[00:06] Hello and welcome to the program. The AI
[00:10] revolution runs on electricity. Lots of
[00:13] it. Vast amounts of it. The data centers
[00:16] being built right now to power the next
[00:18] generation of AI consume as much energy
[00:21] as a small city. Last year, the world's
[00:25] biggest tech companies spent more than
[00:27] $400 billion building them. And they
[00:30] need more of them, many more of them.
[00:32] But there's a problem. In the United
[00:34] States, four in 10 of the data centers
[00:37] that are being planned for this year are
[00:39] at serious risk of delay. Not enough
[00:42] power, not enough equipment, and not
[00:44] enough people to build them. Why? Also
[00:48] on the program this week, Coachella, the
[00:51] world's most glamorous music festival.
[00:53] But some of the online influences you
[00:55] may have seen pictured in the California
[00:57] desert alongside the biggest stars were
[01:00] fake, not real. They don't exist. AI
[01:03] generated to promote brands and to make
[01:05] money. Plus, we'll also talk this week
[01:08] about the late Val Kilmer appearing in a
[01:11] new film one year after his death. with
[01:14] us this week to talk about it, Palmy
[01:16] Olsen, uh who is a technology columnist
[01:19] at Bloomberg. Welcome to you.
[01:20] >> Thank you.
[01:21] >> Also here, Dr. Sasha Luchoni, a computer
[01:24] scientist specializing in AI and its
[01:26] environmental impact and also in the
[01:29] studio with us. Uh good to have her back
[01:31] here. Dr. Stephanie Hair, colleague, uh
[01:34] author and AI expert to give you your
[01:36] full title. Um Pommy, let us start with
[01:39] this issue of data centers. Um we're
[01:41] building lots of them. They are powering
[01:44] the AI revolution.
[01:46] Why are so many of them on hold?
[01:48] >> Yeah. So this is really an issue of
[01:51] bottlenecks and you sort of alluded to
[01:52] it in your introduction. There's the
[01:54] issue of power. So getting access to
[01:57] actual power on electricity grids that
[01:59] are very very old and haven't received a
[02:01] lot of investment over many decades in
[02:03] the United States. Um also equipment,
[02:06] getting access to things like
[02:07] transformers or switch gear, which are
[02:09] the types of equipments you need to
[02:11] actually run data centers. There's also
[02:13] a huge bottleneck getting them. It could
[02:14] take up to five years to get some of
[02:16] that equipment, particularly because a
[02:18] lot of it comes from China and recent US
[02:20] tariffs on Chinese goods has made that
[02:22] even more difficult. And then there's
[02:24] also just getting the people, the
[02:25] talent, um the electricians um and the
[02:28] people with the skill set who can
[02:30] actually uh construct and run these data
[02:33] centers. Um so all those things combined
[02:36] have meant that at a time when there is
[02:38] this rapacious demand for um for energy
[02:42] from tech companies, it's actually very
[02:44] difficult to build them fast enough.
[02:47] >> Does that problem become more acute in
[02:49] the energy crisis we're in currently?
[02:51] >> I think it does to some extent and it's
[02:53] more of a problem for the so-called
[02:56] hyperscalers, the big tech companies
[02:57] like Meta, um Alphabet, Amazon, etc. who
[03:03] are actually the ones who have to shell
[03:04] out for these energy costs. And so
[03:07] they're setting up um these kind of mini
[03:10] nuclear reactors that can actually
[03:12] provide um energy specifically for the
[03:15] data center um and use renewable sources
[03:18] instead of gas. But at the moment even
[03:20] getting those up and running is
[03:23] logistically very very difficult. It's
[03:25] time consuming. There aren't any
[03:26] actually that are operational just yet.
[03:29] and and so right now I think the main
[03:31] source of energy is gas.
[03:33] >> Should we give our viewers just a scale
[03:35] of what we're talking about here? Scott
[03:36] Galloway who might be coming on the
[03:38] program next week over in New York, he
[03:40] says Open AI alone, Stephanie will need
[03:44] 20% of current US electricity capacity
[03:47] at $10 trillion.
[03:50] >> That's extraordinary. It's extraordinary
[03:52] for a company that is yet to turn a
[03:54] profit and which is under huge pressure
[03:56] to demonstrate value ahead of an IPO as
[03:59] well. So it's just worth saying that all
[04:02] of these companies, not just OpenAI,
[04:03] were making very big promises about
[04:06] their data structure buildout plans as
[04:08] of last year. We've already seen the US
[04:10] UK technology deal that's on hold. We've
[04:13] seen OpenAI have to pull out of some of
[04:15] its Stargate, the big US data structure
[04:20] plan. and they've had to pull out on
[04:21] some of those things. Maybe we are
[04:23] walking back. So that question of will
[04:25] the AI bubble burst? It might not burst.
[04:27] It might just sort of slightly
[04:29] >> situation, isn't it? When you think that
[04:30] actually if you're going to plug these
[04:32] into the grid and you don't know whether
[04:35] these companies are going to survive in
[04:37] the scale or perhaps they're even bigger
[04:39] than they are right now, it you it's a
[04:42] difficult thing to plan.
[04:44] >> It's a difficult thing to plan when
[04:45] you're not a planned economy. Which is
[04:47] why when we're looking at the data
[04:48] center roll out in a country like China
[04:50] and comparing that to the United States
[04:52] or indeed here in Europe, we get very
[04:54] different pictures.
[04:55] >> Right. Um Dr. Luchi, Sasha, I'm going to
[04:58] call you Sasha. Does it make sense for
[05:00] the US president to be so vehemently
[05:02] opposed to renewable energy given the
[05:05] scale that we're talking?
[05:07] Well, the problem is is that the data
[05:09] centers are being built so quickly that
[05:11] renewable capacity has trouble keeping
[05:13] up, especially in rural areas,
[05:15] especially out outside of places where
[05:18] renewables are are are typically the
[05:19] case. So, I think that currently the
[05:21] emphasis is build faster, build bigger,
[05:24] and they don't want to wait around for
[05:26] solar or wind, which is why um
[05:28] essentially most uh the data centers
[05:30] that are coming online as quickly are
[05:32] essentially bringing in turbines on the
[05:34] back of trucks, natural gas. It's it's
[05:35] like bring your bring your own energy uh
[05:37] essentially and most of that is and they
[05:39] have the money
[05:39] >> is non-renewable.
[05:40] >> I mean these companies
[05:41] >> they have the money
[05:42] >> they have the money but actually
[05:44] currently there's a bottleneck even when
[05:45] you have the money because there's not
[05:46] enough turbines to to power all these
[05:48] data centers because there's there's a
[05:50] backlog nowadays and and even these uh
[05:52] these turbines can't be produced fast
[05:54] enough to to respond to demand. Palmy at
[05:57] Bloomberg recently you highlighted an
[05:59] issue in Northern Spain um with the data
[06:02] center build out there which has
[06:03] actually been held up as a a model for
[06:05] the rest of Europe but for the people
[06:06] who live around these projects the
[06:09] reality is sometimes very different why
[06:12] >> I think it's a common story we're also
[06:14] seeing in the United States a lot of
[06:15] push back from local residents in areas
[06:17] where companies want to build data
[06:19] centers and in northern Spain um there
[06:21] have been uh the situation is that AWS
[06:24] which is the cloud business of Amazon um
[06:27] sent letters to local people um saying
[06:31] we want to buy your land, giving them
[06:33] sometimes 4 days notice to say yes or
[06:35] no. Um and some of these people in in in
[06:38] northern Spain actually thought it was a
[06:40] scam at first. Um one lady went to her
[06:42] local town hall and even they didn't
[06:44] know. So, it's a real um kind of land
[06:47] grab almost to try and get land that is
[06:50] relatively cheap in an area where energy
[06:53] costs are relatively low and that are
[06:56] sparssely populated as well. It seems
[06:58] like an ideal situation for building a
[07:00] data center, but at the same time there
[07:02] is the reality for people who do live
[07:04] there and there are people who live
[07:06] there um that they have to give up that
[07:08] land or maybe the suddenly you've got
[07:10] this eyes sore in a place that you've
[07:12] lived in for many generations. If you're
[07:14] in a community like that though and
[07:15] you've already struggled to to get
[07:18] natural resources or to to to get
[07:20] electricity to get yourselves on the
[07:22] grid, does the arrival of a big AI
[07:24] company help in that process? Perhaps it
[07:26] it could help a community
[07:28] >> in some respects. And the funny part in
[07:31] that is that governments um local
[07:33] governments often uh frame data center
[07:36] buildouts as being great for jobs. Yeah.
[07:38] But I think you're conflating in that
[07:40] situation permanent jobs with
[07:43] construction jobs which are temporary.
[07:45] And so when you build out a data center,
[07:46] you're you're going to hire me.
[07:47] >> They're not are they not necessarily big
[07:49] employers once the kit is there?
[07:50] >> No, I think in a typical data center you
[07:52] might have about 100 people, most of
[07:54] them cleaners and security people. Um
[07:56] whereas for the buildout, sure,
[07:58] hundreds, maybe thousands of people, but
[08:00] then that's only temporary.
[08:01] >> All right, I'm going to bring in an
[08:02] audience question quite early into the
[08:04] program this week because it it it fits
[08:06] what you're talking about. It's from
[08:08] James in the UK. He says, "Sasha, AI
[08:10] companies continue to minimize this
[08:12] environmental impact." He points
[08:15] specifically to Sam Alman's recent claim
[08:17] that AI's water usage is minimal. James
[08:20] says that's simply not true. He also
[08:22] tells us that younger generations are
[08:24] increasingly boycotting generative AI
[08:26] for environmental reasons. So, here's
[08:28] his question. Should mainstream media be
[08:30] doing more to hold these companies to
[08:33] account?
[08:34] >> Definitely. Actually, a recent Guardian
[08:37] study found that uh the big tech
[08:38] companies were lobbying very very hard
[08:40] against transparency to make sure that I
[08:43] mean citing confidentiality to not
[08:45] include any uh energy figures or water
[08:48] figures about data centers and so we
[08:50] we're we're seeing them play dirty and I
[08:52] think it's time to ask for
[08:54] accountability and I think that
[08:55] especially in a time where people are
[08:57] increasingly sustainability conscious so
[08:59] so you know we make our decisions based
[09:00] on on the environment and and ethical
[09:03] concerns we need this information
[09:05] whether it be for choosing one AI model
[09:07] over the other for for using AI or not
[09:09] using AI, right? There's there's lots of
[09:11] decisions that we make on an everyday
[09:12] basis that we just don't have the
[09:14] information for. And especially since AI
[09:16] has become such a common technology, we
[09:18] definitely need these numbers and and
[09:20] these companies have them. It's just a
[09:22] matter of of giving them maybe positive
[09:24] and and and less positive incentives for
[09:26] sharing them.
[09:26] >> Well, well, let's let's try and choose
[09:28] to look at this positively because we're
[09:30] all using the technology. We're going to
[09:32] use it in our work. So we we need these
[09:34] companies to be successful if we're if
[09:36] we're going to employ AI fully. What
[09:39] does a responsible data center look
[09:41] like, Sasha?
[09:43] >> So you can definitely create them in a
[09:45] way that's u more integrated in the
[09:47] existing infrastructure. So currently um
[09:50] the data centers are being built out in
[09:51] a very kind of bigger is better kind of
[09:54] way. So typically they're outside of
[09:55] cities, they're huge like warehouse
[09:57] sized um but they can really be
[09:59] integrated like the smaller data centers
[10:00] can be in basement. Um, the heat can be
[10:03] reused to heat office buildings or
[10:04] university campuses. It's much easier to
[10:06] use renewable energy or or a mix at
[10:08] least of of renewable energy if if
[10:11] there's less capacity that's needed.
[10:13] >> You think it that maybe part of this
[10:14] answer is then partnering with other
[10:16] companies,
[10:18] >> partnering and rethinking the paradigm.
[10:19] So, currently it's like we need the
[10:21] biggest data centers, we need sovereign
[10:23] AI, we need, you know, bigger, let's
[10:25] build it out. Even in like I mean in
[10:26] Canada it's the same thing. We need our
[10:28] own data center. Let's let's build it
[10:29] out. But instead of thinking of that as
[10:31] as the the you know when you have a
[10:33] hammer everything's a nail I think we
[10:34] should be thinking about the nails that
[10:35] we have and thinking about okay so what
[10:37] do we need this data center for? Is it
[10:39] for a university? Is it for a private
[10:41] company? Is there a way of for example
[10:43] incentivizing uh some mix of renewables
[10:45] or or for example helping them build it
[10:47] out in a way that isn't you know bring
[10:49] your own turbine on a on a on a truck
[10:51] kind of situation. And so I think there
[10:53] are ways of being more more agile if we
[10:56] rethink our way of of doing AI. And it's
[10:58] not only for data sensors. Same thing
[10:59] for AI models. Instead of being like we
[11:01] need the biggest, we need the most
[11:02] energy intensive model for every single
[11:04] task. We can have smaller models for
[11:06] example ondevice models. Uh instead of
[11:09] having every query be dispatched to the
[11:11] cloud, we can have AI models running
[11:12] locally on our smartphones and and
[11:14] computers. So I think we should really
[11:16] be rethinking a little bit the way that
[11:17] we design and deploy AI currently.
[11:20] >> A quick question just to um satisfy my
[11:23] curiosity, Sasha. Um quick answers if
[11:26] you could. There are some country uh
[11:28] companies that are developing these air
[11:30] cooling systems to reduce water
[11:32] consumption. Do they work?
[11:35] >> Yes, but it's often a trade-off of using
[11:37] more energy and less water. So often um
[11:40] it's true that you can for example
[11:42] recycle water. So essentially water gets
[11:44] cycled through and it heats up and you
[11:46] have to cool it down. So either you need
[11:48] uh cooling towers or sometimes you know
[11:50] you cool it down with energy with
[11:52] electricity. And so it's often a
[11:53] trade-off where they're using more
[11:55] energy but but less water. It's a closed
[11:57] loop system.
[11:57] >> Yeah. And at the outset you you said
[11:59] that very often these data centers are
[12:01] outstripping what the renewable industry
[12:03] can provide for them. But there are good
[12:06] examples and I wanted to point to them
[12:08] where data centers have been cited very
[12:11] close to renewable energy. So Iceland is
[12:13] using geothermal, Norway using
[12:16] hydroelectric. Is that an example that
[12:18] other countries should be following?
[12:21] >> Yes. But I think that um very few
[12:23] countries I mean uh in in the current
[12:26] state of things have that extra capacity
[12:28] and also if these data centers continue
[12:30] to be so like for example if a data
[12:31] center uses as much energy as 100,000
[12:33] homes um there's very few grids
[12:36] renewable grids that can take that that
[12:38] can provide that energy on such a short
[12:40] notice. Even for example in Quebec where
[12:42] I live we have hydro but we don't have
[12:43] the extra capacity for you know in two
[12:45] years an extra 100 thousand homes to be
[12:47] built. It has to be gradual. And so it's
[12:49] really the timelines that often don't
[12:51] line up. And this is why natural gas is
[12:53] the cheapest, fastest solution. And and
[12:55] often there are long-term plans. Often
[12:57] it's like, well, in 10 years we're going
[12:58] to do renewables. In 10 years, we're
[13:00] going to do this. But in the meantime,
[13:02] it adds a lot of emissions.
[13:04] >> Okay. Well, you might have questions on
[13:05] what you've been hearing about data
[13:07] centers. You might have some strong
[13:08] thoughts on it. AI decoded atbc.co.uk.
[13:12] Now, since Stephanie has been focusing
[13:14] on clarity and regulation, I've got a
[13:15] story for her. Um, let me show you some
[13:17] images. Uh, these are images that look
[13:20] entirely real, but the people in them
[13:23] are fake.
[13:27] >> My Coachella week was so much fun. Let
[13:30] me take you around. It's a secret. You
[13:32] can be in Coachella as me just by a few
[13:34] prompts. Stay with me till the end for
[13:36] the prompts.
[13:42] They are computerenerated influencers
[13:46] uh who were seen photographed on
[13:49] Instagram alongside some of the most
[13:52] famous people at Coachella uh which is
[13:56] those in the know will know is this very
[13:57] trendy music festival in the desert in
[14:00] one of the desert valleys in California.
[14:03] How many of those engaging in these
[14:05] photographs knew that uh the people they
[14:09] were pictured alongside were fake? I
[14:11] would suggest not very many. I'm not
[14:12] even sure that Coachella knew that there
[14:14] were fake influences uh in the crowd.
[14:16] Stephanie, we've talked about this on
[14:18] the program before about AI generated
[14:21] beauty, the impact it has on young
[14:23] people. This for me, actually, I was
[14:25] reading about it this week. This feels
[14:26] like the next chapter of that.
[14:28] >> Yeah. And again, the law is just not fit
[14:30] for purpose on this. I think we're
[14:32] really going to have to get to a point
[14:33] where we have laws on the books that say
[14:36] if you have someone that's pre
[14:37] pretending to be a human being, it has
[14:39] to be labeled. It just has to because
[14:42] you're dealing with children first of
[14:43] all, so like anyone that's under the age
[14:44] of 18 needs to be protected, but you've
[14:47] also dealing with older people. You're
[14:49] also dealing with the potential for
[14:50] scam, for fraud, for misinformation and
[14:53] disinformation. So this would just solve
[14:55] a lot of things.
[14:56] >> Pommy, who's behind these images? What
[14:58] do they what do they want? From what I
[15:00] understand, it is mostly agencies. It's
[15:03] not, you know, it's not a cottage
[15:05] industry of people working from home.
[15:06] There are, you know, agencies most of
[15:08] the time in in Europe, in places like
[15:11] London, um, and on the continent who are
[15:14] producing these as branding exercises
[15:16] and as an opportunity for a brand to get
[15:18] a sponsorship. But for Yeah, absolutely.
[15:20] Um, if you think about it, an influencer
[15:24] um who has a brand sponsorship deal will
[15:27] be quite costly because if they want to
[15:28] go to Coachella, they want to go
[15:30] business class maybe. Um, they want to
[15:33] get a hotel, they want to get some other
[15:34] freebies. But if you have an influencer
[15:36] who you are sponsoring to hold your can
[15:39] of whatever in the photograph, um,
[15:42] they're not going to have a bad day or
[15:44] get old or look weird in the photo.
[15:46] They're always going to look great. It's
[15:48] funny, ahead of this program, I actually
[15:50] looked at some of these influencers who
[15:51] were in Coachella in Coachella and it
[15:55] was amazing like one of them had about
[15:57] 170,000 followers and had it was
[15:59] pictures of her with Justin Bieber um
[16:02] with the Kardashians with Madonna and no
[16:05] one in the comments was saying this
[16:07] isn't real. They were all the comments
[16:09] were kind of congratulatory
[16:12] um and there was no disclosure at all on
[16:14] the Instagram profile that it was AI
[16:16] generated. So I think a lot of people in
[16:18] good faith would look at it and think
[16:20] this is really
[16:20] >> the obvious problem is Sasha that the
[16:22] very famous person who's gone to
[16:24] Coachella can say to someone who might
[16:27] be advertising kryptonite next to them
[16:29] look I don't want to be advertising
[16:30] kryptonite and they can push them away.
[16:32] They have no choice. They have no say in
[16:34] in an AI generated person being put next
[16:37] to them in a photograph that they pose
[16:39] for unknowingly.
[16:41] >> Yeah. In a in a world of AI agents,
[16:43] humans lose their own agency. I think to
[16:46] some extent and especially famous people
[16:48] because there's so many likenesses of
[16:49] them on the internet that it's very very
[16:51] easy to generate a false image or video
[16:53] now of of a celebrity.
[16:55] >> Didn't Stephanie, didn't we talk about
[16:57] New York bringing in new regulation to
[17:00] stop this? I think you had to put on
[17:02] your website whether you were you were
[17:04] using an AI generated influencer, but
[17:07] there's I mean some of these pictures
[17:08] from Coachella do do that, but plenty of
[17:10] them don't.
[17:11] >> Yeah. And that's the enforcement thing.
[17:13] Like there are all sorts of laws that
[17:14] are obvious
[17:15] >> from state to state is different. Right.
[17:16] >> Exactly. And how you know whose job is
[17:19] it to police that and how are they able
[17:22] to get the accountability that they
[17:24] need? So again this is a case of if you
[17:25] were to take them to court that's going
[17:27] to take years right it's going to cost a
[17:29] lot of money etc. So it's kind of like
[17:31] everything that we saw about
[17:32] accountability with social media not
[17:34] being very effective.
[17:35] >> I mean Coachella themselves could just
[17:37] say enough. You can't do this. It's up
[17:39] to the organizer.
[17:40] >> They absolutely could. And I think, you
[17:42] know, what you mentioned earlier about
[17:43] the uh the celebrities, you might want
[17:45] to just play the world's tiniest violin
[17:47] for these people for being in these
[17:49] photographs and potentially compromising
[17:51] situations. But I mean, ultimately, if
[17:53] they do get upset and if the brands get
[17:55] upset, I think that's perhaps going to
[17:57] be potentially even more effective than
[17:59] actual regulation in um in getting
[18:03] enforcement. So,
[18:04] >> what about the platforms though? I mean,
[18:05] in Instagram, Tik Tok, they're profiting
[18:08] from these engagements.
[18:09] >> Yes. And um technically on these
[18:11] platforms you're you are supposed to
[18:13] disclose if something is AI generated.
[18:15] The fact is nobody actually follows that
[18:18] rule. And Meta does have um automated
[18:22] systems that will try and look for
[18:24] things that are AI generated and tag
[18:26] them. But it's an almost impossible task
[18:28] because there's hundreds of thousands of
[18:30] posts made every day and many many many
[18:33] are slipping through the net. Now the
[18:35] thing is it is possible if they really
[18:37] wanted to. There are ways to put um a
[18:41] cryptographic signature on actual
[18:43] photographic images um called CP2A. Um
[18:48] but that's just not something that the
[18:50] tech companies are investing in because
[18:52] if you think about it, there is a
[18:54] commercial incentive to just let this
[18:56] carry on because to to your point
[18:57] earlier, do people actually like this?
[18:59] Weirdly, the public don't hate AI
[19:02] avatars. They're kind of okay with it
[19:04] >> if they know. I mean, how many people
[19:06] really know? because some of these are
[19:07] really good and they're getting better.
[19:09] >> I think that's just going to make it
[19:10] harder to to deal with.
[19:11] >> So weird. Everyone talks about being
[19:13] authentic and brands are all about your
[19:15] values and then they do this stuff
[19:17] that's so fake and people eat it up.
[19:19] >> Weird.
[19:20] >> Do we still call California the Wild
[19:21] West? No, maybe not.
[19:23] >> But this is the Wild West.
[19:25] >> Yeah, absolutely.
[19:26] >> Now, the late Val Kilmer was one of the
[19:28] greats. Do we agree on that?
[19:30] >> Yeah. Top Gun Batman forever.
[19:32] >> Real genius.
[19:33] >> Yeah. The Doors. I liked him in that. uh
[19:35] one of the Hollywood uh greats, one of
[19:37] the versatile actors of Hollywood as
[19:39] well. And he died a year ago, as many of
[19:41] you will know, age 65 after a long and
[19:43] sad battle with throat cancer. But he
[19:46] had been cast in a film uh a few years
[19:48] earlier. It's called As Deep as the
[19:50] Grave. It's a historical drama about uh
[19:53] the American Southwest. And uh of
[19:56] course, he didn't make it to set because
[19:58] he was he was too ill at the end. But
[20:00] this week, the trailer for that movie
[20:03] debuted at Cineacon in Las Vegas. He's
[20:06] in it. And every scene in which he is in
[20:08] and every line that he speaks, of
[20:10] course, is generated entirely by AI.
[20:18] >> Is me.
[20:29] Hey, hey,
[20:32] hey.
[20:55] Hey,
[21:04] >> don't fear the dead and don't fear me.
[21:07] His children gave their blessing to this
[21:09] Stephanie. Um, and just so everybody
[21:12] knows, the filmmakers followed the
[21:14] guidelines, spoke to the unions. Kilmer,
[21:16] in fact, himself uh embraced AI in his
[21:20] final movie, Top Gun Maverick. uh his
[21:22] his voice was recreated by AI, so he
[21:25] wasn't oblivious to this.
[21:28] Is this the blueprint, do you think, for
[21:30] for for AI in Hollywood, an ethical way
[21:33] of using it, or does it for you open a
[21:35] door that we can't close?
[21:37] >> I think it's just about choice. So, I
[21:40] like the idea that if directors and
[21:42] other artists want to experiment with
[21:43] AI, that they are doing so mindfully,
[21:46] that they're trying to come up with an
[21:47] ethical standard that is no doubt going
[21:49] to be discussed and may eventually be
[21:51] formalized. I also think it's really
[21:53] important for any creative person who
[21:55] doesn't want their likeness, their
[21:58] biometrics or their creative output to
[22:00] be used in this way to be able to say
[22:02] no. Right? So in that case for anybody
[22:05] who's a Hollywood actor listening and
[22:07] watching our show as we know they are
[22:09] >> um they would want to be all the way
[22:11] they want to be speaking with their
[22:12] agents and with their team and their
[22:14] lawyers to be really clear about that.
[22:16] So you know how do you want your
[22:17] likeness being used while you're alive
[22:19] and then how do you want it being used
[22:21] after your death. So in this case Bal
[22:23] Kilmer's children were fine with it. His
[22:25] estate's fine with it and everything was
[22:26] done with everyone being I think as
[22:28] ethical as they can be. Other actors
[22:30] have made different choices. C can I
[22:32] just say Stephanie for the record that
[22:34] if you're going to use my likeness for
[22:35] AI decoded into the future I am happy
[22:38] for that so long as you pay the
[22:40] royalties to my estate just on the
[22:42] record so we're all cleared.
[22:44] >> Um Sasha I mean obviously Hollywood
[22:46] Hollywood actors are are a gift aren't
[22:48] they for AI because there's there's
[22:50] hundreds of hours of film of them.
[22:53] They've been in lots of performances. Uh
[22:55] and so in fact I think this performance
[22:58] that Kilmer in of this movie was it was
[23:01] reconstructed from 40 films hundreds of
[23:04] hours of footage. So as long as you
[23:06] signed up to this sky's is the limit.
[23:09] >> I think that's a very hard ethical
[23:10] question. Um I've seen a lot of papers.
[23:13] I even saw a theater play on this topic
[23:15] especially after death actually. I think
[23:16] that's a really important point. Uh who
[23:18] can opt in? How can you opt out if
[23:20] you're dead? Um, and also what does this
[23:22] mean for the community? Because if
[23:24] there's peer pressure, for example, and
[23:25] I think that actually AI was one of the
[23:27] one of the sticking points during the
[23:29] strikes a couple of years ago, right?
[23:30] Uh, to what extent is there is there
[23:32] union pressure? Is there community
[23:33] pressure to opt in? Can you can you
[23:35] continue opting out in this new new
[23:38] world, right? And it's it's similar to
[23:40] what a lot of workers are facing as
[23:41] well. Um, I'm hearing a lot of people
[23:43] being like, well, I'm forced to use AI
[23:44] in my in my workplace. We even have
[23:46] dashboards for tracking it. And so, it's
[23:48] really this pressure that we're seeing
[23:49] to use AI. And I think that that does uh
[23:52] uh make people give up some of their
[23:54] individual choices if they feel
[23:56] pressured to. So for example, if you're
[23:57] a young actor and you want to uh you
[23:59] know make your make a name for yourself,
[24:01] but you don't want to use AI but you
[24:02] have this peer pressure around you, can
[24:04] you really opt out without having a
[24:05] negative impact on your career?
[24:08] >> Yeah. Yeah. And and I think um Hollywood
[24:10] has already this history of recycling
[24:13] old films and making sequels and making
[24:16] remakes. And I think there is already
[24:17] this tendency to want to maximize
[24:19] profits by going back to whatever works.
[24:21] And if that's the incentive driving the
[24:24] remake of an actor who has died, I think
[24:28] in the end that can actually put younger
[24:30] fresh talent out of work if it's just
[24:32] the kind of the same icons appearing
[24:34] over and over again for the next 100
[24:36] years. Um, and to Sasha's point as well,
[24:39] I think where that pressure comes from,
[24:41] like I've spoken to a company that does
[24:43] virtual reality concerts or Avatar
[24:45] concerts, um, and they have had pressure
[24:48] from the families of artists who have
[24:51] died to try and recreate the the the
[24:55] deceased artist for a concert,
[24:57] >> not knowing whether there's consent or
[24:58] not for that.
[24:59] >> And it's a very gray area. It's a very
[25:01] because we're talking about people who
[25:02] died maybe 10 20 years ago, icons um
[25:06] putting pressure on them as a it's a new
[25:07] revenue source for the family left
[25:09] behind for the estate.
[25:10] >> We're out of time. Pie uh Sasha
[25:13] Stephanie, thank you very much indeed.
[25:15] Um uh AI decoded next week. Uh we think
[25:18] Scott Galloway is coming on. I'm putting
[25:20] that out there so he does come on next
[25:21] week. Um so do tune in for that. If you
[25:24] have any thoughts on anything we've
[25:25] discussed, uh do email us aidbc.co.uk.
[25:29] UK. And I'm going to put on screen for
[25:32] you the QR code for the AI decoded
[25:35] playlist which is on YouTube. Some of
[25:36] you struggling to find it. There it is.
[25:38] If you scan the QR code, you'll be able
[25:39] to find it. All the backup episodes are
[25:42] there. So, do take a look at that. And
[25:44] don't forget if you want to watch us
[25:45] again, we are on the BBC i Player. Uh,
[25:48] that's all the housekeeping. Thank you
[25:50] very much for watching. Thank you to our
[25:51] guests this week. We'll see you next
[25:53] time.