How AI can supercharge the benefits of business intelligence
The promise and ultimate goal of artificial intelligence is to make machine intelligent. With advancement in machine learning, statistical reasoning and pattern recognition, as well as the exponential growth in big data and computing power, AI has become the front and center of technological innovation and business transformation in the second decade of 21st century and beyond.
In this respect, AI is perfectly aligned to the goal of business intelligence, which is to make business more intelligent by augmenting and, in some cases, automating human intelligence. As AI is getting smarter, it is not unreasonable to expect that BI will too.
Traditionally, BI, along with data warehousing and big data technologies, provides systems, tools and processes to help companies harness data from disparate sources and turn them into high quality and actionable information to drive competitive advantage. On the other hand, AI is the simulation of human intelligence processes by machines, including the ability to understand, learn, reason and self-correct.
AI in popular culture often captures the imagination of general public in the form of HAL 9000 as in “2001: A Space Odyssey” or intelligent humanoids such as “Terminator” and “Transformers.” However, it has very little to do with reality and the current state of AI.
Today, all real world AI adoptions and successful implementations are artificial narrow intelligence (ANI) and one area where AI has been quietly making remarkable progress is the expansion of intelligence into business applications and processes via smart algorithms, intelligent bots, cognitive virtual agents and deep learning programs, such as deep neural network (DNN).
In a recent study, IDC predicts “by 2025, at least 90% of new enterprise apps will embed AI; by 2024, over 50% of user interface interactions will use AI-enabled computer vision, speech, natural language processing, and AR/VR.”
Although AI powers many systems and applications ranging from improving business operational efficiency to analyzing patterns and trends or detecting intrusions and fraudulent transactions, the focus here is limited to examine AI’s current and potential future impact on BI and how it could empower BI and decision making in general.
The synergy between AI and BI starts with data. Both AI and BI requires lots of data. One argument that can be made is modern successful AI implementations, such as many deep learning approaches to real world problems, rely entirely and excessively on data.
How to generalize beyond training dataset and acquire intelligence becomes one of the key measures of general intelligence. Current AI is data-driven by nature while BI is data-driven by definition.
From this perspective, data management under traditional BI framework could provide AI with critical data inputs. Furthermore, AI system or intelligent device generated data or data parsed, classified and interpreted by AI systems could in turn be new information sources for BI to better correlate and understand events, e.g. audio, video, image and other multimedia data.
In “Unplugged – The Disconnect of Intelligence and Analytics” (2011), I argued that business intelligence and business analytics (BA) need to converge toward one integrated approach to data-driven decision making as well as proposed a reference architecture. The first major transformation in BI is the modern statistical and analytical methods, tools and algorithms that give rise to BA which reciprocally extends the traditional BI scope and capabilities. Now, AI could transform BI yet again and make it more robust and intelligent.
The most important thing to understand AI is it is not one dimensional but multifaceted, contrary to the common misconception of AI - a single Being. And perhaps more surprisingly to a layman, often it doesn’t have to take humanoid form or even intelligent device.
Whether on Internet or embedded in various online, offline, mobile applications, most AI programs take the shape of machine learning algorithms, such as pattern recognition, anomaly detection, feature reduction or matching, classification, simulation, microtargeting, natural language processing, auto-decisioning and adjudication.
BI could leverage these algorithms and models in building machine learning capabilities, analytical as well as operational. In this area, AI has already started impacting many business applications including BI and business decision making. It will only get deeper into enterprise applications and more sophisticated in terms of intelligence modalities, such as logical reasoning and learning by inference, deduction, analogy, schema, etc.
As AI advances and inches toward general intelligence, new progress will be made in the areas where AI is historically weak and real world problems that can’t be solved currently.
For instance, causal effect analysis is very limited with current AI which tends to identify correlations, not causations, as the old saying goes correlation does not imply causation. Traditional BI methods and tools, without the intervention of human intelligence, are also weak in interpreting cause and effect, such as what actually causes the current quarter revenue downturn, instead of simply reporting it.
Would AI one day help BI in understanding root causes or potential causes? Can AI replace reasoning by association with causal reasoning? If human intelligence and reasoning capability can somehow understand and explain business ups and downs and even cycles, to certain extent with some degree of certainty, then it’s not by any stretch of the imagination that intelligent machines can too one day.
For example, one approach is probabilistic causal model (PCM) proposed by computer scientist Judea Pearl, although it still remains a theoretical framework and lacks real world solutions and business applications. This is one area of AI that needs breakthroughs but the possibilities and benefits of achieving such potential are enormous.
Future AI also could potentially improve in “human level” reasoning, for lack of a better term, and provide “human level” narratives in NLG (natural language generation). The current available applications and tools of NLG are primitive with very limited intelligent and language generating capabilities. But if this technology matures, it could expand current BI’s numerical and graphical capabilities to more natural language oriented descriptive and prescriptive narratives, such as AI-enabled financial and regulatory reporting or automated business narrative and synopsis writing.
Future AI will provide more robust and intelligent what-if analyses to business with real world simulations. Traditional methods based on probability distribution and simulation, such as Monte Carlo Simulation, have limitations. Due to more flexible generalization, optimization and abstraction at high level, intelligent machine based ensemble models could outperform traditional statistical models which use randomness to solve deterministic problems and which are not ideal for complex multi-scenario planning and nondeterministic analysis.
For its long term sustainability, and no matter the state of future development and the exhibited level of intelligence, the AI in BI must be ethical, transparent, explainable and trustworthy to businesses and consumers. Therefore, it is paramount to reduce and, if possible, totally eliminate biases and prejudices in data, processes, systems and algorithms which could impair users’ confidence and sound business decisions.
Moreover, as both AI and BI are moving to cloud, AI powered BI solution architecture in implementation could be more unified and integrated rather than disparate silos to meet business needs and regulatory requirements.
Would one day AI replace BI? In a scenario like Singularity which is all powerful and encompassing superintelligence, according to philosopher Nick Bostrom and futurist Ray Kurzweil, one has to conclude there is such possibility. However, this question perhaps is moot in case such superintelligence does become reality. In the near term, a future that AI will augment and empower BI is more likely. For now, both AI and BI will evolve in their own trajectory while intersecting.
Future generations of AI powered BI systems will be more intelligent and resilient and perhaps self-correcting in learning from data and beyond, reasoning with logic and fact or even counterfact, understanding causality, discovering hidden patterns, uncovering trends, identifying problems undetectable or unforeseeable by humans before they occur, predicting risks and pitfalls, answering quantitative or qualitative business questions in a more natural way or writing narratives comparable to general intelligence.
The full promise of AI will supercharge the future of BI.