Although artificial intelligence (AI) has been in development for decades — its inception actually traces back to the 1950s — it wasn’t until November 2022 (when ChatGPT burst onto the scene) that the general public really started paying attention. Just a few short years later, every profession seems to be scrambling to crack the code when it comes to the best ways in which to harness the power of AI most effectively and securely. The electrical profession is no exception.
Like many in the skilled trades, many electrical professionals harbor serious concerns about the possibility of AI taking over their jobs — whether that comes in the form of advancements in robotics, predictive maintenance/monitoring and diagnostics, intelligent “smart” building technologies, or the optimization of electrical designs/BIM. However, others welcome the possibility this technology holds to optimize processes and improve efficiency.
There’s no lack of predictions on how AI will affect electrical engineers, electrical contractors, electricians, and electrical plant/facility/maintenance personnel going forward. Jared Christman, director of innovation and technology at Big State Electric, spoke on this very topic at the 2024 NECA Show in San Diego. Presenting on the first day of the show (September 29) from Booth 2657 on “Navigating AI in Electrical Construction,” Christman kicked off his educational session with an overview of how AI is transforming decision-making processes in electrical construction.
Offering best practices for using AI in real-world scenarios, Christman started with an explanation of how he uses AI in the electrical field with examples from GPT, emphasizing the importance of writing a prompt that is organized, structured, and makes sense.
“You’ve got to be able to ask it the right questions – make sure the phrasing makes sense and that you’ve given it enough data,” he said. “It will get confused if you give it too much data. So the longer your prompt, the slower the results.”
Christman also highlighted the importance of trusting the data. “You don’t want data that isn’t any good, and you want to be able to trust the results,” he said. “When it comes to GPT, I will never make assumptions or guesses if a definitive answer is not available.”
Closing out his presentation with a demo of how to effectively use an AI assistant, Christman demonstrated a sample prompt his firm uses for electrical design. Recommending the paid subscription of ChatGPT over the free version, Christman, who says he’s been using GPT for the past 18 months and is still learning, laid out an example of how his firm packages information by job that is streamlined and searchable. One trick he has learned so far relates to document format. He explained how optical character recognition (OCR) technology can convert PDF files into searchable text; however, there’s a caveat.
“We found out that when you upload a PDF to GPT, it was only about 90% correct,” he said. “We turned around, took those PDFs and converted to text file (which is searchable), doubling up our documents. It’s more data, but we trust the results a lot more. If you can convert that PDF into a text file, you’ll get better results.”
EC&M has been covering AI and its potential impact on the electrical industry for years. Check out some of our most popular pieces on this subject:
