Generative AI: Get used to it—and use with caution
While artificial intelligence has made remarkable progress, human judgment still reigns.
Although we’ve yet to wrap up the third quarter, it seems safe to declare generative artificial intelligence (or “Gen AI”) the No. 1 buzzword of 2023. This advancement in search technology is most often associated with ChatGPT, but there’s no shortage of players in the space.
Bard, the brand name for Google’s “AI experiment,” has rapidly gained prominence. Microsoft has built Gen AI capabilities into its Bing search engine, and Open AI, the owner of ChatGPT, also offers an AI tool cleverly named DALL-E that creates images.
It may be helpful to think of Gen AI as the logical progression of the chatbot and virtual assistant technology that made great strides in sophistication and acceptance over recent years. Rather than interpreting a flexible set of inquiries and delivering a canned and/or formatted response, Gen AI scours all available online data to craft a customized response.
Software variations are also being developed limiting the scope of data being searched to proprietary information residing behind a company’s firewall. This may ultimately prove more suitable for many commercial purposes.
It’s easy to envision Gen AI technology turbocharging efficiency in many credit union functions, from member inquiries to exception handling, dispute resolution, and fraud detection/prevention.
Many credit union leaders have expressed interest in better understanding Gen AI’s potential. To date, however, such efforts remain in the exploratory phase, including developing policies governing its current use. This is as it should be.
I’m aware of credit union CEOs and marketing leaders informally using Bard and ChatGPT to create initial drafts of employee newsletters, promotional campaigns, and website content.
Gen AI already excels at these types of tasks, offering a starting point for a human expert to refine and personalize. These early adopters find it to be a valuable time saver.
Under no circumstances should Gen AI output be shared with members without detailed human review at this stage, however. This applies to both members engaging directly with the engine, as well as employees leveraging Gen AI to craft outbound communications.
In many ways, the technology’s hype has gotten ahead of its reality. Current iterations are particularly adept at delivering conversational, persuasive responses.
Unfortunately, the accuracy of those responses has not quite kept pace. As is too often the case, the details of the terms and conditions tend to be overlooked.
As Google clearly states on the tool’s landing page, “Bard is an experiment and may give inaccurate or inappropriate responses. You can help make Bard better by leaving feedback.”
Microsoft’s FAQ offers a similar caution: “Bing will sometimes misrepresent the information it finds, and you may see responses that sound convincing but are incomplete, inaccurate, or inappropriate. Use your own judgment and double check the facts before making decisions.”
Make no mistake: The progress made in Gen AI over the past year is nothing short of remarkable. Experts agree the path to commercial viability is years ahead of schedule, and these systems will only get smarter with broader use beyond laboratory conditions.
From a scientific standpoint, every misstep is a “learning opportunity.” In a field like financial services where trust and accuracy are paramount, those missteps cannot come in the form of member interactions.
Gen AI is coming—and it’s coming fast. Now is the time for sandboxing so credit unions are comfortable with the technology and can quickly capitalize on its vast potential when the time is right. For now, human judgment still reigns.
GLEN SARVADY is managing principal at 154 Advisors.