February 10, 2025 • 9 mins
OpenAI Head of Policy Chris Lehane discusses DeepSeek concerns and the future of AI growth. He speaks with Bloomberg's Tom Mackenzie.
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:01):
We're going to now head over to the AI Action
Summit in Paris, where Bloomberg's Tom mackenzie is sitting down
with open AI's chief Global affairs officer, Chris Lahane.
Speaker 2 (00:12):
Guys, thank you very much doing yes at Le Grand
Ballet in Paris. I'll start with that question about deep
sea because it is a key question for people attending
this event. Chris Lhane, head of Global policy of course
at open AI deep Sea, how do you characterise the
impact of deep sea. We're just hearing from Demisisarvice, who
I sat down with earlier. He said their claims around
the way they built and trained this model, the cost
(00:35):
and the amount of chips they were using. He says,
that's exaggerated.
Speaker 1 (00:38):
So I think there can be two things that they
are true here. One that they have built a really
impressive model. It basically competes with what open AI had
put out back in September. Now we've since put out
other more advanced models, but clearly a very capable model.
I think the second thing with Demis said can also
(00:58):
be true, which is perhaps and we've seen news reports
on this, that the costs, that how the technology was derived,
that you know, whether they access certain types of chips,
you know, whether what was initially said was not actually
the case. I do think the big takeaway though, even
if everything I just said turns out to be accurate
(01:19):
and crewe, it is still a very impressive and competitive model.
And so to me, the big, big, big takeaway in
all of this is that this really reaffirmed something that
open ai has been saying since the summer of twenty
twenty four that there are two countries in the world
that can build AI at scale, the US, the CCP
(01:40):
led China. And what that really means is that there
is a global competition right now between whether the world's
going to be built on small, d democratic AI rails
or authoritarian, autocratic AI rails. And that's the big takeaway
for this.
Speaker 2 (01:56):
I want to get to that, but I want to
push you on what you'll see with discovery because I
know this is investigation that has been a problem within
open ai as to whether or not deepseek inappropriately use
some of the inferencing data distilling from your own models.
Have you come to a conclusional We're.
Speaker 1 (02:09):
Still looking at it. Obviously, there's we and we've already
made public that we had seen some evidence of that
taking place, And just so folks understand, because distillation is
not a normal word, at least certainly that I don't
use at the breakfast table all the time. There is
different kinds of distillation. And I'll use an analogy. If
you go to the library and you take out a
(02:29):
book and you learn from that book, and that ultimately
informs some of your work. That's fine. That takes place
a lot of that. That's part of what happens in
the AI space. There's another version of it where you
go in to the library, take the book, keep the book,
put your name on the book, slap a cover on
the book, and hand it out as if it's your book.
And that's the replication. And I think that's what we're
concerned about, and again what we've seen some evidence of
(02:51):
and are continuing to review to have a better understand
We've talked with government officials about it, and we'll share
more as we learn more.
Speaker 2 (02:58):
Well open AIS critics have said, then trend your models
on data, and you haven't been fully transparent in terms
of the use of that data. I don't know you
would push back.
Speaker 1 (03:05):
That's when there's good calories and bad calories there's a
good distallation and problematic distillation.
Speaker 2 (03:10):
Have any of your customers pushed back and said, look,
we're uncomfortable with the pricing of your models at this
at this junct shot.
Speaker 1 (03:17):
Well, let me put it this way. You know, open
ai came out in November of two years ago, twenty
twenty two, right, and within two months was it a
one hundred million users? Right? Were well over three hundred
million today continuing to see you know, that really strong growth.
(03:38):
And what we try to do is have different models
priced at different levels depending on how you ultimately want
to use it. And that's something just given the nature
of how fast the technology is moving and the pace
this moving at that you're always constantly trying to evaluate.
Speaker 2 (03:54):
Prices go down from it.
Speaker 1 (03:56):
Oh yeah, So I think amongst the most interesting things
that we have seen this is a an interesting aspect,
which is the efficiencies of systems, particularly using the reasoning
technology is coming down. Sam, When our CEO put a
blog out last night, maybe technically this morning, I don't know,
but at some point in the last twenty four hours
that sort of looked at three observations of AI. The
(04:18):
first was that the more you spend on compute to
build frontier models. It's pretty logrhithmic. The more powerful of
those gets so you're going to require more and more compute,
more and more investment in infrastructure. Secondly, over the last
year or so, we've seen a big increase in efficiencies
which are bringing down the cost of a token. Of
that I think of a token as a price of
(04:38):
computer as a unit of computing about one hundred and
fifty percent. But even as the costs come down, the
amount of people using it go up, so that then
puts pressures back on computing. Like an analogy here is
car prices come down, more people drive cars, but you
then need more energy, more roads. Third, piecet that has
come out and his observation, sorry, just bear with me
for a second, is that is that the economic productivity
(05:01):
that you're getting in is super exponential.
Speaker 2 (05:03):
Where are you getting the money to spend one hundred
billion initially on star gates, elon must says, you probably
don't have the money, even something Adela says, we're good
for our eight two billion. That's Onion.
Speaker 1 (05:13):
So first of all, we have incredible partners. We have
soft Bank, which is a proven track record of raising
enormous syndicated money from sovereigns and pensions.
Speaker 2 (05:23):
Thought is coming from nails.
Speaker 1 (05:24):
And then and then we have Oracle, right, which actually
builds these. Then you have open aies. What are piece
of this? First of all, in terms of the media question,
there's one hundred billion that's going to be going out
the door in the immediate future. We already have a
facility in Abilene. You guys need to come down. You
can abline Texas. Sorry, we'd love to give you, guys.
You can hang out the Oracle. Guys have done an
(05:45):
awesome job down there. No, no, no, no, why don't
you come see it? Show? We like to show, not tell.
And then and then the open AI piece comes in
two different pieces here, and I think this is to
understand the economic model here. First of all, what we
bring to this is the IP and you can think
(06:06):
of compute the same way. Maybe you can think of
gas regular gas, medium gas, and then premium gas. The
premium gas is what people are gonna pay for. The
premium compute is going to be the most expensive compute
in the world because it's gonna be the highest level compute.
You only get that premium compute with our IP. IP
(06:26):
going too, the chip design IP going into the data centers,
IP going how the clusters are sort of structured. And
then we also are a buyer of the compute, right,
so we commit to buying a certain amount of the
compute that's coming out, which helps the whole economic model work.
Speaker 2 (06:42):
Chris, you worked in the Clinton White House. Is the
Trump administration there go go to great headline is to
the Trump administration right to slash AI regulation.
Speaker 1 (06:52):
I think what the Trump administration is focused on is
one thing, and one thing very clearly, who is going
to prevail in the competition between democratic That's what they
get up and think about every day, at least from
what I have seen and from what I have heard,
And so I think they understand that you do have
to really be leading in leaning into the innovation. If
you think about about where the sort of comparative advantages are.
(07:15):
Right at the end of the day, this is actually
pretty simple. Whoever has access to compute is going to
be in a strong position. What makes up compute, it's data,
it's energy, it's chips, and it's talent. Right, and if
you think of what the PRC has, they have an
enormous amount of data. Authoritarian State Energy ten nuclear facilities
(07:37):
last year, another ten coming on this year. Our chips
are better. They're throwing a ton of money at it talent,
and I think the talent piece is really interesting because
they do have great talent in China. But in capitalist systems,
it is capitalism that unleashes the developer, the builder, the entrepreneur,
(07:58):
the people who are actually making this stuff in the
tools that starve hauled problems. And that's where our advantage is.
Speaker 2 (08:03):
Is that why you've joined up and partnered with Aurae'm
to thinking of the AI principles a Google. They've adjusted
them and Google is no longer ruling out building tech
for defense for the military. Yeah, you're signing up, you
partner with Angurill. How far does that relationship go? How
much are you prepared to embed tech into weapons system?
Speaker 1 (08:19):
Yeah? And we also just announced a partnership with the
National Labs, which are the you know, which play an
incredibly important role in how the US government thinks about
national security. Is on the FILS, Yeah, yes, and the
labs you know obviously like Los Alamos and others, are
very big players in the broader national security ecosystem. So
(08:39):
you know, for US, right, we do want to be
a partner on innovation with the government. We do believe
it's incredibly important that democratic AI prevails, and that means
making sure that the government is getting access to the
highest capabilities. You know, we'll certainly do it consistent with
our values and our principles, but at the end of
the day, like we do believe it is very consistent
with their mission that is to make sure AI benefits everyone.
(09:02):
That you're ensuring that AI is going to be built
in a democratic way with democratic values.
Speaker 2 (09:06):
Chris la Haye, thank you very much. Indeed, Global head
of Policy at Open AI.
We're going to now head over to the AI Action
Summit in Paris, where Bloomberg's Tom mackenzie is sitting down
with open AI's chief Global affairs officer, Chris Lahane.
Speaker 2 (00:12):
Guys, thank you very much doing yes at Le Grand
Ballet in Paris. I'll start with that question about deep
sea because it is a key question for people attending
this event. Chris Lhane, head of Global policy of course
at open AI deep Sea, how do you characterise the
impact of deep sea. We're just hearing from Demisisarvice, who
I sat down with earlier. He said their claims around
the way they built and trained this model, the cost
(00:35):
and the amount of chips they were using. He says,
that's exaggerated.
Speaker 1 (00:38):
So I think there can be two things that they
are true here. One that they have built a really
impressive model. It basically competes with what open AI had
put out back in September. Now we've since put out
other more advanced models, but clearly a very capable model.
I think the second thing with Demis said can also
(00:58):
be true, which is perhaps and we've seen news reports
on this, that the costs, that how the technology was derived,
that you know, whether they access certain types of chips,
you know, whether what was initially said was not actually
the case. I do think the big takeaway though, even
if everything I just said turns out to be accurate
(01:19):
and crewe, it is still a very impressive and competitive model.
And so to me, the big, big, big takeaway in
all of this is that this really reaffirmed something that
open ai has been saying since the summer of twenty
twenty four that there are two countries in the world
that can build AI at scale, the US, the CCP
(01:40):
led China. And what that really means is that there
is a global competition right now between whether the world's
going to be built on small, d democratic AI rails
or authoritarian, autocratic AI rails. And that's the big takeaway
for this.
Speaker 2 (01:56):
I want to get to that, but I want to
push you on what you'll see with discovery because I
know this is investigation that has been a problem within
open ai as to whether or not deepseek inappropriately use
some of the inferencing data distilling from your own models.
Have you come to a conclusional We're.
Speaker 1 (02:09):
Still looking at it. Obviously, there's we and we've already
made public that we had seen some evidence of that
taking place, And just so folks understand, because distillation is
not a normal word, at least certainly that I don't
use at the breakfast table all the time. There is
different kinds of distillation. And I'll use an analogy. If
you go to the library and you take out a
(02:29):
book and you learn from that book, and that ultimately
informs some of your work. That's fine. That takes place
a lot of that. That's part of what happens in
the AI space. There's another version of it where you
go in to the library, take the book, keep the book,
put your name on the book, slap a cover on
the book, and hand it out as if it's your book.
And that's the replication. And I think that's what we're
concerned about, and again what we've seen some evidence of
(02:51):
and are continuing to review to have a better understand
We've talked with government officials about it, and we'll share
more as we learn more.
Speaker 2 (02:58):
Well open AIS critics have said, then trend your models
on data, and you haven't been fully transparent in terms
of the use of that data. I don't know you
would push back.
Speaker 1 (03:05):
That's when there's good calories and bad calories there's a
good distallation and problematic distillation.
Speaker 2 (03:10):
Have any of your customers pushed back and said, look,
we're uncomfortable with the pricing of your models at this
at this junct shot.
Speaker 1 (03:17):
Well, let me put it this way. You know, open
ai came out in November of two years ago, twenty
twenty two, right, and within two months was it a
one hundred million users? Right? Were well over three hundred
million today continuing to see you know, that really strong growth.
(03:38):
And what we try to do is have different models
priced at different levels depending on how you ultimately want
to use it. And that's something just given the nature
of how fast the technology is moving and the pace
this moving at that you're always constantly trying to evaluate.
Speaker 2 (03:54):
Prices go down from it.
Speaker 1 (03:56):
Oh yeah, So I think amongst the most interesting things
that we have seen this is a an interesting aspect,
which is the efficiencies of systems, particularly using the reasoning
technology is coming down. Sam, When our CEO put a
blog out last night, maybe technically this morning, I don't know,
but at some point in the last twenty four hours
that sort of looked at three observations of AI. The
(04:18):
first was that the more you spend on compute to
build frontier models. It's pretty logrhithmic. The more powerful of
those gets so you're going to require more and more compute,
more and more investment in infrastructure. Secondly, over the last
year or so, we've seen a big increase in efficiencies
which are bringing down the cost of a token. Of
that I think of a token as a price of
(04:38):
computer as a unit of computing about one hundred and
fifty percent. But even as the costs come down, the
amount of people using it go up, so that then
puts pressures back on computing. Like an analogy here is
car prices come down, more people drive cars, but you
then need more energy, more roads. Third, piecet that has
come out and his observation, sorry, just bear with me
for a second, is that is that the economic productivity
(05:01):
that you're getting in is super exponential.
Speaker 2 (05:03):
Where are you getting the money to spend one hundred
billion initially on star gates, elon must says, you probably
don't have the money, even something Adela says, we're good
for our eight two billion. That's Onion.
Speaker 1 (05:13):
So first of all, we have incredible partners. We have
soft Bank, which is a proven track record of raising
enormous syndicated money from sovereigns and pensions.
Speaker 2 (05:23):
Thought is coming from nails.
Speaker 1 (05:24):
And then and then we have Oracle, right, which actually
builds these. Then you have open aies. What are piece
of this? First of all, in terms of the media question,
there's one hundred billion that's going to be going out
the door in the immediate future. We already have a
facility in Abilene. You guys need to come down. You
can abline Texas. Sorry, we'd love to give you, guys.
You can hang out the Oracle. Guys have done an
(05:45):
awesome job down there. No, no, no, no, why don't
you come see it? Show? We like to show, not tell.
And then and then the open AI piece comes in
two different pieces here, and I think this is to
understand the economic model here. First of all, what we
bring to this is the IP and you can think
(06:06):
of compute the same way. Maybe you can think of
gas regular gas, medium gas, and then premium gas. The
premium gas is what people are gonna pay for. The
premium compute is going to be the most expensive compute
in the world because it's gonna be the highest level compute.
You only get that premium compute with our IP. IP
(06:26):
going too, the chip design IP going into the data centers,
IP going how the clusters are sort of structured. And
then we also are a buyer of the compute, right,
so we commit to buying a certain amount of the
compute that's coming out, which helps the whole economic model work.
Speaker 2 (06:42):
Chris, you worked in the Clinton White House. Is the
Trump administration there go go to great headline is to
the Trump administration right to slash AI regulation.
Speaker 1 (06:52):
I think what the Trump administration is focused on is
one thing, and one thing very clearly, who is going
to prevail in the competition between democratic That's what they
get up and think about every day, at least from
what I have seen and from what I have heard,
And so I think they understand that you do have
to really be leading in leaning into the innovation. If
you think about about where the sort of comparative advantages are.
(07:15):
Right at the end of the day, this is actually
pretty simple. Whoever has access to compute is going to
be in a strong position. What makes up compute, it's data,
it's energy, it's chips, and it's talent. Right, and if
you think of what the PRC has, they have an
enormous amount of data. Authoritarian State Energy ten nuclear facilities
(07:37):
last year, another ten coming on this year. Our chips
are better. They're throwing a ton of money at it talent,
and I think the talent piece is really interesting because
they do have great talent in China. But in capitalist systems,
it is capitalism that unleashes the developer, the builder, the entrepreneur,
(07:58):
the people who are actually making this stuff in the
tools that starve hauled problems. And that's where our advantage is.
Speaker 2 (08:03):
Is that why you've joined up and partnered with Aurae'm
to thinking of the AI principles a Google. They've adjusted
them and Google is no longer ruling out building tech
for defense for the military. Yeah, you're signing up, you
partner with Angurill. How far does that relationship go? How
much are you prepared to embed tech into weapons system?
Speaker 1 (08:19):
Yeah? And we also just announced a partnership with the
National Labs, which are the you know, which play an
incredibly important role in how the US government thinks about
national security. Is on the FILS, Yeah, yes, and the
labs you know obviously like Los Alamos and others, are
very big players in the broader national security ecosystem. So
(08:39):
you know, for US, right, we do want to be
a partner on innovation with the government. We do believe
it's incredibly important that democratic AI prevails, and that means
making sure that the government is getting access to the
highest capabilities. You know, we'll certainly do it consistent with
our values and our principles, but at the end of
the day, like we do believe it is very consistent
with their mission that is to make sure AI benefits everyone.
(09:02):
That you're ensuring that AI is going to be built
in a democratic way with democratic values.
Speaker 2 (09:06):
Chris la Haye, thank you very much. Indeed, Global head
of Policy at Open AI.
Popular Podcasts
Bookmarked by Reese's Book Club
Welcome to Bookmarked by Reese’s Book Club — the podcast where great stories, bold women, and irresistible conversations collide! Hosted by award-winning journalist Danielle Robay, each week new episodes balance thoughtful literary insight with the fervor of buzzy book trends, pop culture and more. Bookmarked brings together celebrities, tastemakers, influencers and authors from Reese's Book Club and beyond to share stories that transcend the page. Pull up a chair. You’re not just listening — you’re part of the conversation.
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com