• last month

Category

People
Transcript
00:00I think there's two really exciting areas right now.
00:02One is to look at the traditional, you know,
00:06machine learning and data science that's been happening
00:09in ad tech and personalization and recommendation
00:12for a long time.
00:13And look at how you can bring
00:14accelerated computing to that.
00:17What that does for the industry
00:19is increases the efficiency of the computing,
00:22which ultimately can bring down costs,
00:23bring down energy consumption,
00:25and all the things that can grow exponentially
00:29if not looked at and innovated around.
00:31And, you know, we've done this in lots of other industries,
00:34and we've been working really hard
00:35to develop the tools needed
00:37to move that traditional data science workflow
00:40to NVIDIA's accelerated computing platform.
00:44That also brings with it the ability
00:47to have faster data flywheels.
00:50And the concept of the data flywheel isn't a new one,
00:53and I think the advertising industry
00:54is one of the biggest data flywheels of any sector, really.
00:58You can also look at financial services
01:00and healthcare as similar, you know,
01:01volumes of cyclical data information.
01:05As we increase the efficiency
01:07and the speed of processing data,
01:10you can create more intelligence from that data faster.
01:13And this is where we come into the world of generative AI
01:15and the tools that we're building
01:18to enable enterprises,
01:20both the end users of advertising,
01:22so the CMO groups in the industry,
01:25as well as the agencies and the ad tech companies,
01:29to be able to derive huge amounts of value
01:31from that data faster.
01:34That will then enable the ability
01:36to create things like fine-tuned language models
01:39much, much quicker with high levels of accuracy.
01:42That can then drive faster recommendation,
01:44really good consumer interaction.
01:47You know, NVIDIA's been building these toolkits
01:50and SDKs and platforms for a long time.
01:53Our AI enterprise stack is like one of the most,
01:57shall we say, like enterprise-hardened ways
02:00to deploy your own AI journey and your own AI tools.
02:03One of the really important things
02:05in that group of technologies,
02:06especially in this industry,
02:08is the ability to guardrail
02:10those language model interactions,
02:12especially when you're dealing with either consumers
02:14or your own internal teams,
02:15building co-pilots for them or AI agents,
02:18because you need those tools to stay on track, right?
02:22So within that guardrailing toolkit,
02:24you can be really specific about what they'll talk about,
02:27what they'll interact with, and what they'll do.
02:29♪♪

Recommended