Artificial Intelligence (AI) has evolved from a scientific niche to technology that powers recommendations from Netflix, Google and Spotify to self-driving vehicles. While the current applications from smart speakers to text generators are highly impressive it is still far from the AI future that is promised in science fiction movies.
Artificial intelligence that is self-aware, exceeds the capacity of the human brain and is able to master a wide variety of tasks is called Artificial General Intelligence or AIG. Notable examples are The Terminator, the voice assistant in the movie Her, WALL-E and HAL 9000 in 2001: A Space Odyssey.
The current state of the technology is still light years away from reaching that state. Therefore, today’s AI is called narrow AI, weak AI or even dumb AI. While it can be used to automate very specific tasks like driving a car, recognize a face on a photo or discover trends in datasets, it is limited in its capacity. Your car won’t decide at one day it might compose a music piece, and your autonomous vacuum robot won’t start a fight with the house cat. Not by intention, anyway.
Difference between Narrow AI and General AI
Narrow AI, also referred to as Artificial Narrow Intelligence (ANI), is designed to solve one given problem. General AI, or Artificial General Intelligence (AGI), can be defined as a generalised application of artificial intelligence into any domain. In other words, AGI is used to solve any problem that requires AI.
A narrow AI refers to any AI application that can outperform a human in a narrowly defined and structured task. It is also called “weak AI” and is used to perform a single function like an internet search, face recognition, or speech detection. Because of these predetermined constraints, this type of AI application is called ‘narrow’ or ‘weak’.
General AI is known as “strong AI” and it is where AI, as a technology, is heading towards. General AI allows machines to apply knowledge and skills in different contexts. Narrow AI can thus be defined as where AI is right now while general AI is the future of this branch of computer science.
With narrow AI, the scope is limited to running a single, automated, and repetitive task. With general AI, the scope broadens to creating machines capable of reasoning and thinking like humans but the technology is still in its nascent stage.
Artificial General Intelligence (AGI) or strong AI can be used to solve most of the problems associated with weak AI. Since narrow AI focuses on a single task, the performance of algorithms can degrade with minor changes, but general AI escapes that with its ability to adapt to complex contexts.
Narrow AI: Benefits and examples
Narrow AI is essentially the starter pack towards reaching general AI and, hence, all of its benefits will carry forward to higher intelligence. The narrow AI currently used helps improve productivity and efficiency and also supports smarter decision making. They are also used to improve customer experience. Here are a few examples of narrow AI in today’s tech.
The RankBrain algorithms from Google leverage narrow AI to interpret queries and understand user intent to provide accurate search results. The algorithms are designed to adapt to more queries including support for voice queries using different languages and dialects.
Narrow AI algorithms have proven successful in breakthrough disease detection that are more accurate than healthcare professionals. Since these algorithms focus on a single task and are capable of training on a large set of data, they are capable of processing data faster and allowing healthcare professionals to focus on primary care.
One of the most familiar uses of narrow AI can be seen with apps such as Amazon, Netflix and Spotify. These apps use narrow AI algorithms to recommend shopping items, content or music. The algorithms use data to profile and learn user behaviour and find matching attributes to make accurate recommendations.
Image or facial recognition
Image or facial recognition is another domain where narrow AI is used to help with authentication, indexing images, and tagging photos or videos. However, the facial recognition systems used in police work have been found to have unintended consequences.
Autonomous Vehicles have become the next major battlefield for both tech companies as well as automakers. The likes of Tesla, GM, Waymo, and Uber are fighting to reach a level of autonomy where the vehicle can operate without any human input. The holy grail of this tech, also called level 5 autonomy, is one where vehicles will react or act with intention without human intervention. While impressive, this AI will still only be able to get a vehicle from A to B it won’t master chess, write poems or think of a company strategy. That is why even level 5 autonomous driving is still considered narrow AI.
Experts disagree if AIG will ever become a reality. While some say it might happen in the next few decades, others don’t think humanity will ever be able to build a machine that will outsmart ourselves. While we can’t predict the future, notable figures like Elon Musk and Stephen Hawking have warned that a future with hyper intelligent machines might either be the best or the worst thing that could happen.