AI applications take centre stage at the AWS Public Sector Summit in Washington, D.C.

When it comes to tech conferences this year, generative artificial intelligence (AI) is the buzzword one can hardly escape. It was no different at the Amazon Web Services (AWS) Public Sector Summit in Washington DC last week, hosted by a company long at the forefront of AI development, from e-commerce recommendation algorithms to Alexa, the virtual assistant.

At the two-day summit held on June 7 and 8, AWS speakers straddled a balancing act, highlighting the upcoming generative AI applications AWS offers for public sector customers, while cautioning audience members that the AI umbrella extends far beyond the generative. 

“I do encourage people to zoom out and think holistically about the opportunities across the spectrum. In the public sector, AI offers great opportunities to improve the way that government works, improve citizen services, or the way that healthcare professionals deliver care,” said Max Peterson, Vice President of worldwide public sector at AWS, in an interview with GovInsider.

Yet, it was clear that the summit aimed to crystallise AWS’ approach to generative AI, from democratising such applications to customers and exploring the value of these applications to the public sector.

Democratising generative AI innovations

In his keynote address, Peterson highlighted AWS’s efforts to democratise generative AI, from the open-source release of a top-ranking large language model developed with the United Arab Emirates, to a new service which allows AWS customers to build and scale generative AI applications quickly.

The Technology Innovation Institute, the applied research pillar of Abu Dhabi’s Advanced Technology Research Council, recently launched a foundational large language model (LLM), Falcon-40B, which currently ranks as the top performing open source LLM.

The Falcon-40B was trained on Amazon SageMaker, an AWS service that enables developers to build, train, and deploy machine learning models. 

As of June 7, the model has been made available to developers via Amazon SageMaker JumpStart, which they can use to build their own customised generative AI applications, without having to start from scratch.

Peterson also shared on the recent release of Amazon Bedrock, which allows AWS customers to build and scale generative AI models quickly, securely and easily. The service enables users to tap on five foundational models, such as text and image generation models, to create and deploy their own customised applications, privately trained on their own data.

“Customers can bring their own enterprise or government data to the model in a private virtual cloud to train the model and deliver enhanced outcomes for their specific use case,” explains Peterson to GovInsider. 

For instance, a tax agency could securely train such a pre-existing model on the country’s tax regulations and deploy it as a chatbot that can provide tax advice to citizens.

Third, Peterson highlighted Amazon CodeWhisperer, a service that allows developers to generate the code best suited for a task and identify vulnerabilities within code in real-time, improving productivity by 30 per cent. 

Finally, he announced that AWS will be providing an experience-based acceleration programme to train customers in generative AI skills, from developing conversational citizen experiences to implementing best practices.

Generative AI gaining traction in the American public sector

Beyond AWS’ recent generative AI announcements, speakers at a panel on generative AI discussed the ways in which generative AI applications have begun to take root within the United States’ federal government. 

AI has rapidly progressed from being able to provide insights and predict future trends towards generative capabilities and functioning as virtual assistants to end users, said Bratin Saha, Vice President, ML and Engines, AWS.

For instance, the Department of Veterans Affair is currently using generative AI to better communicate the benefits veterans are eligible for, he said. They are currently working with tax authorities in Canada to roll out similar services.

Kathleen Featheringham, Vice President of AI and Machine Learning at system integrator Maximus, shared that the company is working with the government to identify possible AI applications within federal services. She highlighted that this can range from generating first drafts of job descriptions to recommending secure code.

The panel also played host to Republican Senator Mike Rounds, co-chair of the Senate AI Caucus and Representative Ted Lieu, Vice Chair of the House Democratic Caucus, in a rare show of bipartisan agreement.

“There’s nothing particularly partisan about AI,” said Ted Lieu, who had previously developed the first piece of federal legislation written by ChatGPT to draw attention to the need to regulate the technology. Lieu said that he is currently working on creating a bipartisan commission that can make recommendations on AI regulation in the US.

The two politicians affirmed the need for bipartisan cooperation in the realm of AI regulation, while noting the importance of understanding the technology better before introducing significant legislation.

AI applications deployed in extreme environments

Finally, speakers highlighted how AWS has begun deploying AI applications in extreme environments, from outer space to the battlefield.

In his sit-down with GovInsider, Peterson shared that he sees tremendous opportunities for cloud providers to support governments in arenas traditionally considered the domain of large government projects, such as space technology.

For instance, AWS is currently running AI applications on an orbiting satellite.

The models allowed the satellites to quickly identify and remove images that would not be of use, so that ground station employees would only receive the most useful images, he explained. This has increased efficiency by 30 per cent by eliminating a third of the pictures which were unusable.

He highlighted that such edge computing applications could be of use to people working in a range of extreme environments, including “environmental researchers, soldiers, and disaster response workers”. 

At the summit, Peterson also shared that AWS will be working with the Department of Defense on a Joint Warfighting Cloud Capability Contract, which enables the department to tap on cloud services from the enterprise level to the tactical edge, including in the most extreme of military environments. 

The department will be able to tap on AWS’ Snow devices, which provides compute, storage, and other services in remote and limited environments. The new AWS Snowblade device, announced at the summit, has been designed to meet US military requirements for ruggedised equipment.