Should you use ChatGPT (or Gemeni, Claude etc) in business? Yes, but……what are the issues, when do you need to have an Enterprise solution, a RAG solution?
Over the past two years, there have been incredible advances in generative AI tools like ChatGPT that have seen millions of working professionals go from first touch to daily indispensable tool in a very short time.
They have become part of our personal routine, giving us all super powers that were once unthinkable. And for so many, these tools have become just as indispensable in our professional lives as well.
If these tools are so useful and indispensable, why do so many companies have prohibitions or restrictions on the use of some of these amazing tools?
The key issue isn’t so much the tools, it’s the data.
ChatGPT, Gemini, Claude, and other major large language models (LLMs) are trained on enormous amounts of data that is public content. This content contains a tremendous collection of human knowledge from sources like Wikipedia, scientific journals, published books and newspaper articles. It also includes content such as blogs, Reddit, social media influencers, published works of fiction, opinion pieces, company PR, political disinformation campaigns, public falsehoods, special interest groups and more. That’s a wide range of content. Some of it reliable, some of it questionable, and some of it outright biased or inaccurate.
For personal use, this mix might not matter much. But in a business environment, the stakes are higher. You have to ask:
Is the information credible?
Is it factual?
Do I want my company’s decisions—or customer communications—influenced by unverified or biased sources?
There are significant issues with copyright. While some of it is covered by “Fair Use” which at least allows you to use this data for research purposes, that doesn’t cover commercial purposes. In fact, there are huge sets of this knowledge in use by these LLM’s that are protected by copyright and were not paid for or secured by the LLM companies. Some of this is being actively negotiated or in courts.
As an example there are several hundred thousand published books contained in the LLM without proper approval or consent as well as entire libraries of news from major publishers. This fundamentally means that you really can’t use the outputs of these tools in your published commercial works without being in violation of the content creators copyright.
Then there’s the issue of security. When you share your work content into public LLMs, that data could potentially be used to further train the model which might benefit others, including competitors.
That’s why most companies of any scale are cautious, or even prohibitive, about using public generative AI tools in a business setting.
So what’s the business alternative?
More and more companies are adopting Retrieval-Augmented Generation (RAG) solutions to get the benefits of generative AI while maintaining control over their data.
The key difference? RAG solutions only use information that is owned or licensed by the business. This means that the answers provided by the system are derived from trusted and private sources versus the wild west of the internet and public sources. It also means that the business has copyright to the content for commercial use within the system. RAG solutions are typically designed not to share data with the public LLMs, ensure only approved users can access information, and they honor internal permission structures so sensitive data stays protected. These steps help overcome all of the previously mentioned challenges.
At Capacity, we’ve been working with AI in data retrieval for nearly a decade, our Answer Engine® (previously known as Lucy) is an example of a RAG solution to support secure knowledge access. Our Answer Engine is used by companies of all sizes, from Fortune 100 giants to mid-market leaders to help them receive the greatest benefits of the evolution of AI while doing so in a way that delivers trusted content, without copyright restrictions and in conformance with the company’s security standards. We have partnered with them through every phase from requirements, strategy, implementation, onboarding, support and continued evolution..
We’ve seen firsthand how much confusion exists between public AI tools like ChatGPT and business-grade solutions like ours. That’s why education is so important.
To that end, we need to help users absolutely understand when they can use public tools versus when they must use the tools of the company. Further, the users need to understand the “why” so they can fully grasp the reasons and ramifications behind the right tool at the right time for the right job.