Wave of interest in generative AI signals new paradigm for tech industry

AI

Artificial intelligence and machine learning may prove to be much more significant than a wave of suddenly popular technologies. They may be the entire ocean.

“It’s not even a wave; it’s a tidal wave or maybe even the tide itself,” said Clem Delangue (pictured, left), co-founder and chief executive officer of Hugging Face Inc. “AI and machine learning are not something you add to your products; it’s very much a new paradigm to do all technology. Now AI can generate text, it can generate image, it can describe your image, it can do so many new things that weren’t possible before. I really think that in maybe three to five years, there’s not going to be any company not using AI.”

Delangue spoke with theCUBE industry analyst John Furrier at the AWS Startup Showcase: “Top Startups Building Generative AI on AWS” event, during an exclusive “Power Panel” discussion on theCUBE, SiliconANGLE Media’s livestreaming studio. He was joined by Ankur Mehrotra (center), general manager of machine learning at Amazon Web Services Inc., and Ori Goshen (right), co-founder and co-CEO at AI21 Labs, and they discussed the rapidly changing landscape for generative AI. (* Disclosure below.)

Coding with machine learning

Delangue’s company, Hugging Face, is representative of AI’s evolution. The startup has created a platform, similar to GitHub, where developers can host open-source AI models and datasets for training.

“Software engineering as we know it today is a subset of machine learning instead of the other way around,” Delangue said. “Now we are realizing you can actually code with machine learning. Every software engineer can leverage machine learning through open models, and there might be more of these people in a couple of years than there are software engineers today.”

For major technology players such as AWS, the startup ecosystem sprouting around generative AI presents an opportunity. The cloud giant has recently expanded its relationship with Hugging Face, which will build the next version of its language model on AWS.

For its part, AWS has built distributed machine learning capabilities into its SageMaker cloud platform offering and previewed CodeWhisperer technology to create computer code from natural language.

“CodeWhisperer is an AWS service that we announced a few months ago,” Mehrotra said. “It’s a coding companion-as-a-service, which uses generative AI models underneath. We’ve also used machine learning to improve user experiences across different Amazon products, whether its Alexa or Amazon.com.”

Recognizing limitations

Improving the user experience will depend a great deal on being able to train AI models that will add value for a multitude of purposes. The release of ChatGPT into general use has elevated interest in uses for generative AI, perhaps without fully appreciating how new the technology really is, according to Goshen.

“What we’re seeing with these large language models or generative models is that they’re really good at creating stuff, but it’s also important to recognize their limitations,” Goshen said. “They are not as good at reasoning and logic. The next phase is how to make these systems more reliable.”

There are an expanding number of tools becoming readily available for enterprise use in the generative AI field. Goshen’s company has been building state of the art language models with a focus on developing scalable and efficient applications.

“People are starting to now imagine the possibilities and think of their strategy for adopting generative AI technology,” Goshen said. “It probably doesn’t make sense for every company to create their own foundation models. You can basically start by using an existing foundation model, either open source or proprietary, and start deploying it for your needs.”

The innovation engine for generative AI has picked up new momentum in recent months, and code hosting companies like Hugging Face have a front row seat to what is in the pipeline. ControlNet, a new model for implementing fine control in platforms such as image generator Stable Diffusion, is having a major impact in the AI community, according to Delangue, who is also intrigued by the recent movement to develop new tools for interoperability.

“More companies are getting into this capability of chaining different models and different APIs,” he noted. “That’s a very interesting development because it creates new capabilities, possibilities and functionalities that weren’t possible before. Having more interoperable machine learning will open a lot of interesting things in the future.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of the AWS Startup Showcase: “Top Startups Building Generative AI on AWS” event:

(* Disclosure: This is an unsponsored editorial segment. However, theCUBE is a paid media partner for the “Top Startups Building Generative AI on AWS” event. Amazon Web Services and other sponsors of theCUBE’s event coverage have no editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE