How far reaching is the potential of cognitive computing?

Artificial intelligence (AI) is the umbrella term for the set of technologies that make software ‘smart,’ and more technologies are beginning to fall under this umbrella. Cognitive computing is one such subset of AI. Cognitive computing refers to the possibility of creating IT systems that are capable of solving problems without the need of human assistance. Stemming from cognitive science—that is, the study of the conscious mind—cognitive computing aims to simulate human thought processes through self-learning algorithms.

The concept of ‘self-aware’ technology has, for the past 50 or so years, been confined to the realms of science fiction, but is slowly becoming a real possibility. The topic of whether a robot could attain consciousness is still up for debate, but anyone familiar with Stanley Kubrick’s 2001: A Space Odyssey and HAL 9000 will appreciate the risks that sentient AI can pose. Stephen Hawking professed his concerns with creating 'thinking' machines, believing they could redesign themselves at an ever-increasing rate and quickly surpass the biological evolution of humans.

Successfully simulating the brain

Realistically, of course, we are far removed from self-aware robots. In fact, the ‘cognitive’ part of cognitive computing is already stretching our current capabilities with AI.

For us to successfully simulate how the brain functions, we first need to fully understand the brain’s nature and capability. And that is where we reach our first barrier: Our current understanding of how the brain works is surprisingly limited. For starters, current estimates believe there are around 86 billion individual neurons in the human brain—and of those, at least half carry out functions of the brain related to feelings and thoughts. There is currently, however, no safe method of measuring how many neurons are in the brain.

So, the term ‘cognitive’ may be embellishing somewhat, but it remains true that this innovative branch of AI is influenced by ‘human’ processes, and is therefore a big and exciting step forward for artificial intelligence.

The current state of cognitive computing

Cognitive computing involves self-learning systems that use data mining, pattern recognition and natural language processing (NLP) to mimic the way the human brain works. The result would enable automated IT systems to solve problems without direction from humans. But to understand that, we first need to understand the differences between determinism and probabilism—currently the main differentiator in artificial intelligence.

• Determinism—The general concept of “cause and effect.” For every event, there exist conditions that could cause no other event.
• Probabilism—In the absence of certainty, probability is the next-best thing. This is apparent when we either don’t have (or cannot attain) all the information to assess the specifics of an event.

Traditional programmable systems are based on rules that shepherd data through a series of predetermined processes to arrive at set outcomes. That makes them extremely powerful, but deterministic: They can thrive on structured data, but are incapable of processing qualitative data due to its unpredictability.

Cognitive systems, however, are probabilistic—designed to adapt and make sense of the complexities of unstructured information. They interpret information, organize it and suggest explanations and reasoning for their conclusions. The key word here being suggest; they do not offer definitive answers, but weigh information from multiple sources, apply reasoning and then offer a hypothesis.

Cognitive computing is used in numerous artificial intelligence applications, including neural networks, robotics and virtual reality. The most relevant (and impressive) example to date being IBM’s cognitive computer system Watson, which gained widespread attention with its appearance on the TV show Jeopardy! in 2011.

Where do we go now?

Whether it’s uploading 833,333 Dropbox files or watching almost 7 million Snapchat videos, we are creating more data every minute than ever before. The problem is, the majority of that data is unstructured. Cognitive computing is able to turn that problem into potential: finally giving us an effective means to gain insight from the growing sea of data. For an economy in which value is increasingly coming from information, knowledge and services, we can begin to understand the potential cognitive computing possesses.

And IBM may be who to turn to when it comes to cognitive computing’s impact in the future. The technology giant is beginning to build a business in cognitive computing, investing heavily in artificial intelligence, cloud computing and data analytics. A new study, “Your Cognitive Future,” from the IBM Institute for Business Value identified three areas we can expect cognitive computing to have an impact in the near future:

• Engagement—This can fundamentally change the way humans and systems interact. Through taking advantage of peoples’ ability to provide expert assistance and understanding, expert assistance can be provided through the development of deep domain insights and presentation of information in a natural, usable way. Future cognitive systems can act as a tireless assistant, able to consume vast quantities of structured and unstructured data and relay it to us.
• Decision—After the relaying of that information, the next step is decision-making. Those decisions are evidence-based and continually evolve based on new information, outcomes and actions. Cognitive computing systems will be capable of suggesting a set of options to human users, who ultimately make the final decisions. To do so, the systems rely on confidence scores—a quantitative value that represents the merit of a decision after evaluating multiple options—to help users make the best possible choice, including why a particular recommendation was made.
• Discovery—These systems can discover insights that may otherwise go unnoticed. Discovery involves finding insights and understanding the vast amounts of available information across the world. This is a clear need for systems that currently help exploit information more effectively than humans could do on their own. While still in its early stages, some discovery capabilities have already emerged. Advances in this capability can be found already in medical research where robust volumes of information exist.

An intelligent future

The eventual goal for Watson is one in which the system works in tandem with users, performing the analytical heavy lifting while leaving the judgment with humans. With the amount of new data pouring out of the Internet of Things (IoT)—which the company believes will be the largest single source of data on the planet—cognitive computing systems will make it possible to harness this data in real time.