What are the trade offs?
While AI technology is advancing at remarkable speed, with innovative applications emerging all the time, there are limits to its use in the enterprise. Because of the way most businesses have evolved, there is a limit to the number of business problems that are suited to AI. For instance, as a bank, you wouldn’t use AI to raise capital for you but you might use it to improve your ability to predict movements in financial markets.
Many of the organizations that are making impressive advances through the use of AI were created to be data companies — the likes of Google, Facebook, Amazon and Alibaba. These companies were built to gather and analyze date. Here, AI might be able to predict user behavior to benefit advertisers, or perhaps identify images that breach platform rules.
But for more traditional organizations, legacy infrastructure may make it more difficult to collect and analyze data in general.
In most cases, AI applications in the enterprise are based on machine learning systems. These are highly dependent on training data sets and can produce bad results if the training data contains bias or if the training data differs from the data being scored.
However, you may also find the term AI plastered across technologies to make them sound more impressive than they really are. If you invest in that technology it may fail to meet your expectations.