Why should business understand how AI solutions work?
A prospective client wanted a machine learning recommendation system - but had no data at all, because the project was only just starting. A short story about why following AI trends is not enough, and why business needs to understand what these solutions actually require.
A few days ago a prospective client contacted me about building a machine learning solution to increase product monetisation through a recommendation system. Interested in the topic, I scheduled a workshop to define the scope and learn the details of what the client wanted to achieve.
A surprising workshop
After just a few minutes of conversation something happened that clearly surprised me: when I asked what data sets the client had and in what structure, I was told that the project was only just being created, so there was no data at all. 😮
What’s more, the client asked us to:
- 👉 determine what data should be collected to feed the model
- 👉 build a first version of the model that would work and evolve over time as we learned more about the problem domain
What machine learning is and why it needs data
One definition of machine learning describes it as “giving a program the ability to learn automatically and improve from experience (that is, previously provided data and patterns) without being explicitly programmed”.
The description alone makes it clear that you cannot effectively implement a machine learning model without first acquiring data and being able to explore and analyse it. 😢
Business’s problem with AI and machine learning
This conversation illustrates one of the problems business has with machine learning and artificial intelligence (AI). In reality, many people suggest these solutions because of the current trend, without fully understanding whether they make sense for their business.
How we found a solution
In the end we helped the client by proposing a model based on the ideal point method, known from the field of multi-criteria optimisation. So you can see that when you reach people with the right expertise in data analysis, it is possible to develop a solution that properly fits the problem the business actually has. 😎
Conclusions
Remember not to let fashion and popular trends drive your business decisions - rely instead on a deep understanding of the domain and the support of people with the right experience. 🤓
