Page 75 - NO.142銀行家雜誌
P. 75
區塊鏈領域發展非常迅速，銀行業須注意其發展趨勢。（圖／達志影像）
好學習目標，而是直接尋找不同項目中的共 標也毫無意義，因為這既占用了時間，又減
通模式，然後將產生的集群送入一個設好規 少了可以用來改善系統的資源。第三個案例
則的系統，該系統會為每一個集群設定專屬 則讓人看到技術如何改善客戶體驗、加強查
的風險門檻。 帳和藉由標準化決策降低整體風險。
另一個案例是使用人工智慧來大幅減少 如果原本只是為了改善既有的流程而
錯誤警報。如果一個系統需要人類來整理大 使用A I，那麼前述的好處可能就不是那麼顯
量誤判的資料，即使這個系統能找到所有目 著。林靖國說，「使用人工智慧來輔助20年
clusters to a rule-based system, which tailors risk ago is like attaching a rocket to a bicycle in order to go
thresholds for each cluster. to Mars,” says Mr. Lem. Instead, the main purpose of
A separate case used AI to drastically reduce technologies is to discover new sources of value.
false alerts. A system that finds every target case is The blockchain aspect of digital transformation
useless if it forces humans to sort through a massive was represented by ConsenSys, the world’s leading
amount of false positives, which both takes up time pure-play blockchain company. ConsenSys specializes
and reduces energy available to improve the system. in decentralized finance (DeFi) powered by smart
A third case demonstrated the role of technology contracts on Ethereum. DeFi represents “financial
to improve the customer experience, enhance the primitives” such as stable coins, borrowing, lending,
audit trail, and reduce overall risk by standardizing insurance, derivatives, and securities on top of the
decisions. settlement layer (blockchain) and decentralized
These benefits may not be obvious if the exchange (DeX) layer, and below the application layer
motivation is simply to improve existing processes. of decentralized applications (DApps).
"Using AI to supplement applications from 20 years DeFi has grown explosively, from about 100,000
台灣銀行家2021.10月號 75
1 5% JOEE
