From The Lab To The Real World - Challenges In Getting Algorithms To Achieve Their True Potential
There has been significantly growing interests and investments by businesses to use data and algorithms to capture new markets or to create new opportunities in existing markets. This can range from customer and marketing analytics, personalisation of online consumer experiences, financial loan approvals, and many others.
Yet, many initiatives never make it from the prototypes to the real-world application, and many deployed to the real-world fail to meet the original expectations during the internal presentations and approvals. Even worse, some ill-conceived solutions make the headlines for the wrong reasons after discriminatory decisions were highlighted by the public, or an inadequately robust solution results in some catastrophic incidents after deployment.
During the webinar "From the Lab to the Real World - Challenges in Getting Algorithms to Achieve Their True Potential” on 26 October 2022, Gerry Chng, Executive Director, Deloitte & Touche Enterprise Risk Services Pte Ltd spoke about AI Ethics & Governance, and how a Trustworthy AI framework helps in identifying and addressing the potential areas of concern. In his presentation, Gerry said that AI Readiness brings about challenges in:
- Good talents for more comple use cases.
- Proper training and skilling to use foundational models and libraries.
- Common taxonomy to bridge technical and business users.
- Good quality data.
- Different roles for data engineering, model building, Infrastructure, and Application engineers to build a proper end-to-end process.
- Continual monitoring.
- Encapsulating these in a Machine Learning Operations process, and consider the pillars of Trustworthy AI within it.
