WEBVTT 00:00:00.100 --> 00:00:02.669 Ashesh and Vikas, thank you so much for joining us. 00:00:02.936 --> 00:00:04.738 Ashesh, this first question is for you. 00:00:05.071 --> 00:00:08.641 Why is making AI easier and more accessible important? 00:00:09.142 --> 00:00:10.143 Thanks for having me here. 00:00:10.143 --> 00:00:14.147 So I fundamentally believe that AI is a once-in-a-generation 00:00:14.147 --> 00:00:19.019 technology that will improve the productivity that humans can have. 00:00:19.619 --> 00:00:22.956 We'll see the impact of AI in a variety of use cases and a 00:00:22.956 --> 00:00:24.591 variety of industries, right? 00:00:24.591 --> 00:00:29.863 Whether it's agents in a contact center, whether it's students, 00:00:29.863 --> 00:00:31.831 we've already seen them, you know, use it quite 00:00:31.831 --> 00:00:35.068 prolifically, whether it's, you know, retailers making better 00:00:35.068 --> 00:00:36.870 recommendations, whether it's financial services providing 00:00:36.870 --> 00:00:39.272 better services to the customers, or in fact, 00:00:39.272 --> 00:00:42.075 governments providing better access to the citizens. 00:00:42.075 --> 00:00:44.477 So I think the impact of AI is profound. 00:00:44.477 --> 00:00:45.612 It'll be widespread. 00:00:45.678 --> 00:00:48.581 And so it's really important for all of us to figure out ways to 00:00:48.581 --> 00:00:51.518 embrace it to increase our productivity and operate innovation. 00:00:51.951 --> 00:00:55.855 So Vikas, how can companies best leverage open source AI to 00:00:55.855 --> 00:00:56.790 capture value? 00:00:57.590 --> 00:01:00.126 Lisa, I think the important word is choice. 00:01:00.927 --> 00:01:03.730 And you know, Ashesh was just mentioning retailers making 00:01:03.730 --> 00:01:04.497 recommendations. 00:01:05.131 --> 00:01:08.435 I had a discussion with the head of AI for a retailer, and he was 00:01:08.435 --> 00:01:11.104 describing to me—this is a company that uses a lot of 00:01:11.104 --> 00:01:12.305 different AI technology. 00:01:12.572 --> 00:01:15.475 And he said for the simplest of things where, you know, if a 00:01:15.475 --> 00:01:18.378 customer is buying eggs and they need to get a recommendation 00:01:18.378 --> 00:01:21.414 that they should buy milk, they don't need a trillion parameter 00:01:21.414 --> 00:01:24.150 model for that, you know, an open source model, a smaller 00:01:24.150 --> 00:01:25.585 model is more than sufficient. 00:01:25.819 --> 00:01:27.754 And so it's really about choice. 00:01:27.754 --> 00:01:30.757 It's about different situations and different technologies 00:01:30.757 --> 00:01:33.059 that'll be appropriate for those situations. 00:01:33.193 --> 00:01:35.695 And open source, we think is actually a very important part 00:01:35.695 --> 00:01:36.029 of that. 00:01:36.463 --> 00:01:40.567 So for both of you, beyond technology, what can enterprises 00:01:40.567 --> 00:01:41.868 do to create value? 00:01:43.269 --> 00:01:48.041 At BCG, we believe that getting value from AI at scale is an 00:01:48.041 --> 00:01:50.343 organization-wide commitment. 00:01:50.376 --> 00:01:54.013 It's not the responsibility of, you know, a specific individual 00:01:54.013 --> 00:01:55.348 or a specific function. 00:01:55.548 --> 00:01:58.751 And what it really requires is the right level of executive 00:01:58.751 --> 00:02:01.654 sponsorship, the right governance in place, the right 00:02:01.654 --> 00:02:04.657 approach to data—because we're talking, you know, about 00:02:04.657 --> 00:02:07.494 exabytes of data that people are looking to analyze. 00:02:07.794 --> 00:02:09.229 And it's about change management. 00:02:09.295 --> 00:02:12.899 It's making sure that the right policies, procedures, 00:02:12.899 --> 00:02:17.570 enablement, upskilling is in place so that the organization 00:02:17.570 --> 00:02:21.107 can actually get the value from the amazing technology that we 00:02:21.107 --> 00:02:22.775 have now available to us. 00:02:23.610 --> 00:02:25.345 I couldn't agree more with Vikas, right? 00:02:25.345 --> 00:02:26.613 I think choice will be critical. 00:02:27.113 --> 00:02:30.483 I think open source will be very important, which isn't strange 00:02:30.483 --> 00:02:33.820 for me to say. But I think it's also important to really think 00:02:33.820 --> 00:02:37.157 about model sizes and using the right size model for the right 00:02:37.157 --> 00:02:37.624 use case. 00:02:37.624 --> 00:02:40.894 So, you know, we think that's going to be important, but 00:02:40.894 --> 00:02:43.496 coupled with that will be efficiency, right? 00:02:43.496 --> 00:02:46.065 How cost-effective are you delivering? 00:02:46.266 --> 00:02:49.402 What's the appropriate ratio with regard to price and the 00:02:49.402 --> 00:02:51.271 value being delivered against it? 00:02:51.271 --> 00:02:53.506 So I think all those factors will play a really important role. 00:02:53.873 --> 00:02:57.143 But I'll add maybe one more thing to the point you were 00:02:57.143 --> 00:02:59.345 making about organizations. We think it's really, really 00:02:59.345 --> 00:03:03.950 important to enable and skill the users of corporations and 00:03:03.950 --> 00:03:05.952 anyone who's interacting with AI. 00:03:06.186 --> 00:03:09.289 So just like we did with cloud and made everyone comfortable 00:03:09.289 --> 00:03:12.325 with regard to application services around the data center 00:03:12.325 --> 00:03:15.361 and the ones they take advantage of in the cloud, we'll go 00:03:15.361 --> 00:03:17.130 through a similar journey with AI. 00:03:17.130 --> 00:03:21.100 So it's really, really exciting to see and we're looking forward 00:03:21.100 --> 00:03:22.468 to being a part of it. 00:03:22.769 --> 00:03:24.137 Thank you both so much for being here. 00:03:24.571 --> 00:03:24.971 Thank you. 00:03:25.271 --> 00:03:25.605 Thank you.