Minimizing AI Risks in the AEC Industry: Embracing Responsible Practices
Summary
In the AEC industry, companies often wrestle with the decision to develop their own AI solutions versus using generic (off-the-shelf) ones. There are several compelling reasons to pursue proprietary AI development. For one, each company has its distinct expertise and nuances that off-the-shelf AI solutions might not fully capture. Building AI solutions in-house allows businesses to leverage their unique strengths and knowledge, ensuring that AI resonates with their specific requirements. In addition, AEC companies possess unique data sets from their projects, and AI built on specific data is more aligned with the company’s needs than those based on generic data sets. Finally, data confidentiality is a pressing concern. While AI models such as ChatGPT are trained on public datasets, there’s an inherent risk of proprietary data blending into a wider knowledge pool when a company uses third-party solutions, potentially diminishing a company’s competitive advantage. For all of these reasons, in-house AI experimentation and development can potentially protect and amplify a company’s unique value.
Introduction
Over the past few years, especially recently, there’s been a noticeable change in the AI landscape. Surprisingly, as AI capabilities have expanded exponentially, development costs have decreased. Let me share a snapshot of this trend from my own experience: About three years ago, I led a team of six AI engineers to develop an AI solution for design. That POC project took us a full year and came with a $500k price tag. Fast forward a year later, the pace has dramatically shifted. With only two AI engineers, we have developed an AI quality assessment solution, detecting defects in building facades in just a week at a cost of $50K. More recently, with just a single software engineer, we created specialized AI for detecting concrete cracks in merely an hour, costing $5k. Obviously, while our ability to harness AI is growing, the time and cost of development are becoming more efficient and affordable. Thus, more AEC companies are interested in taking advantage of this tremendous opportunity to experiment with and build AI solutions to increase their capacity and profit margins. Besides cost and time, however, why are they running AI experiments in-house?
To answer this question, let’s start by looking at an example from the manufacturing industry. Tesla stands out in the automotive industry, not only because of its electric vehicles but also because of its advanced use of data and AI in manufacturing. At its Fremont factory, AI-powered systems streamline production lines, detect defects, and predict maintenance needs, ensuring smooth operations and high-quality output. What truly sets Tesla apart, however, is its commitment to developing proprietary AI technologies. While many car manufacturers rely on tech vendors for AI solutions, Tesla recognizes the advantages of in-house development. This approach allows Tesla to tailor AI tools to its precise needs, ensuring a perfect fit for its unique manufacturing challenges and data intricacies. And it’s paying off: While Toyota, a long-standing automotive giant, has seen a declining operating margin, Tesla’s has surged, approaching those of luxury legends such as Ferrari. This achievement underscores how Tesla’s strategy has propelled the company ahead of established car industry leaders (Toyota vs. Tesla.)
In the AEC industry, companies experimenting with their own AI solutions can leverage their unique expertise and data sets and protect their data confidentiality.
Experimenting with your own AI Solutions
For many AEC companies, the choice to pursue their own AI initiatives and development, like Tesla’s approach, is influenced by multiple strategic factors.
Unique expertise and know-how: Your company, like every other company in the AEC industry, has its own distinct blend of know-how and expertise, a unique DNA that sets you apart from the competition. Generic AI solutions, while competent, often overlook these idiosyncrasies, usually offering a one-size-fits-all approach. However, you can develop your unique AI solution by drawing from the rich reservoir of your company’s specific know-how and expertise. By doing so, your AI solutions will resonate with your company’s unique strengths, expertise, and nuances. As a result, you can preserve the very essence that makes your business stand out in this industry.
Unique data sets: Every construction project, design blueprint, or engineering challenge has a wealth of unique data points and patterns that can be understood only through your company’s proprietary data. When your company uses its own data to fuel its AI development, the AI solution will be tailored to your business and process needs, reflecting your company’s history, expertise, and intricate industry challenges. These AI solutions, built on your company’s data, not vast datasets from many companies, lead to more integrated and directly relevant outcomes for your company’s unique needs rather than just providing generic insights.
Data confidentiality: Using ChatGPT by OpenAI as an illustration, the AI is trained on extensive public datasets, not on private or proprietary information. Nevertheless, many AEC companies raise concerns: Does OpenAI or similar entities store or utilize my company-specific data to enhance their AI models? While these AI solutions might anonymize data, they still amalgamate knowledge from various sources. Hence, the insights derived are not solely based on your unique data but rather a blend of data from diverse sources that can potentially make certain insights available to all, including competitors. As a result, the distinctive value of your data and processes might merge into a broader, shared repository, potentially diluting your competitive edge. Thus, protecting your data confidentiality is critical to preserving your competitive advantage in today’s AI-driven landscape.
If you’re in the AEC industry and you’re relying solely on off-the-shelf AI solutions, you might be leaving a lot on the table. Your company has a unique DNA with insights and data that off-the-shelf AI solutions can’t possibly get. Sure, tools such as ChatGPT are cool and efficient, but when it comes to truly capturing the essence of your business, nothing beats a tailored AI solution. Not only does it resonate better with your business needs and unique expertise, but it also keeps your data under wraps, ensuring that your competitive edge doesn’t blur into the vast AI landscape. Let your AI be as distinctive as your company’s blueprint.
