Navigating the complexities of AI adoption in the AML space

AI

In a recent post by Moody’s Analytics, the firm asked three key questions about driving efficiency with AI-powered screening.

Distinguishing between artificial intelligence (AI) and machine learning (ML) is essential when considering the digital transformation within computer science. AI encompasses the wider discipline of creating intelligent machines that emulate human intelligence. It is tasked with copying or performing duties that would traditionally require human decision-making or action-taking. Meanwhile, ML, a subdivision of AI, prioritises harnessing large volumes of data and nuanced patterns within this data to develop software models that aid in problem-solving and deriving insights.

These ML algorithms can progressively improve their performance through the incorporation of new data. This adaptation enables systems to recognise patterns, make predictions and execute tasks without the need for direct instruction. Despite the power of ML, it remains one part of the wider AI landscape, which also includes sectors such as natural language processing (NLP), computer vision, and robotics.

In the realm of Moody’s Analytics, we utilise a supervised learning model to enhance screening efficiency. This model is trained with labelled data, where input data aligns with the desired outputs. This allows the model to draw correlations between inputs and outputs by recognising patterns and relationships in the data.

The AI technology can be adapted to ‘know your customer’ (KYC) and anti-money laundering (AML) processes for improved operational efficiency. Moody’s Analytics has created an AI-powered Alert Score for every record requiring a company’s review. This score, ranging from 0 to 1, reflects the confidence level of the screened name, with 0.00 representing a no-match and 1.00 a match.

The Moody’s Analytics Grid is the source of data for the AI Review. Our AI Review global model has been trained with over 12.1m rows of data. In addition, specific firm data can be applied to a local model for deployment within a company, aligning better with its policies and analysts’ behaviour.

The Alert Score can be employed to filter results. Firms can decide to review only those results with a score exceeding 0.25, aiding in the reduction of false positives. This level of tunable screening, defined by a company, can be adjusted to enhance efficiency while maintaining control over the screening process.

Every alert is scored by the model on a scale of 0 to 1. A score nearer to 1 indicates a higher-risk screened name, which may necessitate additional investigation. By establishing a threshold, scores on the lower end suggest a reduced chance of an alert and can be automatically filtered out. The ML technology can thereby function as a level-1 analyst, offering repeatable decision-making to sift out false positives and allowing compliance teams to focus on the investigation of higher risk alerts.

It is crucial for any company implementing AI/ML technology to comprehend its inner workings within the organisation’s processes and procedures, and to enforce governance around its design and implementation. At Moody’s Analytics, we intentionally deploy static models that don’t learn from customer decisions in real-time without human intervention.

The integration of AI within the AML industry comes with its share of challenges, such as significant costs for deploying AI systems, data quality and availability issues, complex regulatory compliance requirements, and difficulties in understanding and explaining the decisions made by AI models. Despite these challenges, the potential benefits of AI and ML in enhancing the efficiency and effectiveness of KYC and AML processes make it an area worth exploring for many financial institutions.

Read the full post here.

Keep up with all the latest FinTech news here

Copyright © 2023 FinTech Global

Enjoying the stories?

Subscribe to our daily FinTech newsletter and get the latest industry news & research

Investors

The following investor(s) were tagged in this article.