As artificial intelligence (AI) continues to evolve, it is increasingly being used in the collision repair industry for tasks such as damage assessment, estimating repair costs, and parts ordering and inventory management. However, many industry experts believe human intelligence (HI) will remain essential, particularly in complex decision-making, customer interactions and quality control.
Josh McFarlin, president and COO of AirPro Diagnostics, and Tim Ronak, senior services consultant at AkzoNobel, discussed the role of AI and HI in collision repair during a presentation held at the SEMA Show in November 2024 as part of the Society of Collison Repair Specialists (SCRS) Repairer Driven Education series.
Looking to the future, they predict there will be a hybrid approach, where AI augments the expertise of skilled technicians, leading to more efficient processes without replacing the invaluable role of human judgment.
McFarlin began the presentation by describing AI and HI. He shared two definitions of AI, one from IBM and the other based on asking an AI chatbot to define itself.
According to IBM, AI is technology that enables computers and machines to simulate human learning, comprehension, problem solving, decision-making, creativity and autonomy.
AI described itself as a branch of computer science that uses algorithms, data and computational power to create machines that can perform tasks that typically require human intelligence.
McFarlin defined HI as the ability to learn from experience, adapt to new situations, understand and handle abstract concepts and use knowledge to manipulate one’s environment.
“Human intelligence, at its best, can be summarized by an unofficial slogan for the U.S. Marines: improvise, adapt and overcome,” he said. “That's what makes human intelligence, in my opinion, different from AI.”
McFarlin shared examples of AI being used in the automotive industry, such as self-driving cars and virtual personal assistants. He explained AI can currently be categorized into narrow and general.
“Narrow AI is designed to perform a single task or very limited set of tasks and lacks the ability to adapt to new situations outside of its programmed domain,” he said. “General AI aims to replicate human-like intelligence and will perform a wide variety of tasks across different domains and is largely theoretical.”
The best-known example of an AI application, according to McFarlin, is likely ChatGPT (Generative, Pre-trained transformer), an AI machine trained on a database that runs on a foundational large language model (LLM). Many in the collision industry are familiar with ChatGPT, including McFarlin and Ronak, who generated images used for their presentation by visiting an AI platform online. They also used AI to help build the presentation.
McFarlin shared some of the overall benefits of using AI in business, such as automating repetitive tasks to reduce human errors, improve productivity and ensure accuracy. AI has also been proven to make data processing and analysis more efficient.
He used the example of monitoring thousands of sales calls. While AI can do this almost instantaneously, McFarlin said it would take a significant amount of time for a human resource to listen to just a sample of those calls.
Another advantage of using AI is the ability to recognize patterns in large datasets.
“Finding patterns in large datasets, especially when they're really large data sets, is not an easy thing to accomplish,” said McFarlin. “It takes time. Being able to automate that process and have it present you results almost instantaneously is very helpful.”
For those interested in exploring how to use AI in their businesses, McFarlin suggests asking AI how to accomplish X, Y or Z and providing options to achieve that result.
“What ChatGPT and other products like it are doing is crawling the web for all the information available, and they're serving that back up to you almost instantaneously as a result so that it's faster than Google,” he explained.
McFarlin then discussed some strengths of HI, such as the importance of empathy and emotional understanding, which is often referred to as emotional intelligence.
He talked about how human intelligence provides creativity and innovation with the ability to think outside the box, understand context and moral implications with ethical decision-making, solve complex problems, and adapt to changing situations.
“Unless you really define the parameters, AI will provide the most direct route from A to B,” noted McFarlin. “It's not necessarily going to tell you whether that route is the best way to handle an employee, so you have to give it some guardrails.”
McFarlin also mentioned some limitations of AI and HI. He said AI lacks common sense, does not have the ability to identify ethical dilemmas, and has biases in algorithms depending on how it is embedded in the data. HI limitations include cognitive biases, emotional responses, indecisiveness, speed, cost and availability.
After discussing the pros and cons of using AI and HI, McFarlin talked about the synergy of using both together and shared examples.
Improved Decision Making: AI can quickly provide data and insights while HI considers context and ethical judgment. For example, AI will help compile the results of sales training calls and a manager or coach can come in and work with employees in a fair and understanding way.
Customer Service: AI can handle repetitive tasks and provide fast response times; HI focuses on more nuanced interactions.
Accurate Forecasting: AI procures data whereas HI brings strategic oversight and context.
Learning Experiences: AI handles real-time feedback and support while HI interprets the outcomes and ensures it is used ethically.
Using AI in the Collision Repair Industry
Ronak discussed some applications for AI in the collision repair industry, the first of which was damage assessment and estimation. AI-powered image recognition software is currently being used to analyze photos and estimate repair costs, potentially reducing the need for in-person inspections. After a collision, a vehicle owner, estimator or adjuster can use AI to capture photos that are uploaded to an app that uses image recognition to evaluate the vehicle’s condition.
The AI model analyzes the images and uses computer vision to find visible damage. It then classifies damage severity based on thousands of pictures of similar damage patterns and their repair histories.
As part of the damage assessment, the AI model estimates repair costs based on parts, labor and other repairs.
“Within seconds, AI provides a breakdown of estimated costs and the time needed for repairs,” Ronak explained. “The estimate includes a preliminary list of damaged parts and repair recommendations, such as ‘repair or replace,’ and an estimated cost.”
Ronak said the estimate can, and should, be adjusted by a human appraiser, when necessary, but AI typically reduces the need for manual adjustment. After being finalized, the estimate is shared with the customer and repair shop, which can streamline the insurance claim and repair process.
“AI-based damage assessment and estimation is increasingly being adopted by insurance companies and repair facilities to enhance accuracy, consistency and speed in handling vehicle repairs,” he said.
AI is also being deployed in subrogation investigations for claims payouts regarding parts pricing and operations paid, which Ronak said has driven increased requirements for supporting documentation.
He noted additional examples of AI such as VIN decoding, diagnostic and calibration recommendations and the use of onboard systems to provide automated notifications of emergency events and self-diagnostics.
Other AI applications include parts ordering and inventory management to look at repair history and parts usage; predictive maintenance and diagnostics to research vehicle telematics and historical data; and repair process optimization to evaluate workloads, repair complexity and parts availability.
Ronak said AI can streamline repair scheduling and suggest optimal workflows. In addition, it can be used for automated quality control and assist with safety compliance by analyzing images of completed work and detecting flaws.
When weighing the pros and cons of using AI and HI, McFarlin and Ronak offered the following to consider:
Efficiency vs. Expertise: The speed of AI or problem-solving abilities of HI.
Cost vs. Customization: Lower recurring costs of AI or potential higher quality via HI.
Consistency vs. Customization: Consistency of AI-driven processes or the adaptability of HI.
Integration vs. Collaboration: How AI and HI can complement each other and what hybrid approaches are available that leverage the strengths of both.
McFarlin and Ronak encouraged attendees to evaluate using AI in their businesses.
“You want AI to be a tool in your toolbox to be something that you can use to augment your process, your workplace and your workflows,” said McFarlin.
“It's going to become even more prevalent,” Ronak pointed out. “Those who figure out how to leverage AI to their advantage are going to be the winners in the world of competition.”
Stacey Phillips Ronak