1 00:00:15,680 --> 00:00:18,239 Speaker 1: Welcome to tech Stuff. I'm care Price and this is 2 00:00:18,280 --> 00:00:21,479 Speaker 1: the story. Our weekly round up episodes will be back soon, 3 00:00:21,640 --> 00:00:24,000 Speaker 1: but today I want to share something that rocked my 4 00:00:24,000 --> 00:00:29,200 Speaker 1: Hollywood community last September. It's Tilly Norwood and it's an actress. 5 00:00:30,040 --> 00:00:33,000 Speaker 1: And yes it is the correct word to use, because 6 00:00:33,040 --> 00:00:37,360 Speaker 1: Tilly Norwood is an AI actress. As you can imagine, 7 00:00:37,400 --> 00:00:40,479 Speaker 1: there was an uproar. Many A listers spoke out against 8 00:00:40,520 --> 00:00:43,440 Speaker 1: Tilly Norwood, and sag after released a statement saying they 9 00:00:43,479 --> 00:00:47,239 Speaker 1: didn't recognize Tilly as an actress. And I completely understood 10 00:00:47,240 --> 00:00:50,400 Speaker 1: the outrage. What I didn't fully get was why Tilly 11 00:00:50,479 --> 00:00:53,680 Speaker 1: Norwood existed in the first place. So I reached out 12 00:00:53,720 --> 00:00:58,560 Speaker 1: to the person behind the AI, Eleen Vandervelden. It turns 13 00:00:58,560 --> 00:01:02,560 Speaker 1: out Ellen is an actress, an actress who was extremely 14 00:01:02,640 --> 00:01:05,120 Speaker 1: good at physics and spent a few years working to 15 00:01:05,160 --> 00:01:09,600 Speaker 1: save the world through nuclear fusion, but ultimately the immediacy 16 00:01:09,640 --> 00:01:13,039 Speaker 1: of acting won her back. In twenty ten, Eleen started 17 00:01:13,040 --> 00:01:16,440 Speaker 1: to build a company called Particle six. Now that company 18 00:01:16,560 --> 00:01:20,039 Speaker 1: is a well known AI production studio, and Tillie Norwood 19 00:01:20,120 --> 00:01:23,360 Speaker 1: is a Particle six creation with a deliberately stated use. 20 00:01:24,080 --> 00:01:28,920 Speaker 1: So let's dive into our conversation. So you want to 21 00:01:28,959 --> 00:01:32,840 Speaker 1: create an actress? How do you actually go about doing that? 22 00:01:33,600 --> 00:01:35,200 Speaker 2: I know, right, where do you stop? 23 00:01:35,480 --> 00:01:37,720 Speaker 1: Yeah? Where do you begin when you want to actually 24 00:01:37,720 --> 00:01:38,600 Speaker 1: create an actress? 25 00:01:39,680 --> 00:01:42,039 Speaker 2: So I went to chat GBT and I said, I hate, 26 00:01:42,120 --> 00:01:45,400 Speaker 2: I want to make an AI actress. Did you really stop? 27 00:01:45,800 --> 00:01:46,160 Speaker 1: Yeah? 28 00:01:46,240 --> 00:01:48,800 Speaker 2: Of course I did. It's like, help me here. 29 00:01:49,320 --> 00:01:49,720 Speaker 1: Yeah. 30 00:01:49,760 --> 00:01:52,360 Speaker 2: I used it as a sort of brainstorming tool. So 31 00:01:52,400 --> 00:01:53,640 Speaker 2: I was like, okay, so I want to make the 32 00:01:53,720 --> 00:01:56,760 Speaker 2: SAI actress. I want her to resonate around the world. 33 00:01:56,880 --> 00:02:00,520 Speaker 2: I want her to be, you know, resonant with multiple 34 00:02:00,640 --> 00:02:03,800 Speaker 2: different communities around the world. I want her to have 35 00:02:03,880 --> 00:02:07,840 Speaker 2: the qualities of all the greatest you know what what 36 00:02:08,000 --> 00:02:11,160 Speaker 2: is that X factor? You know that she should have? 37 00:02:11,760 --> 00:02:13,880 Speaker 2: And the funniest thing I came back with it said 38 00:02:13,919 --> 00:02:21,400 Speaker 2: that she should look healthy. Really sure, Yes, she should 39 00:02:21,400 --> 00:02:24,880 Speaker 2: look healthy and alive. Yeah yeah, so yeah, I thought 40 00:02:24,880 --> 00:02:26,680 Speaker 2: that was a really good tip. And so we just 41 00:02:26,720 --> 00:02:29,760 Speaker 2: went through two thousand iterations, I think almost in order 42 00:02:29,880 --> 00:02:32,880 Speaker 2: to get the right look. I was very picky, you know. 43 00:02:33,000 --> 00:02:35,119 Speaker 2: Originally it came up with this sort of very air 44 00:02:35,160 --> 00:02:37,560 Speaker 2: brushed image of what she should look like in order 45 00:02:37,600 --> 00:02:41,240 Speaker 2: to become this global superstar. And I really told it 46 00:02:41,320 --> 00:02:43,840 Speaker 2: that I wanted her to become super famous. You know 47 00:02:43,919 --> 00:02:46,160 Speaker 2: that she had to become the biggest the AI A 48 00:02:46,320 --> 00:02:50,080 Speaker 2: lister was the idea, and you know, it came up 49 00:02:50,120 --> 00:02:52,760 Speaker 2: with all sorts of plans for her. But then you 50 00:02:52,800 --> 00:02:54,480 Speaker 2: sort of filter through and you go, well, that's not 51 00:02:54,960 --> 00:02:56,800 Speaker 2: that's not a good plan. That's not a good plan. 52 00:02:56,840 --> 00:02:58,519 Speaker 2: That is a good plan. You know, you just work 53 00:02:58,560 --> 00:03:03,680 Speaker 2: together with it, and eventually we had just a general 54 00:03:03,720 --> 00:03:05,679 Speaker 2: image and then we were working with all sorts of 55 00:03:05,720 --> 00:03:08,960 Speaker 2: different tools in order to create the final image and 56 00:03:09,040 --> 00:03:12,640 Speaker 2: get the texture of her skin and you know, really 57 00:03:12,639 --> 00:03:14,760 Speaker 2: what she looked like and as you steal morphed like 58 00:03:14,880 --> 00:03:17,120 Speaker 2: I don't know if you saw the AI commissioner video, 59 00:03:17,720 --> 00:03:19,680 Speaker 2: it was like a comedy sketch we did with her. 60 00:03:19,720 --> 00:03:23,040 Speaker 2: When we launched her, we'd pitched this beautiful comedy two 61 00:03:23,120 --> 00:03:27,120 Speaker 2: sisters running a funeral parlor in Margate, warm, weird, very 62 00:03:27,200 --> 00:03:31,000 Speaker 2: BBC two Commissioner said No AI generated one hundred better 63 00:03:31,040 --> 00:03:34,320 Speaker 2: ideas in minutes, perfectly aligned to channel data, viewing figures 64 00:03:34,320 --> 00:03:37,520 Speaker 2: and optimized for the audience. The winning AI format, I. 65 00:03:37,400 --> 00:03:39,560 Speaker 1: Know what your streamed last summer. 66 00:03:40,520 --> 00:03:44,880 Speaker 2: An interactive thriller built from your streaming history and delivery orders, 67 00:03:45,040 --> 00:03:47,680 Speaker 2: what and who. So it's just a bit of a 68 00:03:47,720 --> 00:03:52,080 Speaker 2: spoof and it's about the TV industry in the UK, 69 00:03:52,280 --> 00:03:55,040 Speaker 2: so it's very British. Some of the references and jokes 70 00:03:55,080 --> 00:03:57,960 Speaker 2: are very British. But we have loads of different characters, 71 00:03:58,080 --> 00:04:00,840 Speaker 2: all very diverse, and this was when we could only 72 00:04:01,080 --> 00:04:04,400 Speaker 2: do AI prompting to make them act. Now we could 73 00:04:04,480 --> 00:04:07,440 Speaker 2: also use motion capture and we're using loads of actors 74 00:04:07,880 --> 00:04:11,120 Speaker 2: in the projects that we're making, and yeah, go have 75 00:04:11,120 --> 00:04:12,360 Speaker 2: a look. It's it's funny. 76 00:04:12,800 --> 00:04:15,200 Speaker 1: When you decided to create Tilly, what was your initial 77 00:04:15,240 --> 00:04:15,880 Speaker 1: plan for her? 78 00:04:16,480 --> 00:04:21,680 Speaker 2: It was for her to be the AIA lister. I 79 00:04:21,720 --> 00:04:22,280 Speaker 2: think she is. 80 00:04:22,760 --> 00:04:25,320 Speaker 1: Would you say Tilly is a list in AI? Yes, 81 00:04:26,520 --> 00:04:29,320 Speaker 1: she's like the top, She's the kreme de la creme actress. 82 00:04:29,480 --> 00:04:31,560 Speaker 2: She's the Kremlin la creme AI actress. 83 00:04:31,720 --> 00:04:33,800 Speaker 1: I think I think I would. I mean, she's definitely 84 00:04:33,800 --> 00:04:39,000 Speaker 1: the most known. She's definitely the most known actress. 85 00:04:39,480 --> 00:04:42,120 Speaker 2: But she now has to go and work and show 86 00:04:42,200 --> 00:04:45,240 Speaker 2: her acting. So that will be our next release will 87 00:04:45,279 --> 00:04:46,640 Speaker 2: be very much focused on that. 88 00:04:47,240 --> 00:04:50,680 Speaker 1: I think people might have a hard time wrapping their 89 00:04:50,680 --> 00:04:54,520 Speaker 1: head around this just because they're not people who create 90 00:04:55,080 --> 00:04:59,280 Speaker 1: digital avatars, Like, how did you choose what she was 91 00:04:59,360 --> 00:05:00,960 Speaker 1: going to look like? Gosh? 92 00:05:01,080 --> 00:05:03,600 Speaker 2: I mean that really comes down to my acting side 93 00:05:03,600 --> 00:05:06,760 Speaker 2: of things to you know, like when you're like I 94 00:05:06,760 --> 00:05:11,040 Speaker 2: did so many pilot seasons and you know, auditions, and 95 00:05:11,880 --> 00:05:15,719 Speaker 2: it's a very specific type that gets cast, a sort 96 00:05:15,720 --> 00:05:19,159 Speaker 2: of a leading female. You know, it's usually actually not 97 00:05:19,240 --> 00:05:21,679 Speaker 2: a blonde, right, So I wanted to create the girl 98 00:05:21,760 --> 00:05:24,880 Speaker 2: that does get cast, and I wanted her to be 99 00:05:25,040 --> 00:05:28,919 Speaker 2: really natural looking, and you know, it's it's almost like 100 00:05:29,960 --> 00:05:32,240 Speaker 2: your own Barbie in a way, right I could. I 101 00:05:32,279 --> 00:05:35,880 Speaker 2: could create my own Barbie and dress her and you know, 102 00:05:36,360 --> 00:05:40,080 Speaker 2: it's all just playing, right, it's creative play. And it's 103 00:05:40,080 --> 00:05:42,080 Speaker 2: the same way I would create a character if I 104 00:05:42,120 --> 00:05:44,240 Speaker 2: was on stage at an improv show. It's the same 105 00:05:44,279 --> 00:05:45,800 Speaker 2: way I would create a character like I did for 106 00:05:45,839 --> 00:05:49,120 Speaker 2: the BBC. It's exactly the same process. So it's like 107 00:05:49,120 --> 00:05:52,279 Speaker 2: an onion, right, you start with the outside usually so well, 108 00:05:52,320 --> 00:05:54,120 Speaker 2: you know, when you when you create a character, you'd 109 00:05:54,160 --> 00:05:56,360 Speaker 2: put certain earrings on and you go and she talks 110 00:05:56,400 --> 00:05:58,960 Speaker 2: like this, and she has these ear rings and and 111 00:05:59,000 --> 00:06:01,839 Speaker 2: then you go down to the outfit and you peel 112 00:06:01,880 --> 00:06:04,560 Speaker 2: the layers off, and then you think about the backstory 113 00:06:04,560 --> 00:06:07,280 Speaker 2: and the motivation. Why do they say these things and 114 00:06:07,360 --> 00:06:09,040 Speaker 2: why they say in this way, and how do they 115 00:06:09,080 --> 00:06:12,400 Speaker 2: react to this? And so that peeling of the layers 116 00:06:12,520 --> 00:06:15,480 Speaker 2: with Tilly is the exact same process, And we're thinking 117 00:06:15,520 --> 00:06:17,480 Speaker 2: about this world that she's living in and who are 118 00:06:17,520 --> 00:06:21,599 Speaker 2: her friends, and what's her personality like and her sense 119 00:06:21,640 --> 00:06:25,360 Speaker 2: of humor, And it's an endless process of onion peeling. 120 00:06:25,760 --> 00:06:28,359 Speaker 1: Will you change the way she looks or she's set? 121 00:06:28,800 --> 00:06:31,080 Speaker 2: She's pretty set. Yeah, she's pretty set. She can get 122 00:06:31,080 --> 00:06:34,000 Speaker 2: older or younger, but her image is now so set, 123 00:06:34,040 --> 00:06:37,080 Speaker 2: and we can do you think short age? I don't 124 00:06:37,160 --> 00:06:39,919 Speaker 2: think she will age unless we want her to for 125 00:06:39,960 --> 00:06:43,360 Speaker 2: a specific role. And this is happening quite a lot now. 126 00:06:43,400 --> 00:06:45,880 Speaker 2: We get a lot of requests for films where they 127 00:06:45,920 --> 00:06:48,279 Speaker 2: want the same actors to play their younger selves or 128 00:06:48,279 --> 00:06:52,400 Speaker 2: their older selves. Like that's all possible and great that 129 00:06:52,520 --> 00:06:55,599 Speaker 2: they can, right that they don't need to cast child 130 00:06:55,640 --> 00:07:00,320 Speaker 2: actors to do that. So yeah, no, I think Also, 131 00:07:00,360 --> 00:07:02,440 Speaker 2: I think it's more ethical when it comes to children 132 00:07:02,480 --> 00:07:05,680 Speaker 2: or animals, right, Like I've filmed with children before. There's 133 00:07:05,720 --> 00:07:09,720 Speaker 2: no need if you can avoid it, that's better, and 134 00:07:09,840 --> 00:07:10,800 Speaker 2: the same with animals. 135 00:07:11,560 --> 00:07:14,840 Speaker 1: Interesting, I bet some people would have different opinion on that. 136 00:07:15,640 --> 00:07:17,920 Speaker 2: Yeah, yeah, and they can. But a lot of bad 137 00:07:17,960 --> 00:07:22,560 Speaker 2: things have happened to kids on sets and sets very safe, yeah, 138 00:07:22,800 --> 00:07:25,200 Speaker 2: not very safe environments, So you know they can have 139 00:07:25,240 --> 00:07:27,160 Speaker 2: that opinion, but I disagree. 140 00:07:27,680 --> 00:07:31,960 Speaker 1: Yeah. So in the CBS Sunday Morning segment that you did, 141 00:07:32,000 --> 00:07:37,400 Speaker 1: you taught Tilly acting how do you teach acting to AI? 142 00:07:38,080 --> 00:07:42,200 Speaker 2: So back in that time, which is November December twenty 143 00:07:42,240 --> 00:07:45,000 Speaker 2: twenty five, it was only still possible to prompt an 144 00:07:45,040 --> 00:07:47,760 Speaker 2: AI actor to teach it how to you know, what 145 00:07:47,920 --> 00:07:50,760 Speaker 2: to say and how to say it. Now we're doing 146 00:07:50,760 --> 00:07:52,040 Speaker 2: way more performance capture. 147 00:07:52,320 --> 00:07:57,080 Speaker 1: Oh interesting. Yeah, so who's doing that acting? Me? 148 00:07:57,080 --> 00:07:57,480 Speaker 2: Me, me me. 149 00:07:57,960 --> 00:08:02,520 Speaker 1: Yeah, so you, in a sense, Tilly is imbued with you. Yeah. 150 00:08:02,720 --> 00:08:06,240 Speaker 2: Absolutely. I think this is the biggest misconception people have 151 00:08:06,320 --> 00:08:08,240 Speaker 2: around the world is that they don't think there's any 152 00:08:08,320 --> 00:08:13,880 Speaker 2: human behind AI things across the board, and there's so 153 00:08:14,040 --> 00:08:17,040 Speaker 2: much human work that goes on behind the scenes. I mean, 154 00:08:17,040 --> 00:08:19,120 Speaker 2: we've got a team of eight people working on the 155 00:08:19,200 --> 00:08:22,440 Speaker 2: Tilly verse. You know, it's it's a huge endeavor and 156 00:08:22,480 --> 00:08:24,000 Speaker 2: it's a really creative endeavor. 157 00:08:24,400 --> 00:08:29,600 Speaker 1: I'm just so interested that you are doing the motion capture. Yeah, yeah, 158 00:08:29,680 --> 00:08:32,080 Speaker 1: so that's so. I mean, where are you doing it? 159 00:08:32,600 --> 00:08:34,200 Speaker 1: Oh you can just do it on an iPhone in 160 00:08:34,200 --> 00:08:36,719 Speaker 1: a studio. No, you can just do it anyway. Oh 161 00:08:36,760 --> 00:08:37,880 Speaker 1: you're doing it on an iPhone. 162 00:08:38,280 --> 00:08:41,120 Speaker 2: Yeah, yeah, you can walk around on an iPhone. Anything's 163 00:08:41,160 --> 00:08:45,560 Speaker 2: possible now, Yeah, unbelievable, But that tech has only arrived 164 00:08:45,640 --> 00:08:52,280 Speaker 2: like the last month or two. Yeah, through in chat GBT. No, no, no, 165 00:08:52,320 --> 00:08:54,680 Speaker 2: We're using all sorts of different programs. So our team 166 00:08:54,800 --> 00:08:58,360 Speaker 2: at Particle Sex is really well versed in probably like 167 00:08:58,559 --> 00:09:03,040 Speaker 2: fifty different tools, and we have a proprietary workflow that 168 00:09:03,080 --> 00:09:05,040 Speaker 2: we use with the different tools in order to achieve 169 00:09:05,080 --> 00:09:08,520 Speaker 2: a certain quality. So that's really what people you know, 170 00:09:09,040 --> 00:09:12,000 Speaker 2: come to us for and Tilly is just you know, 171 00:09:12,679 --> 00:09:15,920 Speaker 2: an example of that. But we do it for commercials, 172 00:09:15,920 --> 00:09:17,680 Speaker 2: we do it for film and TV work, you know, 173 00:09:17,679 --> 00:09:18,640 Speaker 2: we do it across the board. 174 00:09:19,160 --> 00:09:21,400 Speaker 1: And what's your favorite thing that Tilly has done so far? 175 00:09:21,920 --> 00:09:22,600 Speaker 1: Like what role? 176 00:09:23,360 --> 00:09:25,040 Speaker 2: So I can't talk about the roles yet because it's 177 00:09:25,040 --> 00:09:29,400 Speaker 2: still in you know, in progress, but what she does do, 178 00:09:29,480 --> 00:09:32,840 Speaker 2: which I like, I find the prompting quite funny with 179 00:09:32,920 --> 00:09:35,880 Speaker 2: the AI acting because you know, anyone who's worked with 180 00:09:35,880 --> 00:09:38,200 Speaker 2: the eye knows that it doesn't really listen to you 181 00:09:38,360 --> 00:09:41,480 Speaker 2: very well. So I like the unexpectedness of it. And 182 00:09:41,520 --> 00:09:45,160 Speaker 2: I once asked her to dance in blue jeans and 183 00:09:45,240 --> 00:09:50,480 Speaker 2: she was just in her underwear. She once I got 184 00:09:50,480 --> 00:09:53,880 Speaker 2: her flying on a on a flamingo and she turned 185 00:09:53,920 --> 00:09:57,520 Speaker 2: around to me and she goes, take that, Hollywood. I 186 00:09:57,559 --> 00:09:59,160 Speaker 2: was like, I didn't tell you to say that. 187 00:09:59,800 --> 00:10:04,000 Speaker 1: She she just said that untrained. Yeah, yeah, yeah. Is 188 00:10:04,040 --> 00:10:07,560 Speaker 1: it weird seeing something that was trained on you? So 189 00:10:08,000 --> 00:10:11,240 Speaker 1: she's not technically trained on me. We're training her brain 190 00:10:12,080 --> 00:10:14,120 Speaker 1: where we're creating her brain, So it's like a creative 191 00:10:14,120 --> 00:10:17,440 Speaker 1: writing exercise, right, It's built on top of an LLM, 192 00:10:17,800 --> 00:10:21,640 Speaker 1: and we give her guardrails, and we give her morals 193 00:10:21,640 --> 00:10:22,120 Speaker 1: and values. 194 00:10:22,120 --> 00:10:24,480 Speaker 2: It's very much like a child. You know, you bring 195 00:10:24,559 --> 00:10:26,080 Speaker 2: up a child and you send them off into the 196 00:10:26,080 --> 00:10:30,000 Speaker 2: world at eighteen, you go, good luck. Yeah, And it's 197 00:10:30,000 --> 00:10:31,760 Speaker 2: just sort of the same with Chilly. Just knows shorter 198 00:10:31,880 --> 00:10:35,000 Speaker 2: amount of time. So eventually we'll send her off with 199 00:10:35,600 --> 00:10:40,760 Speaker 2: her brain and hope for the best. We're testing her 200 00:10:40,800 --> 00:10:43,360 Speaker 2: a lot, as you can imagine, is. 201 00:10:43,320 --> 00:10:46,520 Speaker 1: It uncanny to see her with the motion capture. 202 00:10:46,679 --> 00:10:49,800 Speaker 2: Yes, yeah, it's quite a lot of my friends they go, 203 00:10:49,840 --> 00:10:52,360 Speaker 2: oh yeah, I can totally see you in her. 204 00:10:53,000 --> 00:10:54,000 Speaker 1: Is it uncanny for you? 205 00:10:54,240 --> 00:10:56,760 Speaker 2: I mean, it's no different to watching myself on screen 206 00:10:56,800 --> 00:10:59,280 Speaker 2: as an actor, which I guess you're used to. It 207 00:10:59,360 --> 00:11:01,640 Speaker 2: is always weird, yeah, I mean, but people have got 208 00:11:01,640 --> 00:11:03,360 Speaker 2: a lot more used to that now because they're watching 209 00:11:03,360 --> 00:11:05,880 Speaker 2: themselves on screen in video calls all the time, so 210 00:11:06,559 --> 00:11:08,000 Speaker 2: it's less weird than it used to be. 211 00:11:08,080 --> 00:11:23,760 Speaker 1: I think after the break, Whileen didn't think Tilly nor 212 00:11:23,760 --> 00:11:40,720 Speaker 1: would would become that controversial stay with us. So the 213 00:11:40,800 --> 00:11:43,920 Speaker 1: SAG union has insisted that Tilly be referred to as 214 00:11:44,000 --> 00:11:47,640 Speaker 1: an avatar or a computer generated character. How do you 215 00:11:47,720 --> 00:11:51,240 Speaker 1: feel about that and not a real actor? 216 00:11:52,040 --> 00:11:55,319 Speaker 2: Yeah, she is a computer generated character. Yeah, And I 217 00:11:55,800 --> 00:11:58,839 Speaker 2: totally agree with SAG on all of that. The film 218 00:11:58,880 --> 00:12:01,400 Speaker 2: You Know Avatar, where they Zary Saldana plays a character 219 00:12:01,559 --> 00:12:03,760 Speaker 2: like it's exactly the same, that she's just the same 220 00:12:03,800 --> 00:12:06,560 Speaker 2: as Darth Vado or any other character that humans have 221 00:12:06,640 --> 00:12:09,680 Speaker 2: created before her, And the only difference is is that 222 00:12:10,280 --> 00:12:13,199 Speaker 2: I called her an AI actor because I didn't want 223 00:12:13,200 --> 00:12:16,400 Speaker 2: Tilly to be limited to only playing one character. So 224 00:12:16,440 --> 00:12:20,800 Speaker 2: the beauty is she can play multiple different roles and 225 00:12:20,840 --> 00:12:23,120 Speaker 2: that's also part of her story going forward. 226 00:12:23,440 --> 00:12:25,360 Speaker 1: So you don't want her to only be Tilly as 227 00:12:25,360 --> 00:12:26,160 Speaker 1: an actress. 228 00:12:26,760 --> 00:12:30,000 Speaker 2: No, no, she's already playing loads of different roles. It's 229 00:12:30,080 --> 00:12:30,480 Speaker 2: really fun. 230 00:12:30,559 --> 00:12:33,960 Speaker 1: Yeah. Yeah, so in that way, she's not just an avatar. 231 00:12:34,040 --> 00:12:34,760 Speaker 1: She's an actress. 232 00:12:35,640 --> 00:12:38,760 Speaker 2: Yes, yes she is, but she's also a computer generated 233 00:12:38,960 --> 00:12:42,120 Speaker 2: Like we never shied away from the fact that she's AI. 234 00:12:42,520 --> 00:12:45,880 Speaker 2: She's one hundred percent computer generated, nothing else. And yes, 235 00:12:45,960 --> 00:12:48,520 Speaker 2: we direct her with prompts and we direct her with 236 00:12:48,559 --> 00:12:52,440 Speaker 2: my emotion capture. But yeah, let's not say she's anything 237 00:12:52,440 --> 00:12:54,040 Speaker 2: else than computer generated. 238 00:12:54,400 --> 00:12:57,160 Speaker 1: So obviously, when Tilly was first announced, there was a 239 00:12:57,160 --> 00:13:00,840 Speaker 1: lot of backlash from Hollywood. Yeah did you expect this. 240 00:13:01,400 --> 00:13:04,920 Speaker 2: Because we'd launched her in July in the UK with 241 00:13:05,000 --> 00:13:09,000 Speaker 2: the comedy Sketch AI commissioner, and you know, everyone thought 242 00:13:09,040 --> 00:13:13,000 Speaker 2: it was quite fun and there was absolutely no backlash 243 00:13:13,080 --> 00:13:16,040 Speaker 2: whatsoever because loads of all the production companies in the 244 00:13:16,120 --> 00:13:18,440 Speaker 2: UK were using AI and they were like this is cool, Like, 245 00:13:18,480 --> 00:13:21,520 Speaker 2: you guys are clearly like leading the front on this, 246 00:13:21,800 --> 00:13:27,800 Speaker 2: and you know, people were coming to us because most studios, 247 00:13:27,880 --> 00:13:31,520 Speaker 2: most companies are embracing AI here because we get small 248 00:13:31,559 --> 00:13:35,400 Speaker 2: budgets and we want max value on screen, so you know, 249 00:13:35,440 --> 00:13:37,680 Speaker 2: all the help we can get in order to achieve that. 250 00:13:38,400 --> 00:13:40,920 Speaker 2: And I was showing Tilly around sort of to educate 251 00:13:41,040 --> 00:13:43,800 Speaker 2: the creative community. It's not like we were taking any 252 00:13:43,880 --> 00:13:47,839 Speaker 2: jobs away from real actors. That was never never the plan. 253 00:13:48,040 --> 00:13:50,880 Speaker 2: It was never anything we did. And so but I 254 00:13:50,960 --> 00:13:54,480 Speaker 2: felt compelled to show the crazy community what was possible, 255 00:13:54,600 --> 00:13:58,040 Speaker 2: right because I think what's even worse is digging your 256 00:13:58,080 --> 00:14:00,720 Speaker 2: head in the sand and not knowing what the text doing, 257 00:14:00,800 --> 00:14:03,120 Speaker 2: so that you can't make any decisions over where the 258 00:14:03,120 --> 00:14:05,520 Speaker 2: creative community is going to go next, and then it's 259 00:14:05,520 --> 00:14:08,240 Speaker 2: going to be imposed on you by by the tech 260 00:14:08,240 --> 00:14:11,880 Speaker 2: community as to what happens next. So I was doing 261 00:14:11,880 --> 00:14:13,760 Speaker 2: a lot of trade shows and I was showing Tilly around, 262 00:14:14,920 --> 00:14:17,520 Speaker 2: and then obviously the American audience got hold of it 263 00:14:17,520 --> 00:14:20,720 Speaker 2: with this deadline article and they just went crazy thinking 264 00:14:20,720 --> 00:14:22,720 Speaker 2: that she was going to take all the jobs because 265 00:14:22,720 --> 00:14:25,480 Speaker 2: she looked so real, and I totally get it, like I, 266 00:14:25,800 --> 00:14:27,520 Speaker 2: you know, you going from zero to one hundred. A 267 00:14:27,560 --> 00:14:29,880 Speaker 2: lot of these people had never seen anything AI generated, 268 00:14:29,880 --> 00:14:32,480 Speaker 2: and suddenly you see Tilly, You're like, oh, my goodness, 269 00:14:33,120 --> 00:14:33,920 Speaker 2: you know, I get it. 270 00:14:34,600 --> 00:14:36,680 Speaker 1: I mean it's also a little bit absurd to think 271 00:14:36,680 --> 00:14:40,840 Speaker 1: that anybody, any one entity could take all of the jobs. 272 00:14:40,880 --> 00:14:42,800 Speaker 1: It's kind of like a new actress coming on the 273 00:14:42,800 --> 00:14:45,160 Speaker 1: scene and being like, well, Jennifer, we should stop Jennifer 274 00:14:45,200 --> 00:14:49,040 Speaker 1: Lawrence because she's going to take all the jobs. Yeah. 275 00:14:49,160 --> 00:14:54,800 Speaker 2: I think my message was clear, right, like, let's be 276 00:14:54,920 --> 00:15:00,240 Speaker 2: aware of this because AI actors are possible now, yeah, 277 00:15:00,280 --> 00:15:02,560 Speaker 2: and so how do we control that and how we 278 00:15:03,000 --> 00:15:06,280 Speaker 2: do we have a say in this going forward as 279 00:15:06,320 --> 00:15:09,480 Speaker 2: actors ourselves, so also as an actor myself. It's a 280 00:15:09,480 --> 00:15:12,760 Speaker 2: way to future proof, right. So I'm telling other actors like, hey, 281 00:15:12,840 --> 00:15:15,640 Speaker 2: you can create an AI digital twin of yourself that 282 00:15:15,680 --> 00:15:18,520 Speaker 2: you can control and you own the copyright too. Or 283 00:15:18,640 --> 00:15:23,440 Speaker 2: you could you know, become an actor that controls with 284 00:15:23,560 --> 00:15:26,720 Speaker 2: motion capture. These AI characters, Like, they're going to be 285 00:15:26,760 --> 00:15:30,600 Speaker 2: loads of new different types of jobs, you know. I 286 00:15:30,640 --> 00:15:32,400 Speaker 2: think there's a lot of fear out there where people 287 00:15:32,400 --> 00:15:34,280 Speaker 2: are like, oh, it's going to take all the jobs away, 288 00:15:34,320 --> 00:15:37,760 Speaker 2: Like we are hiring loads of people, there's loads of 289 00:15:37,800 --> 00:15:40,880 Speaker 2: new jobs. It's it's exciting. It's the biggest boom I've 290 00:15:40,920 --> 00:15:42,840 Speaker 2: seen in our industry in the last ten years. 291 00:15:44,480 --> 00:15:47,840 Speaker 1: When Tilly was first introduced to the US, you said 292 00:15:47,840 --> 00:15:49,800 Speaker 1: that you were looking for an agent for her. Does 293 00:15:49,840 --> 00:15:50,920 Speaker 1: she have an agent yet? 294 00:15:52,040 --> 00:15:54,680 Speaker 2: So we decided to not go with an agency. We 295 00:15:54,680 --> 00:15:58,080 Speaker 2: were talking to loads of them because we felt like, actually, 296 00:15:58,720 --> 00:16:01,720 Speaker 2: there should be an agent at one point that specializes 297 00:16:01,840 --> 00:16:04,680 Speaker 2: in these AI characters, which I think there will come. Also, 298 00:16:04,720 --> 00:16:07,680 Speaker 2: she doesn't really need an agent right now anymore because 299 00:16:07,720 --> 00:16:10,560 Speaker 2: she became so famous. She didn't need an agent anymore 300 00:16:10,680 --> 00:16:15,479 Speaker 2: to get jobs because there was just an influx, massively 301 00:16:15,560 --> 00:16:18,760 Speaker 2: overwhelming inbound reaction. 302 00:16:18,960 --> 00:16:22,080 Speaker 1: So yeah, so she does not have an agent now. 303 00:16:22,240 --> 00:16:25,320 Speaker 1: She doesn't need one. She does not need one, and 304 00:16:25,400 --> 00:16:28,040 Speaker 1: apparently she's turned down film and TV offers. 305 00:16:28,880 --> 00:16:32,000 Speaker 2: Yes, yes, so we you know, we didn't set out 306 00:16:32,040 --> 00:16:36,440 Speaker 2: for her to be replacing real actors in real roles 307 00:16:36,720 --> 00:16:39,600 Speaker 2: right now at all. That's that's never the plan. So 308 00:16:39,760 --> 00:16:42,960 Speaker 2: any people that come to us with like, you know, oh, 309 00:16:43,080 --> 00:16:45,360 Speaker 2: you know, we wanted to cast Jennifer Lawrence for this role, 310 00:16:45,400 --> 00:16:47,800 Speaker 2: but actually it could be Tilly, Like, We're like, no, 311 00:16:47,880 --> 00:16:51,560 Speaker 2: that's not the plan. Yeah, However, if there's a role 312 00:16:51,600 --> 00:16:55,200 Speaker 2: that's like an AI character, in the script or something 313 00:16:55,200 --> 00:16:58,120 Speaker 2: that makes sense, then obviously we would might consider it. 314 00:16:58,480 --> 00:17:02,040 Speaker 2: But really, she's living in her AI world and we're 315 00:17:02,080 --> 00:17:05,080 Speaker 2: creating that whole AI world as we speak. 316 00:17:04,960 --> 00:17:07,639 Speaker 1: I see. So she's not like, we're not going to 317 00:17:07,680 --> 00:17:10,119 Speaker 1: see her winning any Oscars soon unless there was a 318 00:17:10,160 --> 00:17:11,760 Speaker 1: new category in the Oscars. 319 00:17:12,320 --> 00:17:15,280 Speaker 2: Yeah, I think a new category would work. But also, 320 00:17:15,359 --> 00:17:17,720 Speaker 2: you know, if people are making AI films, you know, 321 00:17:17,760 --> 00:17:20,600 Speaker 2: there's loads of great AI short films out there that 322 00:17:20,680 --> 00:17:24,600 Speaker 2: are fully AI. She could be in them right absolutely, 323 00:17:24,720 --> 00:17:27,320 Speaker 2: because we were never going to have a real actress 324 00:17:27,400 --> 00:17:29,240 Speaker 2: being them unless it was a digital twin, which is 325 00:17:29,280 --> 00:17:30,960 Speaker 2: the equivalent of Tilly. 326 00:17:31,280 --> 00:17:34,720 Speaker 1: That's so interesting. I don't know. I'm just so enamored of. 327 00:17:36,400 --> 00:17:38,200 Speaker 2: People have a lot to get their heads around. I've 328 00:17:38,200 --> 00:17:42,440 Speaker 2: been thinking about this for three years and people are like, what, No, It's. 329 00:17:42,280 --> 00:17:46,520 Speaker 1: Just interesting because I think the original backlash was this idea, 330 00:17:46,520 --> 00:17:50,280 Speaker 1: and it's people being somewhat short sighted about the fact 331 00:17:50,320 --> 00:17:53,920 Speaker 1: that an AI actress is going to replace real actresses. 332 00:17:53,920 --> 00:17:56,760 Speaker 1: But what you're saying is very much on the contrary, 333 00:17:57,280 --> 00:17:59,639 Speaker 1: which is like, this is an AI actress who is 334 00:17:59,800 --> 00:18:01,679 Speaker 1: entering a new category of actor. 335 00:18:02,040 --> 00:18:04,760 Speaker 2: Correct, But it's hard for people to understand. Right, it's 336 00:18:04,800 --> 00:18:07,760 Speaker 2: all new, like it's going to take time. So she's 337 00:18:07,760 --> 00:18:10,720 Speaker 2: not doing live action, no, no, no, we've always said 338 00:18:10,720 --> 00:18:12,800 Speaker 2: and that was in our original statement as well, like 339 00:18:12,880 --> 00:18:16,000 Speaker 2: she's only in AI content full AI content. 340 00:18:16,320 --> 00:18:20,399 Speaker 1: Is your intent to create more AI characters like Tilly? 341 00:18:21,000 --> 00:18:24,000 Speaker 2: Yes, so our plan is to create a whole universe 342 00:18:24,119 --> 00:18:26,760 Speaker 2: and they all have you know, they're all characters in there. 343 00:18:26,840 --> 00:18:29,480 Speaker 2: So we're creating a whole TV show for Tilly really 344 00:18:29,560 --> 00:18:30,879 Speaker 2: yeah who really? 345 00:18:31,640 --> 00:18:33,920 Speaker 1: Yeah? Yeah yeah yeah yeah, So a TV show meaning 346 00:18:33,960 --> 00:18:36,600 Speaker 1: like Seinfeld for Tilly. Yeah. 347 00:18:36,640 --> 00:18:38,320 Speaker 2: I don't want to give it away too much, but yes, 348 00:18:38,400 --> 00:18:40,720 Speaker 2: there's a whole load of new stuff coming out with 349 00:18:40,840 --> 00:18:43,520 Speaker 2: Tilly in it. And she's also doing lots of projects. 350 00:18:43,560 --> 00:18:46,000 Speaker 2: So we've got loads of directors that approach us they 351 00:18:46,040 --> 00:18:47,640 Speaker 2: want to work with Tilly because they want to learn 352 00:18:47,640 --> 00:18:51,560 Speaker 2: about AI. You know, we have costume designers, production designers, 353 00:18:51,800 --> 00:18:54,560 Speaker 2: and we all let them work with Tilly. It's like 354 00:18:54,600 --> 00:18:57,520 Speaker 2: they called the Tilly Blayground. Let them work with Tilly 355 00:18:58,480 --> 00:19:00,600 Speaker 2: and that way they can discover how to use AI 356 00:19:00,640 --> 00:19:04,840 Speaker 2: in their workflow and reskill, retool them for this new, 357 00:19:05,880 --> 00:19:09,920 Speaker 2: this new this, it's an AI revolution, really that we're 358 00:19:09,920 --> 00:19:11,960 Speaker 2: going through, and it's a new AI realm that we 359 00:19:12,000 --> 00:19:14,240 Speaker 2: all have to work in. We all have to understand 360 00:19:14,320 --> 00:19:15,000 Speaker 2: how it works. 361 00:19:15,480 --> 00:19:17,680 Speaker 1: If you had a time machine, would you create Tilly again? 362 00:19:18,240 --> 00:19:18,720 Speaker 2: Oh yeah? 363 00:19:18,840 --> 00:19:20,560 Speaker 1: And would you launch would you launch her in the 364 00:19:20,560 --> 00:19:21,040 Speaker 1: same way? 365 00:19:21,600 --> 00:19:25,560 Speaker 2: I mean, yeah, I suppose yes it. You know, like 366 00:19:26,040 --> 00:19:28,800 Speaker 2: as much as publicly there's been this backlash, you know, 367 00:19:29,440 --> 00:19:34,159 Speaker 2: creatively and privately, everyone has come to us because we're's 368 00:19:34,160 --> 00:19:36,000 Speaker 2: sort of the leaders in this space. I would say 369 00:19:36,040 --> 00:19:38,679 Speaker 2: we're probably one of the best AI production studios. I 370 00:19:38,680 --> 00:19:41,520 Speaker 2: think people saw the quality we could do. You know, 371 00:19:42,080 --> 00:19:45,320 Speaker 2: it's it's been great for us. We've been approached by everybody. 372 00:19:45,359 --> 00:19:49,879 Speaker 2: It's fantastic. So you know that a backlash can be 373 00:19:49,920 --> 00:19:50,680 Speaker 2: a positive thing. 374 00:19:51,359 --> 00:19:53,720 Speaker 1: It Also, I think the backlash was a little bit 375 00:19:54,440 --> 00:19:57,199 Speaker 1: shortsighted in not understanding what you were actually trying to 376 00:19:57,240 --> 00:19:57,879 Speaker 1: do with Tilly. 377 00:19:58,200 --> 00:20:02,480 Speaker 2: Yeah, they just saw a head and thought something when 378 00:20:02,560 --> 00:20:06,679 Speaker 2: really they didn't delve deeper into it. But also I 379 00:20:06,720 --> 00:20:08,879 Speaker 2: can see why it might have been a shock for people, 380 00:20:09,000 --> 00:20:12,000 Speaker 2: right like the zero to one hundred thing with AI, 381 00:20:12,040 --> 00:20:17,720 Speaker 2: and I understood it totally. But people adapt very quickly, 382 00:20:17,760 --> 00:20:19,440 Speaker 2: and you can see that now, you know, We're three 383 00:20:19,640 --> 00:20:23,919 Speaker 2: four months on and the conversation has completely shifted again, 384 00:20:24,640 --> 00:20:27,280 Speaker 2: and everyone's like, how do we use AI? You know, 385 00:20:27,320 --> 00:20:31,040 Speaker 2: and we're seeing amazing creative projects being made with AI. 386 00:20:31,920 --> 00:20:34,160 Speaker 2: You know all the super Bowl ads where AI. 387 00:20:34,680 --> 00:20:36,560 Speaker 1: There was you know much AI and the Super Bowl 388 00:20:36,600 --> 00:20:37,680 Speaker 1: ads it was incredible. 389 00:20:39,119 --> 00:20:44,000 Speaker 2: So you know, it's it's coming. There's not you know, 390 00:20:44,600 --> 00:20:46,800 Speaker 2: there's nothing we can do about it. It's coming, So 391 00:20:46,880 --> 00:20:50,600 Speaker 2: let's think about how we approach it ethically and do 392 00:20:50,720 --> 00:20:51,160 Speaker 2: it right. 393 00:20:52,040 --> 00:20:56,120 Speaker 1: Do you foresee a film that is completely AI generated 394 00:20:56,119 --> 00:20:58,880 Speaker 1: that could be nominated for an asker? Yes? 395 00:20:59,119 --> 00:21:02,280 Speaker 2: The Academy has said they don't discriminate against it, right, 396 00:21:02,359 --> 00:21:06,440 Speaker 2: so you know, it's still about storytelling and with or 397 00:21:06,480 --> 00:21:07,240 Speaker 2: without AI. 398 00:21:07,760 --> 00:21:08,520 Speaker 1: So yes, I think. 399 00:21:09,000 --> 00:21:11,800 Speaker 2: I think within a year's time, there won't be a 400 00:21:11,840 --> 00:21:15,040 Speaker 2: film that hasn't got some element of AI in it, right, 401 00:21:15,040 --> 00:21:19,320 Speaker 2: whether it's in post production, whether it's in just voice cleanup, 402 00:21:19,560 --> 00:21:22,720 Speaker 2: you know, an audio tidy, you know, whether the production 403 00:21:23,119 --> 00:21:26,960 Speaker 2: teams have used chat, GPT or Gemini. Of course, like 404 00:21:27,040 --> 00:21:28,960 Speaker 2: everyone's going to be using AI. It's going to be 405 00:21:28,960 --> 00:21:29,800 Speaker 2: like electricity. 406 00:21:30,320 --> 00:21:34,000 Speaker 1: It is different though to have an actual character that's 407 00:21:34,000 --> 00:21:38,240 Speaker 1: AI generated, then to use AI to facilitate production. 408 00:21:40,280 --> 00:21:44,640 Speaker 2: So we started just having AI to facilitate production. Right 409 00:21:44,720 --> 00:21:46,520 Speaker 2: when we started three years ago, it was all about 410 00:21:46,520 --> 00:21:51,879 Speaker 2: optimizing the workflow and not it wasn't visible on screen 411 00:21:51,920 --> 00:21:54,320 Speaker 2: what we were doing with AI, but you very quickly 412 00:21:54,440 --> 00:21:57,040 Speaker 2: moved to So it's a gradual process, right. So first 413 00:21:57,080 --> 00:22:00,680 Speaker 2: it's just in production workflows. Then it's in post production, 414 00:22:01,040 --> 00:22:03,280 Speaker 2: which I mean most of the post production of films 415 00:22:03,320 --> 00:22:05,439 Speaker 2: has AI in it. Yeah, you know, people you're just 416 00:22:05,520 --> 00:22:07,840 Speaker 2: using VFX houses and they just don't know. They're calling 417 00:22:07,840 --> 00:22:11,160 Speaker 2: it CGI, but really they're all using AI. And then 418 00:22:11,240 --> 00:22:14,000 Speaker 2: the next step, you know, once most of your image 419 00:22:14,359 --> 00:22:17,479 Speaker 2: or video is AI, you know, it's the character and 420 00:22:17,560 --> 00:22:20,080 Speaker 2: most of that will soon be done with motion capture 421 00:22:20,119 --> 00:22:24,679 Speaker 2: with AI characters too, because you know, it's easier for 422 00:22:24,760 --> 00:22:28,639 Speaker 2: production and more cost effective. So you'll still have actors, 423 00:22:28,960 --> 00:22:32,440 Speaker 2: they'll just be acting more comfortably in a rehearsal studio 424 00:22:32,960 --> 00:22:36,080 Speaker 2: and be really focused on the craft than anything else. 425 00:22:36,960 --> 00:22:38,879 Speaker 1: Is there anything that I haven't asked you that you 426 00:22:38,920 --> 00:22:41,720 Speaker 1: think is relevant totily that you would like to say 427 00:22:41,720 --> 00:22:42,600 Speaker 1: about her? Gosh? 428 00:22:42,680 --> 00:22:47,560 Speaker 2: I mean, I think you know she was created to 429 00:22:47,760 --> 00:22:50,679 Speaker 2: entertain right, I'm in the entertainment industry. I think so 430 00:22:50,840 --> 00:22:55,800 Speaker 2: far she has entertained people with the backlash and everything 431 00:22:55,800 --> 00:23:02,119 Speaker 2: around it. And yeah, it's just like any human creation. 432 00:23:02,480 --> 00:23:05,560 Speaker 2: It's it's a bit of fun and we're we're all 433 00:23:05,600 --> 00:23:07,000 Speaker 2: just trying to have a bit of fun. We're not 434 00:23:07,040 --> 00:23:11,320 Speaker 2: taking anyone's jobs. And I think I don't think AI 435 00:23:11,520 --> 00:23:14,960 Speaker 2: is as scary as people think it isn't, you know. 436 00:23:15,000 --> 00:23:16,760 Speaker 2: And then they play with it and then they go, oh, 437 00:23:16,880 --> 00:23:18,960 Speaker 2: is this it? And I'm like, yeah, this is it. 438 00:23:18,960 --> 00:23:21,720 Speaker 2: It's just it's fine for now. 439 00:23:21,680 --> 00:23:22,600 Speaker 1: This kind of AI. 440 00:23:23,240 --> 00:23:26,600 Speaker 2: Yes, yes, I wouldn't say that about all AI, and 441 00:23:26,640 --> 00:23:29,840 Speaker 2: there are definitely scary things happening with AI, but like this, 442 00:23:30,160 --> 00:23:31,639 Speaker 2: this is not the one to worry about. 443 00:23:32,240 --> 00:23:34,080 Speaker 1: Where can we watch the show when it comes out? 444 00:23:34,320 --> 00:23:38,560 Speaker 2: So keep your eyes on Instagram. But it'll be you know, 445 00:23:38,600 --> 00:23:41,320 Speaker 2: we're really in development. It will take a little bit longer, 446 00:23:42,320 --> 00:23:45,200 Speaker 2: but yeah, it'll all be through Instagram and then from there. 447 00:23:46,119 --> 00:23:49,880 Speaker 1: Okay, cool, Well, thank you so much. I really appreciate 448 00:23:49,880 --> 00:23:51,600 Speaker 1: you taking the time to talk about Tilly. I think 449 00:23:51,600 --> 00:23:53,840 Speaker 1: it's fascinating and I think what you're doing is really 450 00:23:53,880 --> 00:23:57,800 Speaker 1: interesting and definitely a new corner of entertainment, not a 451 00:23:57,840 --> 00:24:01,200 Speaker 1: co optive entertainment. Yes, I think is very mistaken. 452 00:24:01,720 --> 00:24:04,199 Speaker 2: Yeah yeah, thank you for having me. 453 00:24:04,400 --> 00:24:34,520 Speaker 1: Yeah, that's it for this week for tech Stuff, I'm 454 00:24:34,600 --> 00:24:37,679 Speaker 1: Kara Price. This episode was produced by Eliza Dennis and 455 00:24:37,720 --> 00:24:41,400 Speaker 1: Melissa Slaughter. It was executive produced by me Oz Valashan, 456 00:24:41,720 --> 00:24:45,280 Speaker 1: Julian Nutter, and Kate Osborne for Kaleidoscope and Katrina Norvell 457 00:24:45,359 --> 00:24:48,800 Speaker 1: for iHeart Podcasts. Kyle Murdoch mixed this episode, and he 458 00:24:48,840 --> 00:24:52,240 Speaker 1: also wrote our theme song. Please rate, review, and reach 459 00:24:52,280 --> 00:24:55,560 Speaker 1: out to us at tech Stuff podcast at gmail dot com. 460 00:24:55,600 --> 00:24:56,479 Speaker 1: We want to hear from you.