Japanese banking giant Mizuho has announced it will roll Microsoft’s Azure OpenAI service out to 45,000 employees.
Mizuho announced its consideration of Microsoft’s Azure OpenAI in April, intending to use the technology to accelerate its digital transformation. The company plans to use the technology for coding, drafting documentation, and fielding employee queries regarding procedures and systems.
The institution was also at pains to insist that the environment would be secure and that, as well as Microsoft’s own security and compliance tools, it would be building “an appropriate management system”.
According to reports this morning, the bank will be rolling out access to Azure OpenAI to 45,000 employees this week. In an interview, Toshitake Ushiwatari, general manager of the bank’s planning department, described the decision as “like poking a beehive”, referring to the excitement among employees keen to submit their ideas for consideration.
AI inferencing with AMD EPYC™ processors
Providing an excellent platform for CPU-based AI inferencing
The decision comes at a time when US lawmakers are planning to set limits on staff use of ChatGPT. Following concerns about the implications of the technology, reports have emerged of US lawmakers raising guardrails around the use of ChatGPT.
In a memo obtained by Axios, House of Representatives Chief Administrative Officer, Catherine L. Szpindor, instructed staffers that only ChatGPT Plus was to be used (the product requires a subscription and incorporates privacy features) and that authorization was only provisional at this stage.
Szpindor’s memo went on to require that privacy settings be enabled, the tool should not be integrated into regular workflows and that sensitive data should not be copy and pasted into it.
Why would you want to restrict the use of ChatGPT?
Generative AI tools have proven themselves adept at “generating” content based on a large data set. However, there are two main reasons for restricting the usage of something like ChatGPT.
The first is privacy. What goes into ChatGPT can be added to its data set and so might end up getting reproduced in response to another user’s query. Hence Szpindor’s requirement that sensitive data not be pasted into the tool.
The second is the danger of a generative AI tool outputting something that looks believable but is actually incorrect.
What is the difference between Azure OpenAI and ChatGPT?
It is important not to conflate Azure OpenAI and ChatGPT in this context. While the latter is designed for conversational AI applications, the former makes use of Azure’s security capabilities and can integrate with other Azure services as well as including support for ChatGPT itself.
Azure OpenAI also follows Microsoft’s Responsible AI principles which, according to the company, means that AI systems should be secure and respect privacy, be reliable and safe, and treat all people fairly.
The company additionally states “People should be accountable for AI systems.”
What does this mean for business?
Generative AI is undoubtedly concentrating minds. According to an IBM study released today, half of CEOs surveyed reported that they are already integrating generative AI into digital products and services.
However, 57% were concerned about data security and 48% expressed concerns over accuracy.
Significantly, 75% of the CEOs surveyed felt that the organization with the most advanced generative AI would have a competitive advantage.
This chimes with Ushiwatari’s closing comment: “This is something we have to do, otherwise, we get left behind.”
