AI seems to be everywhere these days.

Since ChatGPT was first released less than four years ago, millions of people have leveraged the power of generative AI in their personal and professional lives.

Individuals use popular AI platforms like OpenAI’s ChatGPT, Anthropic’s Claude, and Google’s Gemini daily to plan vacations, draft term papers, and update their resumes. The technology has become ubiquitous in business, as well, across every industry and sector.

The risk management profession isn’t immune from the artificial intelligence takeover.

According to a recent LightBox survey of environmental professionals across the U.S., nearly half (47%) regularly use AI in their work. Additionally, 3 in 10 (31%) are testing the use of AI for environmental site assessments (ESAs) or related tasks. A strong minority (43%) rely on publicly available large language models (LLMs) like Claude or ChatGPT, whereas a smaller percentage (21%) have begun using custom enterprise or private AI models.

Despite the industry’s rapid adoption of AI for certain tasks, its growing use is raising some serious red flags. The top concerns cited by survey respondents include output accuracy and reliability (83%), legal liability (61%), and data security and privacy (46%).

In the banking and risk management industries, AI clearly has great potential to deliver operational efficiencies and productivity gains. But is it ready for its closeup?

Don’t cut corners with AI

Today, it’s tempting to “outsource” certain tasks to AI. The technology has the capability of absorbing massive amounts of information contained in large document files like ESAs, appraisals, and borrower financials. But without proper oversight and validation, this can be a recipe for disaster.

Recently, a busy lender was confronted with a large influx of SBA deals. To help manage this uptick in volume, they decided to use generative AI to compare the environmental reports against the SBA’s latest Standard Operating Procedures (SOP). Satisfied with the results, they submitted those environmental reports to the SBA as part of the loan packages for approval.

Unfortunately, the SBA rejected several of the environmental reports due to lack of compliance with the SOP. The lender reached out to ORMS in a panic, and we were able to help them properly evaluate the environmental reports and repackage the loans. In the end, the SBA did ultimately approve most of the deals.

Why you can’t rely on AI for environmental due diligence

As the above example demonstrates, it doesn’t pay to cut corners with AI. When it comes to prudent risk management, nothing replaces human expertise and experience. Here’s why:

1. AI is like an intern: Even the most sophisticated LLMs aren’t ready to serve your business needs right out of the box. Think of it like hiring a college intern: They may be smart, energetic, and eager to learn. But it takes time to be trained on the ins and outs of complex, niche businesses like CRE lending and environmental risk management. Often, an intern won’t be providing true business value until they’re ready to return to school. AI models are a lot like that eager intern. They also need to be trained, a process that takes time, patient oversight, and an instructor with expert knowledge of the business. You can’t rush the process.
2. AI can hallucinate: Generative AI models are notorious for providing inaccurate information, seemingly concocted out of thin air. Their answers can also be seriously biased, based on the databases they are trained on, as well as user preferences and points of view (which AI can mirror extremely well). LLM outputs may look good and sound competent, but if they’re not backed up by real, human experience and expertise, they can’t be relied on as being accurate or up to date.
3. AI is only as good as the prompts: The age-old adage in computer science is “garbage in, garbage out.” Even with technology as sophisticated as today’s LLMs, the quality of the output is heavily dependent on the quality of the inputs—the prompts entered by actual users. If an inexperienced staff member asks the wrong questions or only provides partial information, the answers will likely not be accurate or particularly useful.
4. AI may quarterback you to the ten-yard line…but it won’t score for you. There are certain areas where AI can prove very useful. It can scour the Internet for information about an unfamiliar topic and provide a concise summary to help you get up to speed quickly. But until it’s verified, the information presented may be inaccurate, irrelevant to your needs, or woefully out of date. AI is no substitute for real-world human experience and practical hands-on knowledge. To use a football analogy, AI may get you into the “red zone,” but it takes a human expert to carry you over the goal line.
5. Risk management is a relationship business: Banking is an industry built on relationships, and at ORMS we have always been focused on developing deep, meaningful relationships with our clients, colleagues, industry partners, trade associations, and regulators. That’s the only way we know how to operate, and our relationships with well-respected players have allowed us to learn, respond, grow, and flourish.

As smart as it is, AI will never be able to replicate the authentic human relationships we have with our clients, and you have with your borrowers. People are at the heart of our industry, and that will never change.

People are central to all we do

AI is a powerful technology with great potential to deliver cost savings and operational efficiencies to our industry. But it’s essential that we keep humans connected to the process.

At ORMS, we have always focused on serving our clients with speed, accuracy, expertise, and reliability. We continuously evaluate the latest technology and tools to support our mission, and we will continue exploring AI and other emerging technologies in order to keep providing the best service in the business.

AI isn’t a replacement for human experience, expert knowledge, and personal service. At ORMS, we offer relationship-based outsourced risk management services designed to meet your unique needs. Visit orms.com to learn more about our risk management services and how we can support your growth.