Hyperscaler talks up its history in AI and machine learning with an eye on the technology becoming more pervasive across all aspects of business.

Joao-Pierre S. Ruth, Senior Editor

July 30, 2023

5 Min Read
Swami Sivasubramanian, VP of databases, analytics, and machine learning at AWS, delivers his keynote at AWS Summit NYC
Swami Sivasubramanian spoke at the AWS Summit New York about cloud's connection to the rise of AI.

AI needs to live somewhere and hyperscaler AWS seems to want the cloud to be a home for new developments in this space.

Last week’s AWS Summit New York keynote brought out Swami Sivasubramanian, vice president of databases, analytics, and machine learning at AWS, who talked up the slow burn in AI’s popularity to its current rockstar status.

“This technology has reached its tipping point -- the convergence of technological progress and the value of what it can accomplish today,” he said. “That’s because we have a massive proliferation of data and the availability of extremely scalable compute infrastructure.”

AWS conducts these conferences in New York annually, trotting out its latest lineup of products as well as offering a taste of its upcoming plans.

This latest keynote saw disruption from the audience almost from the start. Shouting protesters interrupted Sivasubramanian and other speakers five times that morning, decrying AWS for providing technology to Israel’s government and military, which continues to be embroiled in the Israeli-Palestinian conflict.

This echoed prior protests in July 2022 that disrupted the keynote at the AWS Summit New York, though that time the hot-button issues included AWS providing technology to US Immigration and Customs Enforcement (ICE) and other federal agencies.

Each time protesters raised their voices, they were escorted out and the keynote continued.

Large language models such as generative AI, Sivasubramanian said, can be trained on massive amounts of unlabeled data, used for a wide variety of tasks, and adapted through a customization process called fine-tuning for particular applications. “The ability to easily customize and pre-train models through fine-tuning is an absolute game changer,” he said. “It is substantially faster. It requires a lot less computational time and a lot less data for fine-tuning than spending months of creating a task for a specific machine learning model.”

Based on Sivasubramanian’s keynote, AWS aims to interject itself even more in the AI world by providing services for organizations to work with foundational models to build and scale generative AI applications tailored to their data and needs. He also said customer data is encrypted along with other steps to keep proprietary information from leaking out. “Your data is never used to train the original base model.”

The keynote included some customer testimonials from Gabrielle Tao, senior vice president of product management with Salesforce, and Lindsay Silver, senior vice president of data and commercial technology with Fox Corporation.

Tao said Salesforce leveraged AWS services to create machine learning, generative AI, and to develop a solution to use data to create insights and deliver personalized experiences at hyperscale. “That is certainly not easy today,” she said. “On our planet, there are more devices than there are humans. As more and more disconnected data grows, it becomes harder and harder to connect with customers.”

Silver said media company Fox uses data and AI to further its operations. “We turned a corner in the last couple years. For the first time ever, large language models and GANS [generative adversarial networks] have allowed us to take data and go straight from observation back to a product that we can act on,” he said. “It means generating content, generating responses in chat, generating images that we can take directly back and act on.”

This is a boon to Fox’s business, Silver said. The company draws from data sources across its brands and feeds it into an infrastructure largely built on AWS, he said. Fox also has its own technology in the mix such as a heads-up interface used by sportscasters that uses AI to provide insights and to create graphics with relevant information on the fly that overlays sports broadcasts. “That deep integration is what allows Fox to innovate to create content that really matters to our audiences,” Silver said.

An underpinning element that could help companies derive more value from generative AI is their data, Sivasubramanian said. “While [foundational models] are incredibly powerful out of the box, to be truly useful to your organization, they need access to the right data sources. Your data is your differentiator for gen AI.”

The current attention shining on AI has brought flurries of companies introducing new platforms that make use of it, but this seems to be familiar territory for AWS. After the keynote, Vasi Philomin, vice president of generative AI with AWS, spoke with InformationWeek and reiterated that his company has more than 20 years of history working with AI and machine learning. “What we’re really good at is taking AI and ML and applying it at scale to real-world business problems.”

Recommendations for similar products that shoppers see on Amazon come from machine learning, Philomin said. Further, users of Amazon’s Alexa virtual assistant have billions of interactions with the technology each week. “That’s all machine learning applied behind the scenes,” he said.

The breakout popularity of generative AI, Philomin said, stems from the availability of compute, data, and fundamental improvements in machine learning. “These models require a lot of compute for training and for inference,” he said. “With cloud vendors, we have the ability for anybody to spin up 1,000 machines.” Philomin also said the cloud is a natural place for data to accumulate. “When you have a lot of data, you can gain a lot of value from it -- insights,” he said. “Generative AI allows you to do that.”

Updates to the Transformer architecture model and growth in the size of the models mean the models have emergent behavior, Philomin said. “They need very limited data to learn a new task. You can teach it in a very quick way to learn something new. Most of it is built into the model already -- you just have to unlock it with additional data.”

AWS plans to continue its investment in AI resources, Philomin said, aiming to adapt to the evolving landscape alongside its customers. “We believe that generative AI is going to be transforming every function in every business in every industry,” he said.

What to Read Next:

Cloud Equity Group’s Sean Frank Talks AI Mingling with the Cloud

Leveraging AI, Machine Learning to Enhance Cloud Interoperability

Tapping AI to Alleviate Cloud Spend

About the Author(s)

Joao-Pierre S. Ruth

Senior Editor

Joao-Pierre S. Ruth covers tech policy, including ethics, privacy, legislation, and risk; fintech; code strategy; and cloud & edge computing for InformationWeek. He has been a journalist for more than 25 years, reporting on business and technology first in New Jersey, then covering the New York tech startup community, and later as a freelancer for such outlets as TheStreet, Investopedia, and Street Fight. Follow him on Twitter: @jpruth.

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like

More Insights