Artificial intelligence is here. AI-driven capabilities that not too long ago may have seemed like science fiction are now being discussed, created, and applied in businesses across the globe. As artificial intelligence and the way we think about it continues to evolve, an artificial intelligence definition may feel difficult to pinpoint. For all the headlines and boardroom discussions dominated by the subject, a working artificial intelligence definition still seems out of grasp for those individuals and executives within an organization that may not directly engage in AI projects or initiatives.
Breaking Down the Definition of Artificial Intelligence
At a high level, AI can be defined as the field of study and techniques that teach computers how to learn, reason, perceive, communicate and make decisions like humans do. That is a very lofty definition, but it can be broken down further into three categories based on level of cognitive ability. The three classifications are artificial super-intelligence, artificial general intelligence, and narrow artificial intelligence.
Artificial super intelligence (ASI) is the concept of a computer, system, or algorithm with cognitive ability that surpasses that of a human. ASI is the most complex and distant category of AI from what we have today. It often receives the most attention in the media because of a lack of understanding about how complex it is to achieve. It will take significant breakthroughs in technology and computing before ASI can become a reality.
One step down from ASI is artificial general intelligence (AGI), sometimes also referred to as strong AI. These applications of artificial intelligence would have a nearly identical cognitive ability compared with a human. This means that these computers, systems and algorithms would have the ability to reason with an environmental outlook on the situation, think in an abstract manner, and develop innovative solutions or theories without obvious precedence towards a certain line of reasoning.
The final category, narrow AI (also known as weak AI), is the only category that has truly been successfully applied in business today. This is AI that is developed and trained with a specific and measurable task to be achieved. Narrow AI can provide substantial automation of routine tasks and is already delivering significant business value today. This form of artificial intelligence provides demonstrative value in environments with strong data management practices in place, and immense volumes of data inputs.
Artificial Intelligence Methods in Action
Within narrow AI, the most significant, accessible opportunities for value lie in conversational AI, machine learning, and natural language processing.
Conversational AI is the field of training computers to converse with human beings in the most realistic and natural context possible. This application is at the forefront of the consumer space in AI, as sales of devices like the Amazon Echo and Google Assistant are only increasing. Another exciting usage of AI that could eventually help streamline phone conversation is Google Duplex. Google has built a tool capable of placing reservations and appointments using a natural tone of voice, with seemingly human-like conversation capabilities. This is a significant advancement that will certainly find use in mobile devices or even call center operations in the future.
Two common applications of AI are machine learning and natural language processing. Machine learning is arguably the most prevalent technique used today. It leverages algorithms and statistical techniques to incrementally improve at a task or goal, using large data sets for training, and continuous feedback loops.
Natural language processing works in a similar fashion, performing complex analysis to extract meaning from volumes of natural language data. GSK is leading the charge in applying these two from a life sciences perspective. In the past two years alone, they have announced partnerships with Exscientia and Insilico Medicine, and are the first pharma organization to join the ATOM Consortium; a data-driven collaborative effort with National Cancer Institute and The Department of Energy. GSK’s goals across these programs include identifying novel small molecules to deliver pre-clinical compound candidates across ten disease-related areas, as well as reducing the timeline from drug target through patient-ready drug to a single year.
The application of artificial intelligence solutions in the drug development process, and life sciences as a whole, will continue at a rapid pace. It is critical for life sciences organizations to quickly begin exploring, evaluating, and developing artificial intelligence capabilities as a driver of strategic and operational value.
Co-author and contributions by Spencer Anderson