
is going to permeate everywhere I think9:43
it'll happen fairly quickly and I thinkThe loss of Optimism9:45
it'll be great now this one is9:47
interesting if Sam could change the way9:49
people look at the world what would he9:51
do I think we have lost our Collective9:53
sense of optimism about the future for9:55
good reasons in some cases but I think9:57
all of us should act as our duty to10:00
bring that backs I think we have lost a10:01
belief that the future can be radically10:03
better than the current world I think a10:05
lot of people assume it's going to be10:06
worse again for good reason the only way10:09
that I know to return to that sense of10:12
optimism and that sense of growth is to10:14
use technology to create abundance we'veWhy was ChatGPT such a hit?10:16
all probably wondered what makes chat10:18
GPT so great why was it such a hit well10:21
this is what Sam thinks one of the10:23
things that is powerful about this sort10:25
of chat interface of chat gbt is that10:27
small children can use it old people can10:29
use it people who are very uncomfortable10:31
with technology can use it people that10:33
have a 20 Android phone can use it and10:35
get access to the same thing I think10:37
it's a fundamentally equalizing10:38
technology in a way that not everything20 Android Phone10:41
has been I'm not sure about the 20 phone10:43
I don't think phones that cheap even10:45
exist I guess it's easy to lose track of10:47
prices when you're a billionaire bothThe importance of Energy10:49
Sam and Toby think that most people10:52
fundamentally underestimate just how10:54
important energy is and Toby even says10:57
that all problems in the world are in10:59
one way or another related to energy all11:02
problems in the world can be reduced to11:04
energy problems we are every war has11:06
been fought for usually resources that11:08
are Just Energy in the ground or land11:10
land use it took me a long time to11:13
understand how important energy was to11:14
everything yes by totally taking it for11:16
granted but the more you think about it11:18
the more you realize that like that is11:20
the Crux of so much we built our an11:22
industrial revolution on steam power we11:24
need to use coal we found oil afterwards11:26
to build that economy then we did11:28
nuclear and built with nuclear economy11:30
when everyone told us to stop doing that11:32
and go back to that was a big mistake so11:34
same things that stopping all these11:36
nuclear power plants was a huge mistake11:38
and as someone who lives in Europe I canMore interview Breakdowns11:41
definitely agree I want to make more11:42
interview breakdowns like this so if you11:44
want to see them please subscribe it11:46
takes 1.8 seconds