Decision support systems represent a distinct style of machine learning system implementation with a different design approach from systems targeting the full automation of a decision making process.
One example of this distinction is automated equity trading systems that buy and sell stocks with no human intervention vs. equity research, filtering and ranking systems that provide decision support to a human investor. In both cases machine learning might be central to the implementation but the nature of the system design and the development methodology will be meaningfully different.
As Enterprise’s look to take full advantage of machine learning we will see it implemented in a mix of both fully automated and decision support systems.
The need for both types of systems within the same domain will be common. For example, in making decisions on loans a bank may have a fully automated system to provide immediate responses to credit card applications but may have a decision support system for structuring complex commercial real estate financings.
This duality will be true not just in business enterprises, but also in healthcare, legal, government, military, etc. With decision support being the approach used for the most complex and consequential decisions.