With The Power Of AI, RAFM Analysts Have a Wingman but No Autopilot

AI Enhances RAFM Analysts: A Synergistic Approach in Telecom

Artificial intelligence image on blue fpnt with white text

As practitioners of revenue assurance and fraud management, we at LATRO have been in conversations about whether AI could potentially replace the work of RAFM analysts and investigators and free them up to do other, more complex tasks. Once an AI is trained, it can work tirelessly, scanning networks 24x7x365 for indicators of fraud or revenue leakage. AIs never take a break and never need to take a vacation. ChatGPT and other LLMs (Large Language Models), have also received a lot of hype recently. Many IT departments quickly poured tons of corporate data into them and were able to retrieve sensible answers from them. However, so far, the answer to whether AI will replace RAFM personnel is no, mainly because they can do things it cannot.

The truth is that most AIs are mediocre at most things, which makes a lot of sense. When trained on large datasets, like telecom data, AIs regress to the average of that dataset. Furthermore, training AIs to detect instances of fraud and revenue leakage is quite difficult. This is because those datasets are often quite skewed, containing only a small number of true positives. Obtaining many good, clear examples of revenue leakage or fraudulent activity is the equivalent of finding modern-day needles in a haystack. To continue with the analogy, today, there are not enough needles in the haystack, so AI can learn enough about it, and why trained analysts are extremely valuable and needed.

As of today, the takeaway could be that AI will only support mundane tasks with average results at best. However, the dance between trained professionals and AI has only begun. The mainstream journey of LLM-based AI only started about 14 months ago, and we are already at a level that most people did not anticipate for this decade. This speed for sure will continue at a different pace and probably will not have a linear development. Roadblocks will occur, and further breakthroughs and significant leaps will be needed to excel the capabilities of AI further. No matter which speed and quality of improvement you believe will happen, what we see today is the worst it will be from now on.

It’s probably fair to say that AI is not yet the autopilot that people hoped for, it rather presents itself as a reliable and relentless wingman. It can be a strong tool for trained professionals to spar on insights, automate easy and mundane tasks, and, as a result, achieve significant savings in time.

The need for experienced fraud management and revenue assurance professionals who are able to think creatively is unbowed. RAFM analysts will always possess a spark of curiosity that drives them to investigate issues that just look “off”. That is something AI is not able to do—at least not yet.

To learn how LATRO can help your team increase its effectiveness and consistently perform above average, contact us and inquire about our RA and FM maturity assessments.

By Jacob Howell, Director of Corporate Strategy at LATRO.