The concept of artificial intelligence is applied to business in many different ways, and big shifts in artificial intelligence research can be extremely useful for making progress in business-facing software capabilities. Connectionism is a new direction in which much artificial intelligence research is proceeding, and it's one that is likely to dramatically change the tools and resources that businesses use to benefit from the capabilities of artificial intelligence solutions.
Connectionism is a philosophy of artificial intelligence that promotes modeling the human brain by creating small artificial units corresponding to human neurons and groups of neurons in the brain. One of the fundamental aspects of connectionism is the insistence that high-level behavioral and cognitive systems can be built using small individual units tied together in a combined network. With that in mind, the rise of the artificial neural network (ANN) does much to promote connectionism and Hebbian theory, named after mathematician Donald Hebb and his work in the 1940s.
Connectionism suggests that the artificial neural network is going to have critical applications in artificial intelligence advancement. Scientists already have detailed ANN models at their disposal, and artificial neural networks are enhancing machine learning in many different fields. When it comes to enterprise use of artificial intelligence, connectionism can really change the underlying ways that assistive technologies work.
Looking at traditional enterprise business intelligence tools, we can see that many of them are based on some fairly traditional methods, including probabilistic tools. One of these is Bayesian logic, which utilizes cause-and-effect and decision trees and manipulates big data sets according to this logic to create decision-support results (see an article on the popular use of Bayesian logic in business here).
Perhaps the biggest way that connectionism will impact artificial intelligence in business is that it will replace many of these Bayesian logic models and probabilistic models with models that work on the basis of artificial neural networks. The artificial neural network is a collection of small pieces that individually have little meaning. There's not a lot of logic built into the individual units – instead, the network ties together the outputs of these units, and makes it into a logical result. With that in mind, business artificial intelligence tools built on connectionism will be fundamentally different from those that have been popularly used in the past (see this instructive thread on Quora). Rather than achieving computational results through logic, they will achieve these results through running complex machine learning algorithms through an artificial neural network and examining the outcomes.
Some experts contend that the rise of connectionism has a lot to do with limitations in modern research on logical artificial intelligence. In other words, since researchers had maximized a lot of the potential of traditional AI, connectionism and the artificial neural network provided a means to move forward and continue to enhance and broaden how these technologies work and “think.” Models based on connectionism are bringing us much closer to a full simulation of the human brain and biological thought process, which is why these innovations are going to be so important for all kinds of business artificial intelligence – for example, the outputs that a business will utilize from sales force automation or customer relationship management or supply chain or facilities management tools will all be based on these much different models.
Is AI and machine learning something your company could benefit from? Visit AltaML to find out.