Emerging Technologies and Innovation in Analytics
What you must know
Understand the capabilities and limitations of machine learning (supervised, unsupervised, reinforcement), natural language processing, robotic process automation (RPA), generative AI, blockchain and distributed ledgers, cloud computing, the Internet of Things and process mining. Know the distinction between RPA (rule-based task automation) and AI (learning-based inference). Understand model risk: training data bias, overfitting, drift, black-box opacity and the need for human oversight. Know how continuous auditing and monitoring use automation to test transactions in near real time. Understand data ethics in AI, including accountability for outputs and the need for validation and audit trails. Be aware of the professional scepticism required when relying on third-party or vendor models.
Traps in questions
A common trap is assuming AI output is objective - models reflect their training data and design choices. Another is confusing RPA with machine learning; RPA does not learn. Questions may overstate blockchain as a guarantee of data accuracy - it ensures integrity of records, not truthfulness of inputs ('garbage in, garbage out'). Also watch for the misconception that automation removes the need for professional judgement.
Rule of thumb
For any emerging technology, ask three questions: What is the input quality? What is the model or rule doing? Who is accountable for the output?
Worked example
A company deploys an AI tool to flag fraudulent expense claims. The tool was trained on historical claims, which under-represented a particular region. The auditor should test for bias, review the validation approach, confirm human review of flagged items, and document governance over the model - not simply accept the tool's output.