1 Executive Summary
Artificial intelligence (AI) is too broad of a concept to assess as a whole. To discuss AI more accurately, we need to recognize it as an umbrella term that includes several different disciplines. These disciplines are Natural Language Processing (NLP), Machine Learning (ML), machine reasoning, computer vision, and robotics. The disciplines that belong to AI are fluid, changing as new capacities are defined for AI. The first three disciplines deal more closely with the intelligence aspect of AI, while the last two handle designing physical structures that allow AI systems to interact with the real world.
These disciplines build on each other to make up narrow AI, the ability for a system to independently complete a defined task. The pursuit of artificial general intelligence – systems that are capable of intuitively reacting to situations they have not been trained for in a human-like, intelligent way – will likely require a new set of disciplines. It remains to be seen how the path to artificial general intelligence will be achieved.
When judging the maturity of AI disciplines and capabilities, it must be made clear if a technology is assessed based on its defined application – such as translating from one language to another – or on its ability to match or exceed human intelligence and performance for that task. This Leadership Brief is written for the enterprise audience, with the understanding that many readers are considering adopting one or more of these AI technologies. Thus, we measure for the technology’s narrow application, its ability to do what it was trained to do.
Explainability is a key element of maturity in AI technologies. AI cannot be mature without the ability to track exactly what influenced and produced a decision, prediction, or action. Proper care and attention to PII data is also an expectation of all mature AI technologies, especially as they deal in large volumes of training, enterprise, and personal data.
