How artificial intelligence creates value today?
The word “artificial intelligence” makes most people think of large language models and products based on them such as ChatGPT. But did you know that AI methods based on machine learning was a rising trend already before the launch of ChatGPT? Pioneers are already using these successfully and extracting real value from them.
Using these methods requires not only mathematical understanding of the underlying algorithms, but also of the business process that is being improved. It is actually rather easy to make an AI algorithm crunch through some business data, but interpreting the results and ensuring that the desired outcome is achieved requires skill. But what can AI methods accomplish?
Predictive analytics takes you from past patterns to smart decisions
AI algorithms are particularly useful in the form of predictive analytics. By feeding the model enough high-quality historical data it learns to spot signals that are invisible even to human experts and can predict things like order surges. As a result the machine can optimize resource allocation in such a way that all orders are filled while simultaneously avoiding waste.
Recommender systems drive engagement and sales
Most people are familiar with recommender systems from services such as TikTok, Spotify, Netflix, and Amazon. Lots of resources are being poured into developing the algorithms powering these systems as their business potential is enormous. The most obvious use case is upselling the customer by showing them additional things they might be interested in, but additionally the algorithms can be tuned to optimize features like the best ratio between interesting content and high profit margins.
ANOMALY DETECTION spots what others miss
Anomaly detection has become a popular and successful domain where AI is applied. The algorithm monitors a business process and alerts when a deviation is noticed. These methods can help catch fraud, human errors, machine faults, manufacturing mistakes and much more.
FOR REAL value already today focus on the problem, not AI hype
AI based on language models is no doubt a useful technology that I use at work every day. It is however a rather new technique that is still searching for the best and most valuable use cases. While others are thinking about how and where to utilize agents, I suggest searching for business problems to solve and choosing the appropriate method for that, rather than trying to fit a problem to the solution. Large language models and agents might revolutionize the playfield tomorrow, but there are methods available that can provide real value already today.
Post by Otto
“I’m a software engineer with an interest in artificial intelligence, spanning from machine learning and evolutionary algorithms to logic-driven AI. I enjoy exploring how these technologies intersect with large-scale distributed systems.”