Rakuten & OpenAI: What’s next in GenAI?

OpenAI has officially landed in Japan. A strategic AI partner of Rakuten from fall 2023, the company behind ChatGPT has set up shop in Tokyo – their first office in Asia.

Rakuten and OpenAI marked the occasion by hosting an event at Rakuten Crimson House, which kicked off with a lecture from OpenAI Solutions Engineer Evan Weiss for Rakuten’s e-commerce and travel consultants. During the talk, Weiss shared some of the many innovative ways that AI tools could supercharge workflows and allow them to further empower merchants and accommodation providers on Rakuten’s market-leading e-commerce and travel platforms.

The lecture was followed by an insightful conversation between OpenAI COO Brad Lightcap and Rakuten Group Chief AI & Data Officer Ting Cai, who discussed the new office opening, answered questions about how Rakuten and OpenAI can do even more to support clients and partners, and even speculated on future AI trends and technologies.

Next stop: Tokyo

“Japan has long been a fascination for us,” Lightcap told the Rakuten audience. “It’s a market that has a rich history of adopting new technology, of innovating on critical technology and critical manufacturing.”

Lightcap noted the incredible demand the company has seen from the Japanese market. “We have two million people here using ChatGPT every week, which is astounding,” he revealed. “We wanted to be really close to where we felt we had an opportunity to work very closely with potential customers, with users, with developers.”

OpenAI’s expansion is a continuation of its astonishing growth. Rakuten’s Cai noted that the company had just a few hundred employees as the partnership kicked off. Just a few months later, that number is over a thousand.

OpenAI is more than just research

Lightcap credits OpenAI’s rapid growth to its unique combination of research and engineering.

“Everyone thinks of OpenAI as a research organization. We are a research organization, but we’re a research and engineering organization. We learned early on the value of really, really strong engineering.”

The company’s engineering expertise has allowed them to build systems that can keep up with the sheer scale and complexity of the models they research.

One major turning point for the company was its decision to focus not on small models for specific tasks, but a single, massive model to do everything.

“We started the company early on with a very core mission of being able to build AGI (artificial general intelligence) and being able to ensure that the benefits of AGI were very broadly distributed, and that the technology was built and deployed very safely.”

A generally intelligent system should be capable of learning, allowing it to take on new tasks without prior knowledge.

“If you get to that point, then there’s kind of no reason to have these hyper-specialized small models,” Lightcap reasoned. “Using a thousand models that are each for a specific task is somewhat contrary to this idea of a generally intelligent model that can learn how to do many tasks.”

Future ChatGPT will handle tasks end to end

Lightcap made clear that OpenAI was not about to sit back and rest on its already extraordinary laurels.

“Even today, we still actually consider the technology to be in its infancy,” he said. “I don’t think we’ve really seen fully what this technology is capable of, and I think the rate at which it improves will surprise us.”

How could ChatGPT improve even further?

“Ultimately, the models have to be able to go off and solve more complex tasks,” he continued. “If you think about it, today you can use ChatGPT, for example, to solve fairly basic things. It can help you summarize text, it can help with translation, it can help you write something… but those are very narrow tasks and they only represent small percentages of what comprises a total workflow.”

Lightcap wants to see ChatGPT evolve into something that can handle tasks from end to end.

“The work that we do every day is actually much more complex,” he noted. “So how do you improve the model’s ability to be assistive across that entire workflow and to start helping you think through how to integrate each of those tasks to come up with new knowledge, new ideas, and solve complex problems?”

Empowering clients and partners with novel use cases

ChatGPT and other AI tools are already being utilized in innovative ways, including within Rakuten. Rakuten has been proactively pursuing an initiative called AI-nization, through which AI tools are not only being adopted internally, but used to empower partners, such as the merchants of Rakuten Ichiba and hotels, traditional Japanese inns and other accommodation facilities on Rakuten Travel.

“Both Rakuten and Open AI, we share the mission of bringing value to society. In order for millions of people to benefit from the latest models, it takes engineering, but also more importantly, tools.”

Ting Cai, Chief AI & Data Officer of Rakuten Group

“There’s this diversity of applications from personal consumer use cases all the way through to very real business use cases,” Lightcap added. “I think one that I really love is how Rakuten has thought a lot about building tools for merchants. There’s something very special about aligning the capabilities that the technology has to enable people, to make them more productive, make them more efficient.”

Rakuten’s work empowering merchants harmonizes well with OpenAI’s mission to use AI tools to empower society, Lightcap noted.

“Merchants have a million things to do every day. Running a business is really hard, but it’s really important that merchants exist. They’re critical to the fabric of our society,” he stressed.

“So how do you marry tools and the capabilities that AI brings to the needs that merchants have, when it comes to running a business, and operating a business at growing scale? We think that’s a tremendous opportunity and actually is one of my favorite use cases.”

“You don’t have to know how to write a line of code, but you basically have a 24-hour assistant that you can access virtually for free who can write you as much code as you can dream up, and that’s only going to get better, and it’s going to get better in a way that’s accessible.”

Brad Lightcap, COO of OpenAI

With AI tools available to the entire world, Lightcap is eager to see how people adapt.

“There’s going to be this tremendously large, overarching theme of growth and opportunity. It will change fundamentally how we do work and how we grow companies and build products. I think we’ll adapt pretty well, and overall I’m very positive and optimistic about the future.”

Taking the AI plunge

Rakuten’s Chief AI & Data Officer Cai has a clear vision for how Rakuten wants to employ AI tech going forward. “At Rakuten, we believe we should augment human creativity with the power of AI – humans and AI working together.”

“We want to think big, and we also want to stay very concrete. Especially for Rakuten, we are here to deliver concrete value to our business partners and our customers,” he continued. “Rakuten also very much values entrepreneurship and innovation, and we want to create value for society. In that regard, I feel like OpenAI and Rakuten have a lot of things in common.”

OpenAI has also been prolific with its partnerships, maintaining mutually beneficial relationships to empower their partners while improving their own tools. “We really are a company that can bring people into the process of developing the technology, into the process of distributing its benefits,” Lightcap remarked.

Lightcap’s advice for companies looking to take the AI plunge?

“Don’t start necessarily trying to solve every problem in your business with AI. Start to solve a few that really matter and be very focused in how you do that,” Lightcap said. “I think one of the most important things that any company can do is just giving their people access to the best product, and starting to just let them experiment and figure out where it can add value.”