Understanding Decision Support Intelligence
The dream of all decisioning solutions is to expedite decision-making, such that important decisions are made as correctly and as quickly as possible. In the abstract, its perfect form would be a machine that can gather all relevant data, draw correct inferences, and make accurate decisions at a faster speed than any human could manage. But the ideal world rarely lines up with the real world, and at present, the real world is far from attaining full decision automation.
Of course, we have the technology to automate some decisions. The question is where, or whether, those decisions should be left to the machines.
A helpful guideline when considering this question is to think of prospective use cases along two axes: complexity and consequence. As a rule of thumb, the more complex the decision, and the greater the consequences it carries, the less you want to entrust it to a machine. With this principle in mind, we can define three different categories of decision assistance, arranged from the greatest amount of machine intervention to the least.
Decision Automation
The level where machines intervene the most in the decision-making process is “decision automation.” In this scenario, the machine adjudicates decisions by itself, following predefined, programmed criteria. In general, this approach is suitable for use cases with low complexity and little consequences – in other words, situations where the chance of error is slimmer, and the consequences from potential errors are not drastic.
Decision Augmentation
A useful way to frame the question of what decisions to automate is to think not in terms of completely automating the decision-making process, but of what would assist human decision-making. In other words, the goal isn’t to have machines call the shots. Rather, the aim is for machines to provide what you need for your own decisions to be swifter, more informed, and more accurate. We call this concept “decision augmentation,” and it’s suitable for use cases of medium complexity and consequence.
Decision Support
For situations of the highest complexity and greatest consequence, machine intervention should be at its least intrusive. Humans need to be making the decisions in such cases, both to ensure accountability and minimize the possibility of misjudgment. But to help with making those decisions, machines can help with smaller tasks. Hence, we call this category “decision support,” because in it, machines aren’t really involved with the decision-making process – instead, they’re supporting your decisions. Ideally, a decision support solution tells you how and why it arrived at its conclusions, a concept known as “provenance.” This way, you can judge its reasoning for yourself, take a closer look at surprising or unintuitive findings, and defend your course of action in high-consequence situations.
Deciding Where to Automate
We’re now left with the question of what decisions we should leave to the machines. Where can their abilities be most useful? In other words, what strengths do they have that we can leverage to help with decision-making?
Broadly speaking, machines tend to be far better than humans at rote tasks with easy-to-follow decision trees. For example, automatically locking an account after a certain number of unsuccessful login attempts is a good use of automation – the logic is straightforward, and the consequences no worse than minor inconvenience in the event of a mistake. Similarly, it could be useful to automate generating a notification for a credit card owner if two purchases are made that are close in time but distant in geography. Again, the logic behind the decision is clear, and the consequences small in the event of an error.
In terms of decision augmentation and support, this means that machines are best suited for low-level but high-volume tasks in the decision-making chain. They’re good for doing the groundwork of investigative analytics: reviewing information, identifying patterns, and flagging certain findings. This philosophy informs Cogynt, Cogility’s continuous intelligence platform.
Decision Augmentation and Support with Cogynt
Cogynt is designed to assist with medium- to high-consequence and complexity scenarios. Three key features distinguish Cogynt’s approach and make it optimally effective:
- Cogynt works from an expert-defined model. This model defines what to look for among your data stream and what to do with those findings. Models provide analytic bedrock that does not drift or skew, ensuring that Cogynt always adheres to an organization’s policies and priorities.
- Cogynt’s intelligence includes event provenance. Provenance is the information underlying an event or discovery. By including the details that led Cogynt to its findings, it allows analysts to create detailed, contextualized assessments for decision makers.
- Cogynt provides context. Cogynt’s decisions aren’t made in a vacuum. Every decision is made in a broader context that includes a wealth of data and events. Cogynt provides this context with every finding, ensuring that decision-makers can make informed choices and articulate or defend them.
Cogynt harnesses the power of data streaming and the precision of Apache Kafka’s distributed event stream processing to ingest immense volumes of data, filter out the noise, and deliver actionable intelligence. By instructing Cogynt to look for certain patterns and trends within your data, it can alert you in real time when important indicators are found. Cogynt can be further configured so that systems refine their own findings through machine learning modeling, or weigh probabilities through predictive analysis, to make an even more helpful partner in decision intelligence.
To see what Cogynt can do for your organization, request a demo today.
