Reducing the Threat of Fraud Through AI with "Technology Shichifukujin"
What Is DCON?
The National College of Technology Deep Learning Contest, known as DCON, is a business creation contest where technical college students create products that combine their manufacturing skills with deep learning and compete. The judges, active venture capitalists, evaluate technological and business potential to find the winner.
This year, in 2024, "Technology Shichifukujin" from TMCIT won first place for their product FraudShield AI, an easily installed device that detects fraud in telephone calls.
Technical College: An Environment to Invite Innovation
TMCIT is a technical college with eight main courses, each lasting five years. First-year students learn manufacturing basics before choosing a course in their second year. The campus fosters innovation, through practical learning and technical after-school activities providing an ideal environment for students like Technology Shichifukujin to develop and showcase their skills.
As a technical college, TMCIT encourages students to explore learning opportunities like DCON while emphasizing responsibility. Though students join after junior high, they are treated like university students, accountable for their academic careers. This responsibility also brings opportunities, such as studying abroad through overseas programs, like the IEP (International Education Program) and the GCP (Global Communication Program).
Who Is Technology Shichifukujin?
Technology Shichifukujin is a group of nine students, all under 20, studying at TMCIT. Using AI chatbot ChatGPT, they chose the name to represent the seven team members allowed in the contest, likening their diverse interests to the distinct virtues of the Seven Gods of Luck in Japanese culture.
The team of third year students won DCON. Team leader Nishitani Sotetsu chose members based on key skills like programming and circuit board creation, ensuring open communication needed to meet the contest deadlines over six months of collaboration.
Despite feeling at a disadvantage due to their inexperience, they succeeded through hard work and the invaluable advice of DCON advisors and group mentor Yanagihara Takashi, president of Ridge-i Inc., a company that specializes in developing cutting-edge technologies to address critical business and social issues. His frequent visits and advice were instrumental in their success.
Entering DCON at the age of 16, the team faced challenges but leveraged their fresh perspectives and acrobatic thinking to explore AI's potential, debate its applications, and drive innovation.
The contest helped them develop key skills, including business proposal creation and advanced programming. Unlike their prior experience, the FraudShield AI project required innovative problem-solving and collaboration, opening new opportunities for growth.
The Birth of FraudShield AI
Nishitani's own experience inspired the idea of FraudShield AI. A year ago, the group leader paid for an ESTA (Electronic System for Travel Authorization) to study abroad in the college's IEP through what turned out to be a fraudulent website and consequently lost around 9,000 yen.
Nishitani revealed that while the financial cost was relatively minor, the emotional toll was profound. Initially believing fraud was only experienced by the elderly, he now aims to raise awareness and develop a solution. His research into fraud's financial and emotional consequences, including self-blame and family disputes leading to severe outcomes, fueled his commitment.
FraudShield AI in Action
Fraud is prevalent in Japanese society. In 2023, among the 19,038 cases of special fraud recognized, 77.5% of initial contact was by telephone, and 90.5% of those calls were made to the victim's landline.
FraudShield AI detects fraudulent calls in real-time, displaying fraud risk through increasing LED lights and notifying users with a buzzer once fraud is confirmed. Designed for users of all ages, especially the elderly, the interface was refined through trial and error, including changing the word "fraud" from kanji to katakana to enhance readability, addressing challenges posed by the complex kanji character strokes.
The interface is easy to use and set up and only requires the user to connect the microphone to the landline phone. Users can also connect the device to their family's LINE account, sending notifications if the buzzer sounds, allowing family members to check on the user and decide on next steps.
Leader Nishitani hopes that the circulation of FraudShield AI will put people on their guard against suspicious calls and increase awareness of fraud, leading to decreased cases.