There is currently a huge amount of discussion around Artificial Intelligence (AI) and questioning, “Will artificial intelligence replace business intelligence?”
A morning’s dose of LinkedIn posts proclaims that in the not too distant future, fresh-out-of-varsity data scientists can earn a million dollars a year, that South Korean bots will replace all jobs by 2028, and we have a shortfall of 300,000 skilled data scientists across the world. Mark Zuckerberg and Elon Musk are arguing over whether AI is more dangerous to the planet than nuclear weapons, and Deep Knowledge, a Japanese venture capital company, recently appointed an artificial intelligence app to their executive board.
AI solutions are mostly used to augment human decision-making, and have shown amazing potential to help the planet (think cancer-cell identification and crime prediction models in cities), as well as to disrupt existing business models (for example, Amazon’s on-line product recommendations).
Technology expectations are high, and investors are pouring money into start-ups with an AI edge.
Business Intelligence (BI), in comparison, might seem a little dull. Most companies have been through an initial wave of report building and maybe even a few dashboards over the last ten years, and there have definitely been some ‘tried BI, it didn’t work’ stories.
This perception may be because BI projects have often focused on implementing a platform, and pushed hard by the IT department. They knew that business needed information, but the emphasis was on building the road first, rather than the car.
With analytics-based AI, there is more focus on solving specific business problems – such as how can companies better predict cash flow, or which customers should be targeted to ensure renewable contracts aren’t cancelled. Many separate, different-coloured cars are being built quickly, and people are driving them in all directions.
But is one better than the other? In fact, the AI capabilities relevant for analytics (data science), should be used to augment traditional BI, not replace it. You cannot do product recommendations if your customer master data is not high quality. A graph of forecast sales figures without the context of historical sales might be missing the point.
Traditional data warehouses feed the algorithms that generate employee clusters. In other words, you need both roads and cars to provide efficient transportation.
Predicting the future, even with the benefit of data science, is not easy, but it is certain that businesses are still going to need solutions that help drive performance in the pursuit of goal achievement. For an operations manager, this might be a dashboard showing current production against target (BI), with an indication of future production based on a model that predicts artisan absenteeism (AI). For a sales manager, an algorithm, which can best-match telesales agents against prospects (AI), will definitely help, but it will still be necessary to show revenue figures for the month to date (BI) to see how close she is to succeeding.
What is certain is that both AI and BI skills are needed more than ever, and individuals that can master both will be in great demand in our insight-hungry world.