How AWS is capitalizing on generative AI

Amazon Web Services announced their large scale simulation system is now available for government clients. AWS Worldwide Public Sector Vice President Max Peterson sits down with Yahoo Finance's Dan Howley to expand on how AWS is capitalizing on generative AI.

Video transcript

DIANE KING HALL: Amazon Web Services with a big announcement this morning. Their large scale simulation system is now available for government clients, with compliance requirements introduced last year. AWS SimSpace Weaver allows for testing of large-scale scenarios and using those results to plan and train for real world responses.

Now, governments at various levels can test using their sensitive data. Big news for disaster and emergency scenario planning. Our own Dan Howley spoke with Max Peterson, Vise President of Worldwide Public Sector at AWS about the system. Take a listen.

MAX PETERSON: Everybody's going to be excited to join us here at the Worldwide Public Sector D.C. Summit. And part of it is because of generative AI. Amazon has made a number of different announcements. Customers are using-- used to using Amazon machine learning and generative AI and they may not even realize it, where you do personalized shopping, whether you talk to Alexa, whether you go through Amazon Go and check out.

So at the summit here today, we're going to be talking about the latest AWS services to help customers build and deploy generative AI solutions. Bedrock provides a simple, easy way to let people deploy gen AI solutions using pre-trained models. The AWS Cloud Infrastructure is the most cost-effective way for you to train and run these types of models, with Inferentia and with Trainium.

And finally, we've got customers that are building actual large foundation models, not just using them, on AWS. And so we're excited to be able to talk about the Technology Innovation Institute, which is the technology and research arm of the Abu Dhabi government, that trained the Falcon 40B large language model on AWS SageMaker.

And they have surpassed the sort of performance, training, and cost barriers that many others haven't. And we're excited to acknowledge and congratulate them on the fact that they've been recognized.

On the Hugging Face large language model index as the world's number one LLM right now.

DAN HOWLEY: So, you know, talk about that the generative AI aspect and the training and deployment. You know, that's obviously something that's very costly. We've discussed it with, you know, regards to OpenAI Google, Microsoft. What does this do then for companies that, you know, outside of your main competition? What does it do for companies then when you're talking about the training, the deployment of generative AI of these large language models.

MAX PETERSON: Many customers are just not going to have the time or the resources or the inclination to train their own large language models, their own foundation models. And that's where we believe that Amazon Bedrock is going to be the way for more customers to be able to actually put enterprise and government class AI and generative AI applications into use.

Bedrock makes it easy for customers to use pre-trained models, like Amazon's Titan One or Titan Two models, or like models from Stability AI. And these pre-trained models have a known data provenance. They have known uses.

They're ready to be used on AWS in a customer's private virtual cloud with the customer's confidential and oftentimes private information to be able to train and enhance these already good models to use them for your purposes. To be able to have a much better conversational discussion with the citizen about government services.

Or to be able to use in a case of an assistant or a copilot and quickly summarize a vast amount of information so that the person can make better and faster informed decision, whether you're a law enforcement official or whether you're a defense analyst.

DAN HOWLEY: One of the announcements you guys have is something called SlimSpace Weaver. It's basically a way to provide governments with security and compliance requirements. It meets those. But it's meant to be kind of a simulation and training platform. Can you kind of explain what that is, the idea of digital twins kind of the government, public sector getting in on that. And what does that do for them?

MAX PETERSON: Yeah. One of the most dynamic areas is with governments who realize now that they need to have much more comprehensive planning exercises.

So, for instance, with digital twins and using SimSpace Weaver, governments can model thousands or millions of different variables and create really complex simulations in 2 or 3D to allow them to do things like complex traffic flows, you know, in cities around the world and making it easier to be able to model what happens with a construction change or with a road closure. And how do you wind up using the simulated digital twin to create a better outcome in a live environment.

Many of these things like emergency planning can be conducted much more capably using a simulated environment, a digital twin because you would never create these kind of large-scale disaster scenarios, you know, in reality. But you still need to be able to prepare for them.

So what's exciting about the SimSpace Weaver announcement at the summit is that it's now available in the US GovCloud, which is the cloud that is used for security and compliance by national, state, and local government customers and companies that are building solutions for these types of customers. So AWS SimSpace Weaver is now available in the AWS GovCloud.

DAN HOWLEY: So you have SimSpace Weaver, you have generative AI, I guess, just as a final question. We talk about generative AI a lot, where, you know, it comes to businesses, it comes to, you know, creativity. Where does it go as far as the public sector goes? Where does it benefit the public sector?

MAX PETERSON: Well, as I talked about just a second ago, there are a lot of applications where government or health care organizations or educational institutions can use generative AI to deliver better service for citizens, a more immersive conversational environment, for instance, or better outcomes for students to be able to help them quickly find the right resources and aid their studies or better outcomes in health care by being able to quickly summarize vast amounts of information for that clinician or the medical practitioner so that they can focus on the patient and not focus on all of the information gathering and assessment pieces that tools like machine learning AI and generative AI are going to help customers with.

DIANE KING HALL: All right. That was our own Dan Howley, who's speaking with Max Peterson, vise president of Worldwide Public sector at AWS about the system.