Artificial Intelligence in Financial Technology: A Systematic Review of Applications, Challenges, and Future Directions
This systematic review of 156 studies from 2018 to 2025 demonstrates that AI significantly enhances predictive accuracy and automation across six key Fintech domains while highlighting critical challenges like algorithmic bias and explainability, ultimately offering a research agenda and regulatory insights for stakeholders.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine the world of money and banking as a massive, bustling city. For decades, the rules of this city were written in stone: if you had a long history of paying bills on time, you got a loan; if you didn't, you didn't. But recently, a new kind of "super-intelligent city planner" has arrived. This planner is Artificial Intelligence (AI), and it is completely rewriting how the city of Fintech (financial technology) operates.
This paper is like a massive map drawn by a researcher named Petros Degefa Mulu. He didn't just look at one street; he studied 156 different reports and studies from 2018 to 2025 to see exactly how this new planner is changing the city.
Here is the breakdown of what the paper found, using simple analogies:
1. The Super-Planner is Faster and Smarter (The Good News)
The paper found that this AI planner is much better at predicting the future than the old, manual methods.
- The Credit Score Analogy: Imagine a loan officer trying to guess if you will pay back a loan. The old way was like looking at a single photo of your bank account. The new AI way is like watching a 24-hour movie of your life—your phone usage, your utility bills, even your social habits.
- The Result: The paper says these AI models are like having a crystal ball that is 8.3% more accurate than the old crystal ball. In the world of catching fraudsters, AI is like a detective who can spot a criminal ring 24% better than a team of human detectives working together.
- Where it shines: It's great at spotting fraud (like a security guard who never sleeps), trading stocks (like a race car driver who reacts in milliseconds), and helping people who don't have traditional bank records get loans.
2. The "Black Box" Problem (The Mystery)
While the AI planner is great at making decisions, it's terrible at explaining why it made them.
- The Analogy: Imagine you apply for a loan, and the AI says "No." You ask, "Why?" The AI replies, "Because the numbers in my brain said so." It's a black box. You can see the input (your application) and the output (the rejection), but you can't see the gears turning inside.
- The Risk: In a city, if a traffic light turns red for no reason, people get confused. In finance, if a bank can't explain why they denied your loan, it's unfair and against the rules. The paper notes that even the tools we use to try to open this box (called "Explainable AI") sometimes give us confusing or wrong answers.
3. The Ghost in the Machine (Bias and Fairness)
This is the paper's biggest warning. AI learns from history, and history is messy.
- The Analogy: Imagine you teach a new student to judge people based on old yearbooks. If those yearbooks show that the school used to reject people from a certain neighborhood, the new student will learn to reject them too, even if they are now good students.
- The Reality: The paper found that AI often copies the bad habits of the past. If the data it was trained on had racial or gender bias, the AI will amplify that bias. It's like a photocopier that makes the smudges on the original page even darker. One study mentioned in the paper showed that an AI rejected minority applicants 23% more often than others, not because they were bad risks, but because the AI was using "proxy" clues that were unfairly linked to their race.
4. The Domino Effect (Systemic Risk)
The paper warns that if everyone uses the same AI planner, the whole city could crash at once.
- The Analogy: Imagine every traffic light in the city is controlled by the same computer program. If that program gets confused by a strange cloud pattern, every light turns green at the same time. Chaos!
- The Risk: If all banks use similar AI to trade stocks, they might all decide to sell at the exact same second. This could cause a "flash crash," where the value of money drops instantly, not because of a real problem, but because the robots all panicked together.
5. The Rulebook is Playing Catch-Up
The city is moving faster than the law can write new rules.
- The Analogy: The AI planner is driving a Formula 1 car, but the traffic police are still writing rules for horse-drawn carriages.
- The Current State: The paper looked at five different countries (EU, US, UK, China, Singapore).
- Europe is trying to be strict, classifying credit scoring as "high risk" and demanding strict safety checks.
- The US is more focused on checking the "model risk" (making sure the math works).
- China is very hands-on, directing the AI and controlling where data lives.
- Singapore is trying to be friendly to innovation but wants to keep things fair.
- The Gap: The paper says the rules are uneven, and this creates a "governance gap" where bad actors might slip through the cracks.
6. The New Frontier: The "Hallucinating" Assistant
The paper also looks at the newest trend: Generative AI (like the chatbots you might talk to).
- The Analogy: Imagine hiring a very confident but slightly confused tour guide. They can write a beautiful speech about the city, but sometimes they make up facts.
- The Warning: The paper found that when these AI assistants try to give specific financial advice (like tax rules), they get the facts wrong about 11% of the time. That's like a tour guide telling you the museum is closed when it's actually open, or vice versa. This is dangerous for people's money.
The Bottom Line
The paper concludes that AI is a powerful engine for the financial city. It makes things faster, cheaper, and more accurate. But, it's an engine that can run off the road if we aren't careful.
The author argues that we can't just let the engine run wild. We need three things to happen at the same time:
- Better Engineering: Make the AI fairer and more transparent.
- Better Rules: Update the law to handle these new machines.
- Better Teamwork: Lawyers, computer scientists, and bankers need to talk to each other to make sure the city stays safe for everyone.
The paper ends with a call to action: We need to make sure this technology helps everyone, not just a few, and doesn't accidentally break the global financial system.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.