AI is not an Elixir

Artificial Intelligence (AI) has reached the peak of the hype cycle. Company Boards are now asking their leaders to map out their own AI strategies. With that said, many organizations do not have a clear understanding of what AI is and how to use it. In the case of Marketing there are many ways to benefit from AI. To ensure you are using AI in the right way, it is essential for you to understand how AI works and when to use it. 

AI is now a blanket term for most things beyond a simple Excel spreadsheet. You are likely experiencing this right now. To sound data-smart and data-savvy your vendors, agencies and possibly your very own employees are overreaching the notion of AI. But how can you tell if it is real or vaporware? It is not easy, though you can start with asking a few questions: what was the data and what algorithm was used to process that data? Wait, you didn’t know that AI is not THE algorithm? Technically, no! AI represents many variations of machine assisted algorithms that work to emulate a logical human decision. 

There are several algorithms one can use, and it greatly depends on the type of data. So, if you want to kick the tires of someone selling you AI, just ask what data was used and what algorithm was used. It can be that simple. 

For example:

If you have unstructured text data, you would use Natural Language Processing

If you have image data, you would use Image Recognition Processing

If you have transaction data, you would use Autoregressive Modeling

And so on…

Beyond the type of data and algorithms, there are two additional things to consider: 

1. While it is telling to know the type of data, it is more telling to know if the data is clean, unbiased and accurate. AI works on good data, it fails using bad data.

2. While there are many uses for AI, applying AI to the wrong use case will result in failure and hurt the project. Not all can be solved with AI.

AI is the process of constant calibration. The beauty of AI is that it uses calibration to improve prediction as part of an always-on feedback loop. A good use case for AI is online search. As you search, the search engine supplies you with the predicted results, once you click – this behavior reinforces the prediction and improves the accuracy. With that said, AI does not work for fixed data. Let’s say a one-time customer segmentation or a customer survey. For these projects Cluster Analysis or a form of Factor Analysis is way more appropriate.

The future of AI can only succeed if we start using it in the right way. AI is a not an elixir. Applying AI for the wrong use cases will hurt its reputation, your reputation and deliver erroneous results.

Good luck and please reach out with any questions or comments!