Google, AWS AI leaders discuss promise, perils of generative AI

Generative AI isn’t some overhyped short-lived trend; it could transform the world as we know it. That’s according to executives from Google and Amazon.

In a wide-ranging conversation with VentureBeat founder Matt Marshall at the VentureBeat Transform 2023 conference today, Gerrit Kazmaier, VP data and analytics at Google Cloud, and Matt Wood, VP of product at Amazon Web Services (AWS), discussed the opportunities and risks of generative AI.

Wood called gen AI “the single largest, most transformative technology which is going to change how we interact with data and information and each other, probably since the advent of the very earliest web browser.”

Data is the foundation for generative AI success

Kazmaier is also extremely enthusiastic about gen AI as a way for organizations to unlock the value of data in ways that were not easy or even possible before. He noted that generative AI models are now available to anyone.

“You can go to AWS or you can go to Google and you’re gonna get more or less the same [capabilities], which is great because it [generative AI] has some innate capabilities that have not been seen before,” Kazmaier said.

>>Follow all our VentureBeat Transform 2023 coverage<<

He noted that generative AI sparks creativity and allows organizations to do “cool new things,” but emphasized that data is the foundation for training AI — and for corporate differentiation.

“The data that you have, how you curate it and how you manage that, interconnected with large language models (LLMs), is, I think, the true leverage function in this entire journey,” Kazmaier said. “As a data guy, this is just a fantastic moment because it will allow us to activate way more data in many more business processes.”

Bucketloads of opportunities for generative AI

At the Transform session both Wood and Kazmaier identified multiple “buckets” of opportunities for enterprises to benefit from generative AI.

Wood outlined use cases for creating content, new personalization options for search and customer experience, helping experts work more efficiently, and creating opportunities for collaborative problem solving.

Kazmaier talked about productivity, where generative AI can have a profound impact. One of the key step changes in productivity for him is enabling non-coders to generate code and applications. AWS’s CodeWhisperer application helps with code development, while Google has several efforts including its Generative AI Studio.

Data is also a key opportunity for gen AI, specifically unlocking the value of unstructured data, according to Kazmaier. Working with unstructured data is often difficult; generative AI simplifies it.

Kazmaier also cited the ability to build entirely new products as a huge opportunity for generative AI. Wood echoed and expanded upon this. He said that while large organizations like Google and Amazon have been able to benefit from machine learning (ML) for years, generative AI makes ML easier to use and accessible to anyone.

“I think that this is the single largest step forward in the ease of use and accessibility of machine learning, ever,” Wood said. “I suspect in the next year, five years, three years, who knows, six months, I think we’re going to see hundreds of new organizations emerge that are starting to apply these techniques, defining products in all of these different industries, in ways that are going to appear magical.”

Hallucination is a big risk, but so too is lack of planning

As for the risks of generative AI, both execs cited hallucination as an obvious concern. But Kazmaier noted that in his view the biggest risk is underestimating the long-term impact of generative AI, and thinking of it as just an incremental step.

Another potential risk can also come from not being open to experimentation with all the different approaches and models that exist today and will tomorrow, whether or not those models are open or closed.

“It’s super-early, to be candid. If this is a marathon, I don’t think we’re three steps into the marathon. And I don’t think you pick a winner three steps into the marathon,” Wood said.