Is there a difference between Analytics and AI ?
Yes there is.
Traditional Analytics (which i have been doing for 35 + years) is the process of analyzing vast amounts of data (structured and unstructured) with the desired outcome of reporting on and analyzing a new result, patterns and finding new relationships between variables. Assumptions are made by humans, and then data is analyzed. Traditional Analytics is based on past events (days/weeks/months/years + ago), it does not attempt to predict the future (like AI algorithms often do). As an example, Analytics can help a company’s marketing or sales team to increase more focused efforts (marketing or selling something else/new) to this specific target market.
Predictions or Predictive Analytics are based on Analytics against historical data and then rely on human interactions, assumptions, knowledge of past patterns and potentially even intuition to query data, identify key variables, find new or hidden patterns, and then create and deliver assumptions, predictions or a hypothesis. The capability for an organization to deliver Predictive Analytics is constrained by the time, quality, quantity, skills, ability, experience and cost constraints associated with an often finite number of highly skilled data analysts.
AI, for example Machine Learning or Deep Learning Algorithms, tends to focus on delivering a Prediction, Predictions or a Hypothesis as a desired outcome. A core objective is to “learn:, ie get better from more “experience” (training) and of course by leveraging more data. AI Model Frameworks and new AI Models are selected and/or created, AI Models are then Trained, and then the AI Models are put into production. Once an AI model is running there is little or no need for human intervention during the process of creation of the Model result (Prediction or Hypothesis). For example, if you want to identify a certain set of people that appear in a collection of 100,000 images or hours of Video, you would run an AI model against that data and the output would be a prediction or hypothesis of a series of names of the individuals that are included in those images. There is very little involvement by Humans in the ‘prediction’ phase (Model Inferencing), and the results may come back very quickly. Could a human being also do this ? Yes of course, but it may take hours/days/weeks and the person doing it may actually produce a result that is much less accurate and of course much more costly and slower than the AI Model. In many ways, today’s AI use cases tend to be focused on trying to get software to re-create something certain (not all) people could do but when leveraging AI software you can (hopefully) do this specific task faster, safer, more accurately, with a higher quality, or at much lower costs or at much higher levels of scale. For example, a good use case for leveraging an AI Algorithm would be to create brand new Marketing and Sales campaigns hyper-personalized to specific individuals behaviors, values, wants, needs, desires, personalities, etc. If done properly, the results could be achieved with limited Human intervention and result in much higher levels of business value.
The most interesting thing i find about analytics (even though I’ve been involved with doing it for over 35 years) is the use cases still tend to focus on discovering or uncovering something brand new. One example here is Fraud Ring detection — Once a single instance of Fraud is detected (say by using an AI Algorithm), you can go back and examine historical data and then uncover and unwind a larger, broader and more pervasive Fraud Ring by leveraging Traditional Analytics.
Net, yes there is a difference between Analytics and AI, however i recommend you leverage the best of both worlds going forward. All opinions are my own.
