WEBVTT - Smart Talks with IBM: L’Oréal and IBM: AI-Powered Beauty 0:00:00.160 --> 0:00:02.960 Hey everyone, it's Robert and Joe here. Today we've got 0:00:02.960 --> 0:00:05.240 something a little bit different to share with you. It's 0:00:05.280 --> 0:00:09.400 a new season of the Smart Talks with IBM podcast series. 0:00:09.720 --> 0:00:13.520 This season on Smart Talks with IBM, Malcolm Gladwell is back, 0:00:13.520 --> 0:00:15.760 and this time he's taking the show on the road. 0:00:15.880 --> 0:00:19.160 Malcolm is stepping outside the studio to explore how IBM 0:00:19.239 --> 0:00:22.880 clients are using artificial intelligence to solve real world challenges 0:00:23.079 --> 0:00:25.520 and transform the way they do business. 0:00:25.480 --> 0:00:29.560 From accelerating scientific breakthroughs to reimagining education. It's a fresh 0:00:29.600 --> 0:00:33.040 look at innovation in action, where big ideas meet cutting 0:00:33.120 --> 0:00:33.960 edge solutions. 0:00:34.320 --> 0:00:37.160 You'll hear from industry leaders, creative thinkers, and of course 0:00:37.280 --> 0:00:40.920 Malcolm Gladwell himself as he guides you through each story. 0:00:41.440 --> 0:00:44.680 New episodes of Smart Talks with IBM drop every month 0:00:44.680 --> 0:00:48.279 on the iHeartRadio app, Apple Podcasts, or wherever you get 0:00:48.320 --> 0:00:52.240 your podcasts. Learn more at IBM dot com slash smart Talks. 0:00:52.880 --> 0:00:56.240 This is a paid advertisement from IBM. 0:00:56.520 --> 0:01:00.560 To understand why the cosmetics supergiant Lorel Group is teaming 0:01:00.640 --> 0:01:03.640 up with IBM, you must first take a closer look 0:01:03.680 --> 0:01:07.600 at its products. Take lipstick, for example, it's one of 0:01:07.600 --> 0:01:10.800 those things that seems straightforward. A waxy cylinder that you 0:01:10.880 --> 0:01:13.880 rub on your lips to turn them a different color. Easy, right, 0:01:14.440 --> 0:01:18.200 Well maybe not, as my colleague Lucy Sullivan found out 0:01:18.400 --> 0:01:21.480 when I sent her an assignment to Lorel's North America 0:01:21.560 --> 0:01:23.279 Research and Innovation Center. 0:01:24.360 --> 0:01:30.319 All right, I'm reporting live from the Loreal visitor's parking lot. 0:01:31.240 --> 0:01:35.000 Malcolm told me that he would be sending me to Paris, 0:01:35.120 --> 0:01:40.560 France for this Looreal excursion, but instead I am in Clark, 0:01:40.760 --> 0:01:43.680 New Jersey. Pass a lot of strip malls on the 0:01:43.720 --> 0:01:47.119 way here. But to be fair to Clark, New Jersey 0:01:47.200 --> 0:01:50.440 and Lorel, this is a beautiful compound. It kind of 0:01:50.440 --> 0:01:52.720 looks like a spa. 0:01:53.440 --> 0:01:57.000 Lucy went into the center and was blown away. The 0:01:57.040 --> 0:02:02.280 facility houses about six hundred scientists and experts across skincare, makeup, fragrance, 0:02:02.360 --> 0:02:06.440 hair care, innovative packaging, and tech. It is one of 0:02:06.480 --> 0:02:09.760 the largest formulation lab spaces in the industry. It's the 0:02:09.800 --> 0:02:14.240 size of six basketball courts. The reason Loreel's facility is 0:02:14.280 --> 0:02:17.240 so big and has so many people is that everything 0:02:17.280 --> 0:02:20.160 Loriel does to bring a product to market happens here, 0:02:20.760 --> 0:02:25.360 from molecule discovery and product development to consumer testing. The 0:02:25.400 --> 0:02:28.960 center even has its own mini factory. My conception of 0:02:29.000 --> 0:02:32.320 lipstick that it's just a waxy stick was plain wrong. 0:02:33.120 --> 0:02:36.960 Lipstick is a high performance product born from years of research, 0:02:37.320 --> 0:02:43.200 consumer insights, and precision science. Lipstick isn't simple. It's incredibly complex, 0:02:43.800 --> 0:02:46.280 and one of the main reasons it's so complex is 0:02:46.320 --> 0:02:49.160 just a nature of fashion trends. The kind of lipstick 0:02:49.200 --> 0:02:51.639 consumers want is constantly changing. 0:02:52.200 --> 0:02:54.480 A lot of our consumer insights with Lorial is like, 0:02:54.680 --> 0:02:56.520 where are consumers going in the future. 0:02:56.919 --> 0:03:00.440 This is Nadine Gomez. She is vice president for Low's 0:03:00.480 --> 0:03:02.680 research and innovation development team. 0:03:03.120 --> 0:03:05.440 Our chemists are working on five six years down the line. 0:03:05.680 --> 0:03:09.280 We predicted that consumers wanted more of a softer look 0:03:09.280 --> 0:03:10.200 on their lips as well. 0:03:10.320 --> 0:03:11.880 So how do you predict something like that. 0:03:12.400 --> 0:03:16.720 We see slow signals from fashion houses and social media 0:03:16.800 --> 0:03:18.680 and things like that. We kind of see that trend 0:03:18.760 --> 0:03:21.840 evolving a little bit, and then we know at five 0:03:21.919 --> 0:03:23.240 six years it's going to become big. 0:03:24.280 --> 0:03:27.000 Lucy talked with her about the origins of one of 0:03:27.040 --> 0:03:30.600 their products, Mabe Lene matt Ink liquid lipstick. 0:03:31.200 --> 0:03:33.520 Our competitors have two steps. The first step is a 0:03:33.520 --> 0:03:36.160 base code. It's super opique. You get the color and 0:03:36.200 --> 0:03:39.720 you get the madity, but it's very very drying your lips. 0:03:39.800 --> 0:03:41.800 You cannot wear that, honestly more than ten minutes. It 0:03:41.800 --> 0:03:44.560 feels like your lips are like aching at one point. 0:03:44.800 --> 0:03:46.400 So we had to develop a top cot and you'll 0:03:46.400 --> 0:03:48.160 see many of our competitors did the same thing. It's 0:03:48.160 --> 0:03:50.200 like a bomb. You put it on top, it's super comfortable. 0:03:50.840 --> 0:03:54.120 But we also noticed that consumers kind of get tired 0:03:54.320 --> 0:03:56.720 of reapplying a bomb, So we're like, what can we 0:03:56.760 --> 0:03:59.120 do to create this two step into one step? 0:04:00.000 --> 0:04:02.720 Well, Loriel had a challenge, how do you make a 0:04:02.800 --> 0:04:07.360 comfortable liquid Matt lipstick that doesn't require consumers to reapply 0:04:07.520 --> 0:04:11.000 a top layer of bomb. Solving this type of problem 0:04:11.200 --> 0:04:14.040 takes a lot of resources and a lot of expertise, 0:04:14.400 --> 0:04:18.599 and crucially, it takes time. Remember Nadine said that working 0:04:18.640 --> 0:04:21.600 on a breakthrough product such as Matt inc can take 0:04:21.839 --> 0:04:26.520 years before it comes out. But can this process be accelerated, 0:04:26.960 --> 0:04:30.720 taken further, be even more sustainable. That's what IBM and 0:04:30.800 --> 0:04:36.080 Lareel are hoping to find out. My name is Malcolm Gladwell. 0:04:36.200 --> 0:04:39.360 You're listening to the latest episode of Smart Talks with IBM, 0:04:39.680 --> 0:04:42.159 where we offer our listeners a glimpse behind the curtain 0:04:42.200 --> 0:04:46.159 of the world of technology. In our last episode, we 0:04:46.200 --> 0:04:49.599 talked about how an AI assistant created with IBM, Watson 0:04:49.800 --> 0:04:54.560 X helps future teachers practice responsive teaching by simulating classroom 0:04:54.640 --> 0:04:59.320 interactions with students. In this episode, we take you on 0:04:59.360 --> 0:05:03.920 an even more or unexpected journey into the world of cosmetics, 0:05:04.200 --> 0:05:08.679 hair care, skincare, fragrance, makeup and how a custom AI 0:05:08.800 --> 0:05:12.880 model could help Loreel's researchers shape the future of what 0:05:12.920 --> 0:05:22.800 we put on our faces every morning. I want to 0:05:22.800 --> 0:05:25.640 say on lipstick a moment longer to help illustrate what 0:05:25.720 --> 0:05:29.679 goes into Loriel's product development, and let's focus on matt 0:05:29.760 --> 0:05:34.000 Inc lipstick. Loreel wanted to create something that was comfortable 0:05:34.440 --> 0:05:36.039 and could be applied in one step. 0:05:36.680 --> 0:05:38.719 So to go from two step to one step, we 0:05:38.760 --> 0:05:41.160 had to look cross functionally and try to figure out 0:05:41.200 --> 0:05:43.600 what can we bring into the product to make it 0:05:43.640 --> 0:05:46.839 more comfortable and luckily we have many different types of 0:05:46.839 --> 0:05:47.760 products at Lorel. 0:05:48.160 --> 0:05:51.400 That's Alex good, a senior chemist who leads the lip 0:05:51.400 --> 0:05:54.600 products team in North America. She says the trick to 0:05:54.680 --> 0:05:59.000 making matt incwork was finding an elastomer, a substance they 0:05:59.000 --> 0:06:01.000 were already using in foundation. 0:06:01.920 --> 0:06:04.760 We have this elastomer that can give you like more 0:06:04.800 --> 0:06:07.480 comfortable and make it feel like there's like something on 0:06:07.480 --> 0:06:08.440 your lips, like a cushion. 0:06:09.240 --> 0:06:12.760 She handed Lucy two jars. The first jar contained the 0:06:12.800 --> 0:06:15.279 former version of the product that was used in super 0:06:15.320 --> 0:06:18.720 State twenty four. By the way, this is exactly why 0:06:18.720 --> 0:06:22.719 I sent Lucy to the lab in my place the samples. 0:06:22.680 --> 0:06:24.440 And I actually have something for you to try here, 0:06:24.960 --> 0:06:29.039 so you can try this is what was the initial product. 0:06:29.360 --> 0:06:32.920 Okay, so this is like it sort of looks like okay, 0:06:32.920 --> 0:06:35.800 it is. It looks like vacline that has like more 0:06:35.839 --> 0:06:38.679 of a color. It's kind of a beige, looks like 0:06:38.720 --> 0:06:39.239 some skin. 0:06:39.400 --> 0:06:39.799 Okay. 0:06:40.040 --> 0:06:42.400 So this was from the two steps this would go 0:06:42.400 --> 0:06:44.800 on after Oh okay, right. 0:06:46.640 --> 0:06:49.560 Islet Okay, So. 0:06:49.480 --> 0:06:52.720 It feels like very wet. As you can see, it's 0:06:52.760 --> 0:06:55.680 kind of it's gonna absorb into your skin and leave 0:06:56.200 --> 0:06:57.720 and then you're gonna feel the dryness of. 0:06:57.680 --> 0:06:58.680 The product once it's long. 0:06:58.839 --> 0:06:59.080 Okay. 0:06:59.080 --> 0:07:01.479 So we're gonna move from the clay product that you 0:07:01.520 --> 0:07:04.440 have on your hand now to the elastimmer. 0:07:04.000 --> 0:07:07.320 Or you try half hour this jar held the elastomer 0:07:07.520 --> 0:07:10.200 that Lareel had spent years developing in the lab. 0:07:11.240 --> 0:07:15.400 This one is it clear, looks like Aqua four, a 0:07:15.520 --> 0:07:17.320 much player and. 0:07:17.320 --> 0:07:20.520 You can see yeah, physical layer that you're putting on 0:07:20.560 --> 0:07:21.000 your aids. 0:07:21.120 --> 0:07:25.360 Yeah, so that's much thicker. It kind of like clumps together. Yeah, 0:07:26.320 --> 0:07:28.160 it was more of a cloudy it's less shimmery though 0:07:29.400 --> 0:07:30.080 that's intended. 0:07:30.320 --> 0:07:35.200 Yes, So this is a like a powder this dispersed 0:07:35.320 --> 0:07:39.600 in dimethicone and it creates like a comfort on your 0:07:39.600 --> 0:07:41.280 lips a fields, like there's something. 0:07:41.040 --> 0:07:43.360 There for a barrier to keep the. 0:07:42.920 --> 0:07:46.160 Film form on. And that's like the key ingredient that 0:07:46.240 --> 0:07:49.160 came from Foundation that we transferred into lipstick to give 0:07:49.240 --> 0:07:53.000 us this innovative product ahead of the market. Yeah, this 0:07:53.120 --> 0:07:56.040 is what gives it comfort. So the difference between super 0:07:56.040 --> 0:07:58.679 State twenty four and matt Inc is really the comfort. 0:07:58.720 --> 0:08:02.040 They both last a long time, but this matt Inc 0:08:02.080 --> 0:08:04.720 you don't have to apply the bomb over and over again, 0:08:04.840 --> 0:08:07.120 so you can apply matting once for the day and you're. 0:08:07.000 --> 0:08:07.720 Good for it. 0:08:08.400 --> 0:08:11.800 Alex Good is under selling it here, once for the 0:08:11.880 --> 0:08:16.120 day and you're good. That's a liquid lipstick revolution. Literally, 0:08:16.320 --> 0:08:19.720 millions of loreal consumers around the world have worn matt ink. 0:08:19.920 --> 0:08:24.920 It's a blockbuster. It's also a marvel of science. The 0:08:24.960 --> 0:08:28.760 world's first liquid lipstick was developed in the nineteen thirties, 0:08:28.800 --> 0:08:31.160 and it was actually just a stain for your lips, 0:08:31.760 --> 0:08:35.320 barely counts as lipstick. Then came another wave of liquid 0:08:35.360 --> 0:08:38.280 lipstick when they were able to make it matt That 0:08:38.400 --> 0:08:41.400 was a two step version. It felt heavy on your lips. 0:08:41.800 --> 0:08:45.520 You had to keep reapplying the top coat. It was inconvenient. 0:08:46.640 --> 0:08:49.640 Lorel tackled that challenge in the lab, with chemists like 0:08:49.760 --> 0:08:54.200 Alex and the Dean leaving the charge their breakthrough matt Inc. 0:08:55.040 --> 0:08:58.560 But creating matt Inc took a long time, trial and error, 0:08:58.760 --> 0:09:03.160 the hard work of science experimentation. As Nadine told Lucy, 0:09:03.480 --> 0:09:06.199 the lipstick team had to put the new product to 0:09:06.520 --> 0:09:07.560 extensive tests. 0:09:08.040 --> 0:09:10.960 We do a very robustability system here. You know, we 0:09:11.000 --> 0:09:14.360 have color odor appearance. We monitor this in extreme conditions. 0:09:14.720 --> 0:09:17.320 We simulate a forty five degrees celsius and that can 0:09:17.360 --> 0:09:19.520 be something like a three year shelf life. I'm saying 0:09:20.120 --> 0:09:23.040 we simulate your real life product, like if you leave 0:09:23.080 --> 0:09:25.680 your lip gloss in the car in Arizona's one hundred 0:09:25.679 --> 0:09:27.480 and twelve degrees worth three days. Is it still going 0:09:27.520 --> 0:09:30.440 to perform? Is it gonna smell? Is it gonna look granted? 0:09:30.559 --> 0:09:31.600 Is it gonna change colors? 0:09:31.840 --> 0:09:32.600 We do all that? 0:09:33.240 --> 0:09:37.080 See what I mean? Lipstick is complex. Most people would 0:09:37.120 --> 0:09:40.040 never consider it a piece of technology, but one lip 0:09:40.040 --> 0:09:43.520 product has millions of data points. 0:09:43.280 --> 0:09:45.040 So much science behind. And you can see here how 0:09:45.040 --> 0:09:47.360 many scientists we have. You know, some of them have PhDs, 0:09:47.360 --> 0:09:50.559 some of them have master's degrees chemistry, biology, psychology. 0:09:50.600 --> 0:09:54.319 Also, when I first heard about this collaboration between LORI 0:09:54.520 --> 0:09:58.120 L and IBM, I was surprised. I thought, these are 0:09:58.160 --> 0:10:01.520 two very different companies. What do they really have in common? 0:10:02.040 --> 0:10:03.520 You guys? 0:10:03.800 --> 0:10:04.040 Yeah. 0:10:04.200 --> 0:10:06.400 To find out, I went to the IBM research Center 0:10:06.440 --> 0:10:08.880 outside New York City, which I have to say is 0:10:08.920 --> 0:10:11.280 one of the coolest buildings I've ever been in. A 0:10:11.320 --> 0:10:15.920 semi circular modernist masterpiece with a long curving wall of windows, 0:10:16.120 --> 0:10:19.160 looks like something out of a Stanley Kubrick movie. I 0:10:19.240 --> 0:10:21.560 was there to talk with two experts from research and 0:10:21.600 --> 0:10:27.520 innovation at LOREL Metheu Cassier and Gabriel Bertoli metthew IS 0:10:27.640 --> 0:10:31.880 VP for Digital and Transformation, Gabrielle is a Chief Digital 0:10:31.920 --> 0:10:36.440 Transformation Officer for Formulation. These are the people whose jobs 0:10:36.440 --> 0:10:40.120 are to oversee big changes within the company. And Methu 0:10:40.200 --> 0:10:42.000 told me to try on some lipstick. 0:10:42.440 --> 0:10:44.280 I'm gonna make you try this one. 0:10:44.679 --> 0:10:48.760 Okay, this is super stay vinin Final Inc. 0:10:49.000 --> 0:10:50.080 Yeah, so that's a glosse. 0:10:50.160 --> 0:10:51.679 Never in my life put on lipsey. I've no idea 0:10:51.720 --> 0:10:52.120 what I'm doing. 0:10:52.120 --> 0:10:52.800 You don't have to put it. 0:10:52.840 --> 0:10:53.800 You can try it virtually. 0:10:54.679 --> 0:10:58.079 Oh this may not be news to people who buy makeup, 0:10:58.320 --> 0:11:00.760 but it was news to me. You can try on 0:11:00.840 --> 0:11:06.040 loreal products virtually. They call it augmented beauty. Oh my goodness. 0:11:06.720 --> 0:11:09.040 That is the strangest thing I've ever said. I look 0:11:09.120 --> 0:11:10.920 quite fetching, that's the way. 0:11:10.920 --> 0:11:14.680 It amazing. And I can just hit you can choose 0:11:14.679 --> 0:11:15.560 your color absolutely. 0:11:16.280 --> 0:11:18.640 So I'm on a little app it's looking at me 0:11:18.960 --> 0:11:21.280 and it's just showing me exactly how I would look 0:11:21.280 --> 0:11:23.800 with different shades of lipstick. So the odd idea of 0:11:23.840 --> 0:11:25.760 going into a store and trying on each one, you 0:11:25.840 --> 0:11:27.400 cannot do that from home, if you're not even at 0:11:27.440 --> 0:11:27.800 the store. 0:11:28.280 --> 0:11:31.280 Yeah, absolutely, that's all purpose. If you want to manage 0:11:31.320 --> 0:11:33.400 a trend that would go for something more like pitch. 0:11:34.440 --> 0:11:37.520 You think I'm a peach person. I don't know that 0:11:37.600 --> 0:11:40.000 looks I have to say that looks kind of natural. 0:11:40.000 --> 0:11:43.160 It just is enhanced. It's given me a boys share 0:11:43.200 --> 0:11:48.320 I would not otherwise have. This is why Loreel says 0:11:48.360 --> 0:11:52.880 it creates beauty products and beauty experiences. Loriel is a 0:11:52.920 --> 0:11:57.360 beauty tech company. Over the last decade, Loriel has seized 0:11:57.360 --> 0:12:01.760 the power of AI and more recently, generative AI technology 0:12:01.760 --> 0:12:06.040 has become a driving force alongside science and creativity. And 0:12:06.080 --> 0:12:09.959 while some of this digital technology is relatively new, Matthew 0:12:10.200 --> 0:12:13.920 helped me see that IBM and Lorel have always had 0:12:13.920 --> 0:12:14.680 a lot in common. 0:12:15.400 --> 0:12:18.800 So the original creator of Loyal, Jentulier, was a chemist 0:12:18.800 --> 0:12:21.839 in nineteen or nine, so one hundred and sixteen years ago, 0:12:22.480 --> 0:12:25.640 and he created this new air color type for the 0:12:25.720 --> 0:12:29.080 market in France, and then little by little, it has 0:12:29.120 --> 0:12:31.360 been always a very scientific company. So if you look 0:12:31.360 --> 0:12:34.440 a little bit at key facts, we invented sun filters 0:12:34.679 --> 0:12:37.920 in the nineteen thirties, there was a very very big 0:12:37.960 --> 0:12:40.440 milestone where we also invented not only product, but a 0:12:40.480 --> 0:12:45.040 reconstructed skin. So if you look at nineteen seventeen nine, 0:12:45.640 --> 0:12:48.400 we've been the created this reconstructed skin that helped us 0:12:48.440 --> 0:12:50.560 to go out of animal testing very fast and by 0:12:50.600 --> 0:12:53.560 the way, before the law even asked it to cosmetic companies. 0:12:54.160 --> 0:12:57.640 And then more recently, because it's a history of innovation, 0:12:57.760 --> 0:13:00.480 we launch on new molecules like one you can find 0:13:00.480 --> 0:13:03.959 in naoche pose Milibi three, which is really helping people 0:13:04.000 --> 0:13:07.360 to find against some you know, spots they could have 0:13:07.400 --> 0:13:09.920 on their skin. It's all about like big mountation, how 0:13:09.960 --> 0:13:10.599 to regulate it. 0:13:11.320 --> 0:13:14.840 Loreel and IBM were both started in the early twentieth century, 0:13:15.240 --> 0:13:18.840 Loreel in nineteen oh nine and IBM in nineteen eleven. 0:13:19.320 --> 0:13:23.199 Both companies have long standing histories of innovation, of using 0:13:23.240 --> 0:13:26.400 trial and error to improve everything they do. The two 0:13:26.400 --> 0:13:29.199 companies have been doing that in parallel for more than 0:13:29.200 --> 0:13:33.520 a century until recently. When does that start when doll 0:13:33.559 --> 0:13:35.160 Lorel and IBM start working together. 0:13:36.000 --> 0:13:38.880 So we started in twenty twenty three at the end 0:13:38.920 --> 0:13:39.320 of the year. 0:13:39.360 --> 0:13:43.520 But you know, really the discussion is really recent, absolutely absolutely, 0:13:43.880 --> 0:13:46.320 it's really recent in reality. You know, I would say 0:13:47.000 --> 0:13:49.880 the first really interaction happened at the beginning of twenty 0:13:49.920 --> 0:13:50.480 twenty four. 0:13:51.280 --> 0:13:54.720 This is Gabriel Bertoli, who I spoke to alongside Matthew. 0:13:55.440 --> 0:13:58.600 What really played a key role here is we wanted 0:13:58.640 --> 0:14:02.840 to bring from a logic perspective to R and D together, 0:14:04.080 --> 0:14:07.520 which normally, you know companies like us, you just go 0:14:07.640 --> 0:14:10.439 to a provider. You know, it's a customer and the 0:14:10.520 --> 0:14:12.440 supplier and new work they delivered to you. 0:14:12.760 --> 0:14:14.480 Here the concept was totally different. 0:14:15.360 --> 0:14:18.839 Mid two said that the collaboration began with simple conversations. 0:14:19.120 --> 0:14:22.000 So if you look at the way IBM entered into 0:14:22.600 --> 0:14:26.840 Loreal Labs, it's started by interviewing people what would help 0:14:26.880 --> 0:14:27.680 you to do your job? 0:14:27.960 --> 0:14:29.120 What is your business need? 0:14:29.640 --> 0:14:32.120 So it was, by the way, two months ago, a 0:14:32.320 --> 0:14:36.120 long series of interviews and from all the people around 0:14:36.160 --> 0:14:39.160 the world we have in research in Brazil, in India, 0:14:39.440 --> 0:14:44.120 in China, Japan, US, France of course, So we really 0:14:44.160 --> 0:14:45.760 want to make sure that at the end of the day, 0:14:46.120 --> 0:14:48.200 this new model, this new tool that we will give 0:14:48.240 --> 0:14:50.960 to people is really people centrick in the way that 0:14:51.040 --> 0:14:52.480 it selves their daily need. 0:14:53.120 --> 0:14:57.120 More the point, Lourel has leveraged technology for decades and 0:14:57.200 --> 0:15:02.680 accumulated amounted of scientific knowledge, everything from consumer aspirations and 0:15:02.720 --> 0:15:06.040 market trends to the results of all the experiments conducted 0:15:06.120 --> 0:15:10.200 during product development. To which formulations melt in a hot car. 0:15:10.960 --> 0:15:14.000 It's hard to get your head around. Loreal isn't just 0:15:14.040 --> 0:15:18.080 a cosmetics company. It's a beauty data powerhouse. 0:15:19.000 --> 0:15:23.840 If we have sixteen thousand terabat of data coming from 0:15:24.160 --> 0:15:31.760 consumer insights, coming from market research coming from sales, well 0:15:31.880 --> 0:15:36.720 with the new technology, maybe by aligning those two and 0:15:36.840 --> 0:15:39.840 using best in class technology you can solve that problem. 0:15:39.920 --> 0:15:42.640 So you say you have sixteen terabytes of data, put 0:15:42.640 --> 0:15:44.560 that in perspective. How much data is that? 0:15:45.160 --> 0:15:45.480 Give me? 0:15:46.800 --> 0:15:50.760 This is one hundred year of Looreal data based on 0:15:50.960 --> 0:15:55.440 the last forty years of data in the systems. So 0:15:55.720 --> 0:15:57.560 this is really I mean, we're talking about one hundred 0:15:57.640 --> 0:15:59.280 year of data that only Lorial have. 0:16:00.160 --> 0:16:01.760 Take the example of the ellipsis. 0:16:01.840 --> 0:16:04.920 I mean, you know, if ellipsex can be between twenty 0:16:04.960 --> 0:16:08.480 and thirty arrow material, each raw material will have I 0:16:08.480 --> 0:16:14.320 would say ten or fifteen way of doing things. 0:16:15.760 --> 0:16:18.240 Gabrielle is talking about how things used to be done. 0:16:18.560 --> 0:16:22.440 Researchers at Lorel needed roughly twenty five ingredients for a 0:16:22.480 --> 0:16:25.600 new lipstick formulation, but they have to choose from a 0:16:25.600 --> 0:16:29.920 pool of hundreds, if not thousands, of raw materials, and 0:16:30.000 --> 0:16:32.240 even after they settle on the ones they want. They 0:16:32.240 --> 0:16:34.640 have to figure out how much of each ingredient they 0:16:34.680 --> 0:16:39.040 need and in what form, what molecular weight, what combination. 0:16:39.800 --> 0:16:42.760 It's not just a math problem. It's a problem that 0:16:42.840 --> 0:16:50.360 requires balancing multiple perspectives safety, performance, quality, compliance standards, sustainability 0:16:50.440 --> 0:16:54.240 and more. It can take years. But what if you 0:16:54.240 --> 0:16:58.120 could simulate hundreds of cars parked in a sweltering heat. 0:16:58.360 --> 0:17:00.920 What if you could do all those files and errors 0:17:01.240 --> 0:17:05.240 virtually over and over and over again. What if instead 0:17:05.280 --> 0:17:09.080 of mixing materials together by hand, you could ask AI 0:17:09.200 --> 0:17:13.120 to predict what combinations might work best and then try 0:17:13.200 --> 0:17:14.119 those out first. 0:17:14.720 --> 0:17:17.720 This is ten on the power of twenty five. 0:17:18.680 --> 0:17:22.359 This is one hundred billion of years for a human 0:17:23.359 --> 0:17:27.280 to do a change in the formula or the possibility 0:17:27.359 --> 0:17:31.640 they have. You can only do this by using technology, 0:17:32.800 --> 0:17:35.040 power of technology and data that you have. 0:17:35.720 --> 0:17:38.960 This, Matthew says, is where IBM can come in to 0:17:39.080 --> 0:17:43.760 help take things further. Using artificial intelligence. IBM can help 0:17:43.800 --> 0:17:47.480 Lorio create a custom AI model that helps to crunch 0:17:47.560 --> 0:17:51.480 those numbers, to be a companion to the researchers, to 0:17:51.520 --> 0:17:52.600 give them superpowers. 0:17:52.840 --> 0:17:54.840 We don't want to replace the intuition of The sentis 0:17:54.960 --> 0:17:57.919 we just want to make sure that this intuition is 0:17:57.960 --> 0:18:02.280 really augmented by some inclination poor that, as Gabrielle said, 0:18:02.280 --> 0:18:04.840 then does those ten at the poor of twenty five 0:18:05.040 --> 0:18:08.320 solution and seem probably try this one, this one, this one, 0:18:08.520 --> 0:18:11.480 it looks like a better solution and Thentimately that's really 0:18:11.520 --> 0:18:13.639 the decision of the chemist to make it happen. 0:18:16.320 --> 0:18:19.400 Well. To make a predictive AI model that can give 0:18:19.480 --> 0:18:23.480 Lorel researchers those superpowers, you'd need that mountain of data, 0:18:24.040 --> 0:18:28.240 years worth of laboratory testing and all Loreal's data digitized 0:18:28.359 --> 0:18:32.400 and AI ready. You'd need to train artificial intelligence on 0:18:32.600 --> 0:18:35.399 everything the company has already done in order for it 0:18:35.560 --> 0:18:37.080 to predict what it could do. 0:18:37.600 --> 0:18:41.880 Loriial has one hundred years of worse of data, fifty 0:18:42.000 --> 0:18:43.680 years of digitized EGGDA. 0:18:44.400 --> 0:18:48.040 This is Mariam Ashuri, Senior director of Product Management for 0:18:48.160 --> 0:18:51.600 IBM Watson X. Loreal has the data and part of 0:18:51.640 --> 0:18:54.480 IBM's job is to help put that data to work, 0:18:54.960 --> 0:18:59.240 which involves ensuring data quality. Mariam talked about the concept 0:18:59.480 --> 0:19:00.919 of a ready data. 0:19:01.600 --> 0:19:04.439 The sole purpose of this data engineering pipeline is to 0:19:04.560 --> 0:19:09.040 clean the data, and we call them AI ready data 0:19:09.240 --> 0:19:13.040 makes them ready to be consumed by AI, so basically 0:19:13.080 --> 0:19:16.800 looking into biases in the data to fix the distribution, 0:19:17.040 --> 0:19:20.480 looking into guard brains that you're putting into place in 0:19:20.560 --> 0:19:23.000 terms of removing personal information. 0:19:23.840 --> 0:19:26.920 Mariam that explained that a custom model like the one 0:19:27.000 --> 0:19:30.199 IBM is creating with Lorel can be more efficient and 0:19:30.359 --> 0:19:33.640 targeted than the larger general purpose AI models. 0:19:33.840 --> 0:19:37.200 You've heard about large language models. The reason that they 0:19:37.240 --> 0:19:40.800 call them large language model is they are exposed into 0:19:42.200 --> 0:19:46.240 really large amount of data. So the larger the model, 0:19:46.280 --> 0:19:50.240 the more capable the models are, but also the larger 0:19:50.280 --> 0:19:55.040 computed requires that translatestand increase carbon footprint and energy consumption 0:19:55.200 --> 0:19:59.840 that translates, stand increase latency that's your response time, that translates, 0:20:00.040 --> 0:20:05.000 and increase costs. So we started seeing that enterprises started 0:20:05.760 --> 0:20:09.840 grabbing a much smaller model customize it on their proprietary 0:20:09.920 --> 0:20:13.399 data that's the data, their DOMAINO specific data, or the 0:20:13.480 --> 0:20:17.919 data about their users to create something differentiated that is 0:20:18.000 --> 0:20:22.119 applicable to a real world use case but also delivers 0:20:22.200 --> 0:20:25.280 the performance that they needed for a fraction of the costs. 0:20:25.680 --> 0:20:28.600 And that's why there's been a lot of push around 0:20:28.800 --> 0:20:32.800 using custom models versus very large general purpose models. 0:20:33.560 --> 0:20:37.000 So how is a custom model created? Miriam says, you 0:20:37.040 --> 0:20:40.159 start with the base model. Imagine you're buying a car. 0:20:40.640 --> 0:20:42.800 You could get a minivan, or a sedan or a 0:20:42.800 --> 0:20:45.679 sports car, and then you get to customize it. You 0:20:45.680 --> 0:20:48.960 could add a sunroof, leather seats, or a rearview camera. 0:20:49.440 --> 0:20:51.520 Turns out you could do the same thing with your 0:20:51.560 --> 0:20:55.080 AI model. You pick a base and then you customize it. 0:20:55.480 --> 0:20:58.520 You tune it on the data unique to your organization. 0:20:58.960 --> 0:21:02.800 We do believe that one model doesn't fit all use cases. 0:21:03.480 --> 0:21:07.439 You want to truly have access to any model anywhere, 0:21:07.520 --> 0:21:11.920 and by any model anywhere. I really mean any model anywhere, 0:21:12.000 --> 0:21:16.879 open source, proprietary, low call out your machine. Wherever the 0:21:16.920 --> 0:21:20.360 model is. You want to host it yourself, because then 0:21:20.560 --> 0:21:23.399 you would be able to take advantage of the best 0:21:23.440 --> 0:21:26.280 of the technology at any point and pick the right 0:21:26.359 --> 0:21:27.840 model for the target use case. 0:21:28.200 --> 0:21:31.600 So a custom model tuned on Lorel's data would be 0:21:31.680 --> 0:21:35.840 more targeted and efficient than a general purpose model. It 0:21:35.840 --> 0:21:41.040 would understand the researchers world and provide transparency into its workings. 0:21:41.400 --> 0:21:44.240 That's part of the magic. And what could a custom 0:21:44.359 --> 0:21:48.320 AI foundation model do for a company like Lorel. 0:21:48.840 --> 0:21:54.280 Accord with that moder is contain the complexity of the formulation. 0:21:54.960 --> 0:21:59.000 That's game La Moline, an IBM distinguished engineer and one 0:21:59.080 --> 0:22:01.240 of the people working on the AI model. 0:22:01.480 --> 0:22:07.119 And to hype as the formulator to go not only faster, 0:22:07.560 --> 0:22:11.960 but also I would say, be able to include more 0:22:12.000 --> 0:22:17.040 complexity or so in their formulation, more personalization, more certain ability, 0:22:17.640 --> 0:22:21.560 better selected ingredient. So it's really a tool to help 0:22:21.680 --> 0:22:26.679