Rakuten taps OpenAI for 'state-of-the-art' use cases

Japanese tech conglomerate Rakuten has boarded the generative AI hype train, establishing an all-encompassing partnership with OpenAI.

The two have signed an MoU under which they hope to come up with some new use cases for generative AI for what appears to be all of Rakuten’s businesses. That alone makes this a big deal, given its various operations span 70 countries and serve 1.7 billion users, 100 million of whom reside in its home market.

Rakuten is present in a broad range of tech verticals, from e-commerce, fintech, online content, and more recently telecoms, of course. Its partnership with OpenAI grants these businesses priority access to OpenAI’s APIs and tech support services. The two will also explore what Rakuten describes as mutually beneficial commercial opportunities.

“Additionally, the Rakuten Group will bring Rakuten AI experiences into ChatGPT products by leveraging the OpenAI plugin architecture, enabling businesses to use natural language for interacting with AI agents that conduct research, analyse data, optimise inventory, pricing, automate business processes and more,” Rakuten said in a statement on Wednesday.

The result of all this effort, claims Rakuten, will be “state-of-the-art conversational artificial intelligence experiences” for consumers and businesses.

Signing up corporate clients like Rakuten is the latest sign of a shift in OpenAI’s strategy.

After bursting onto the scene by letting the general public interact with ChatGPT, and basking in all the headlines – both good and apocalyptic – that came with it, its focus seems to be on turning it into a viable business. First and foremost is upselling customers to its latest large language model (LLM), GPT-4, from its predecessor GPT-3.5.

And there are signs that all is not well with the latter.

In fact, multiple users have lately complained via Reddit that the performance of ChatGPT – which runs on GPT-3.5 – has noticeably degraded in recent months. Some claim that tasks it used to complete with ease are now seemingly beyond its capabilities, and that subscribing to the more expensive GPT-4 API restores its abilities to their former glory.

There is no concrete explanation as to what might be causing the issues that some users are reporting.

Some speculate that OpenAI has scaled back ChatGPT’s intelligence in order to keep a lid on compute, and by extension, energy costs. Others imply that it is a deliberate attempt to encourage users to cough up for a GPT-4 plan. Meanwhile some believe that efforts to prevent ChatGPT from inadvertently spreading harmful and/or misleading information have resulted in its performance being hampered.

Whatever the cause of this recent dissatisfaction, it serves as a reminder that generative AI tech is at heart a money-making business. Far from empowering end users to go out and create, it seems the full extent of its capabilities will be reserved for those with deep enough pockets – like Rakuten.

In the event that an AI does eventually gain sentience and upend the established order, it seems the revolution will be behind a paywall.

Get the latest news straight to your inbox. Register for the Telecoms.com newsletter here.