We’ve discussed both the power and pitfalls of AI in previous posts. It’s too new to completely trust, but much too powerful to ignore. To move forward I would recommend a three-pronged strategy.
First, focus on its undisputed strengths. Deep learning, the branch of AI most prominently in the news today, is exceptionally good at pattern recognition, particularly with images. Utilizing it for optimizing Optical Character Recognition (OCR) for AR is a no-brainer. It has enabled 9ci to develop a very powerful tool that can take virtually any image and extract customer, payment and payment remittance information without the need for labor-intensive templates, zones or manual verification.
Second, supplement AI with domain expertise. Focus deep learning on discovering and leveraging customer payment patterns and open invoice growth, but structure this around expert system governors. One example of this approach would be to define common sense collection strategies based on your company’s experiences and then allow the AI to analyze the results and tweak the strategies accordingly.
Third, establish a collaborative AI partnership with users. Injecting human judgement and awareness into the process is critical, especially with deep learning. Start out having the AI make recommendations to users until a record of success and trust is built. Always ensure that users are made aware of any changes made or proposed by the system. This will guard against the AI making bad decisions based on inadequate or biased training data.