AI has been all the buzz for the past few years, and while there are tons of questionable ways that this technology can be used—including but not limited to world domination that would eventually result in the demise of the human race—there are indeed a few use cases that can make life easier.
If you happen to visit the Philippines—which is where I’m from—you’ll notice that there are tons upon tons of motorcycles roaming the streets. Over here, two-wheelers are many people’s primary means of mobility, and that’s mainly because of how easy it is to buy a motorcycle here.
Dealers grant loans to pretty much anyone, but that often comes at a cost. In the ever-increasing likelihood that these folks can’t pay up, well, repo agents head over to their homes and repossess their bikes. Of course, this comes after a few friendly letters that quickly escalate to strongly worded demand letters.
The result is an expansive second-hand market that consists of both repossessed motorcycles and bikes for sale on classifieds like Facebook Marketplace. Naturally, making sure you get the best bang for your buck can be quite daunting.
And this is where AI comes into the picture.
The Philippine Department of Science and Technology (DOST), in partnership with the Asian Institute of Management (AIM), has developed a tool that makes use of AI to assess the value of second-hand and repossessed vehicles. It’s called the Automated Repossessed Motorcycle Assessment System or ARMAS.
But how does it work?
Well at first, it all seems cut and dry. The system asks for key pieces of information about the motorcycle in question: make, model, model year, and date of sale or repossession. It’s similar to many other online services like insurance premium calculators and whatnot.
But things get pretty interesting from here. The system will then ask you to upload pictures of the motorcycle from various angles. Here, the algorithm will analyze whether or not the bike has been damaged, if there are any aftermarket accessories on it, or if it’s still sporting its original colorway.
Last but not least, ARMAS will ask you to fire up the motorcycle, wherein the system will “listen” to the bike’s sound. We can only assume that the algorithm has been trained to watch out for odd noises coming from the engine, sounds that someone unfamiliar with motorcycles may just shrug off.
It was reported that in the past, human assessors, along with the help of trained mechanics and technicians, would need around an hour to assess the value of repossessed motorcycles. As such, ARMAS gives assessors a pretty solid baseline to work with, pointing scrutineers towards potential issues.
RideApart's tech-cynical leader Jonathon Klein, however, has his doubts on whether this will be helpful at all, saying, "The current crop of AI systems have been shown to hallucinate and give bad advice. See my story on Google's AI offering its thoughts on motorcycle gear safety. But this seems like it'd be harder to program, as there's more nuance required in evaluating a motorcycle's condition. And training the system to 'listen' to the engine for odd noises would be difficult, if not impossible, if you've recorded the video near the public."
At present, ARMAS isn’t yet available to the general public—a shame, as I would’ve loved to try the system for myself. It’s currently being used internally by some banks and credit companies that specialize in the procurement of repossessed vehicles.
But hey, it’s likely just a matter of time until this software makes its way to the general public.
