“It’s actually quite difficult to build a really good generative AI application” – Amazon CEO outlines his AI vision and challenges
- Amazon CEO Andy Jassy speaks during the opening keynote of AWS re:Invent 2024
- Jassy outlines Amazon’s AI work and what it has learned
- Amazon priority is “technology that really matters to customers”
Amazon CEO Andy Jassy has outlined some of the company’s biggest challenges when it comes to using AI.
As part of a guest appearance in the opening keynote of AWS re:Invent 2024, Jassy highlighted the usefulness of “practical AI” in helping his customers.
Jassy outlined multiple examples of how the e-commerce giant is using AI internally, and also shared some key learnings from the company’s experiences with the technology – as well as unveiling the new foundational Nova models.
Andy Jassy and Amazon AI
“We’ve been using AI extensively within the company for the last 25 years,” Jassy noted, “but the way we think about technology – and this applies to AI as well – is that we don’t use it because we think it’s cool. , we use it because we’re trying to solve customer problems.”
“That’s why when we talk about AI, we typically say less about beating the world’s best chess player – and more about giving you better recommendations, or equipping our pickers in our fulfillment centers. . or for our Just Walk Out technology”
“We prioritize technology that we think really matters to customers, and with the explosion of generative AI over the past few years, we’ve taken the same approach: there’s a ton of innovation, but what we’re trying to do is solve problems for you – what we consider practical AI.”
During his time on stage, Jassy highlighted a number of examples of Amazon’s use of generative AI, from customer service to seller page creation and inventory management.
Jassy also focused on Rufus, its generative AI chatbot, which is said to be getting better and smarter at recommending products to customers around the world, and a focus on robotics – which is becoming much more important in its fulfillment centers and ensuring a huge increase in efficiency and efficiency. productivity.
But he noted that this work is not without challenges – even with the wealth of resources and knowledge that Amazon possesses.
“It’s actually quite difficult to build a really good generative AI application. You need a good model, but you also need to have the right guardrails, deliver the right message fluidly, and you need to have the right user interface .”
“We keep learning the same lesson over and over again: there will never be one instrument to rule the world.”
“In AWS, we’re going to give you the very best combination… as we always do.”