Artificial intelligence (AI) has provided a critical competitive advantage for those organizations able and willing to use it. AI has gained significant momentum in the last few years, acting as personal assistants for some, while processing business transactions and providing technical services to others. AI systems have the ability to manage large amounts of data in a number of ways. Different types of artificial intelligence have been evolved to handle a variety of tasks, ranging from facial recognition to drug design to driving cars.

In terms of logistics, an AI can optimize the routing of delivery traffic, thereby improving fuel efficiency and providing faster delivery times. It has become a valuable response tool, providing customer service centers with a phone answering service. In the world of sales, combining customer demographics with past transaction data and social media can result in recommendations tailored to the customer. An artificial intelligence can improve predictive maintenance, analyzing large amounts of data from images and audio to detect anomalies in auto engines or assembly lines. Specific deep learning techniques can be used to tailor an AI for accomplishing specific goals and tasks.

AI & Deep Learning

Currently, artificial intelligence is trained through the use of machine learning and deep learning. Deep learning algorithms require massive amounts of data to learn from, and the increase in available data is one reason deep learning abilities have expanded in recent years. Additionally, deep learning algorithms are helped by the stronger computing power available today. “Artificial Intelligence as a Service” has offered smaller organizations access to AI technology and algorithms, supporting deep learning training without a large investment. Deep learning architectures have been used to train AI for:

• computer vision
• speech recognition
• natural language processing
• social network filtering
• bioinformatics
• machine translation
• drug design
• visual inspections
• medical image analysis
• board game programs

Narrow and General AI

Narrow AI is a type of AI that outperforms humans in narrowly defined tasks. Self-driving cars and facial recognition are two examples of narrow AI. Many businesses have invested in narrow AI to improve efficiency, reduce costs, and automate various tasks.

General AI is an ideal. It describes an artificial intelligence capable of applying experience and knowledge in different contexts. It is modeled after human intelligence and supports autonomous learning and problem-solving. General AI has not yet been achieved, but is gradually coming closer as a reality.

Deep Learning and Generalization

Deep learning models use multiple-layered artificial neural networks. These neural networks screen for  patterns through the various layers of abstraction. The abstractions contained in the layers are the reason deep neural networks are preferred for dealing with large and complex amounts of data.

Generalization is a form of classification where all, most, or some of a particular group carries characteristics that are common to the group, or, put another way, generalization is the process of identifying the parts of a whole, as belonging to the whole. For example, a bird falls into the category of animals (a part of the whole). A flower does not. The process of generalization includes abstraction (reducing something to its essential characteristics). A bird chooses to move, animals choose to move. Flowers do not have the option of choosing to move. The concept of generalization also applies knowledge from previous experiences to new circumstances, or thinking beyond the original problem and making predictions.

Generalization is a form of broad recognition that relies on a few characteristics to identify things. For example, if something is moving in a deliberate way, it gets recognized as an animal. If it’s green and swaying in the breeze, it gets recognized as a plant.

Training deep learning AIs to recognize a photo of a bird uses the lowest layers to identify edges and light/dark gradients, etc. The higher levels learn how to combine those into patterns. Higher levels can learn how patterns combine to make recognizable forms, and combine forms to recognize other animals. The more characteristics used, the more accurate the recognition and identification.

How Artificial Intelligence Is Being Used

According to Amir Husain, CEO and founder of SparkCognition, a machine learning company:

“Artificial intelligence is kind of the second coming of software. It’s a form of software that makes decisions on its own, that’s able to act even in situations not foreseen by the programmers. Artificial intelligence has a wider latitude of decision-making ability as opposed to traditional software.”

Some of the different uses for AI include:

Big Data Research: Artificial intelligence can help us make sense of massive amounts of data, including unstructured data. AI has helped organizations find new insights that had been locked away in stored data. Hidden within the data lies the potential to develop amazing businesses and resolve some of the world’s largest challenges.

Customer Management Systems: Artificial intelligence is being used to alter customer relationship management systems. Some software systems, such as Zoho or Salesforce, require significant human maintenance to remain accurate. However, when an AI is applied to these platforms, they are transformed into auto-correcting, self-updating systems that efficiently store and manage data, without constant glitches.

In the Classroom: A promising innovation is the concept of a personalized AI tutor for each student. Because a single teacher cannot work simultaneously with every student, an AI tutor would help students to get extra help in areas where they need it.

Aviation: The AOD (Air Operations Division) uses AI for training purposes. Artificial intelligence is currently being used as mission management aids, combat and training simulators, and support for tactical decisions. Airplane simulators use artificial intelligence to process data taken from simulated training flights, as well as simulated aircraft warfare.

Self-driving Automobiles: AI driven cars and trucks are not yet an option. Nalin Gupta, the Director of Business Development at Ridecell, stated, “Safety is crucial when it comes to autonomous vehicles, and for the public to embrace AVs, they have to be safer compared to human-driven vehicles.” He was referring to the tragic accident in 2018, when an autonomous Uber killed a pedestrian, and the incident with Jeremy Banner, who died in 2019 while engaging the autopilot feature of his car.

Financial Trading: Several banks and proprietary trading firms currently have entire portfolios being managed by AI systems. Additionally, complex AI systems are used in “algorithmic trading.” They make trading decisions several times faster than humans are capable of, and can make millions of trades per day without human intervention. This is referred to as high-frequency trading, and represents a fast growing sector in financial trading. 

Sensors: AI has been combined with a variety of sensor technologies supporting both smart cities and several manufacturing industries. Sensors are included in the IoT (the Internet of Things) and are used to collect data that the AI processes and uses for decisions. Sensors can be used to monitor such things as traffic flows, when lighting is needed, problems with a conveyor belt, and even available parking.

Hospitals and Medicine: Artificial intelligence now helps people with diabetes to regulate their blood sugar. AI automates prescription refills and connects call center customers with the person most qualified to answer their questions. As algorithmic advances, computing power, and data proliferation continue to evolve, the variety of opportunities continue to expand. They are predicted to include:

• Design treatment plans
• Big data research — mining medical records to provide more useful information
• Companion robots to care for the elderly
• Predicting the likelihood of death from surgical procedures
• Heart sound analysis
• Drug creation

Personal Finance: Products have been developed that use artificial intelligence to help people deal with their personal finances. Digit, for example, uses an app powered by an AI that automatically helps consumers optimize their spending habits based on personal goals and behaviors. The app analyzes  monthly income, spending habits, and current balance. It then makes its own decision, and may transfer money to the savings account.

Dr. Hossein Rahnama, the CEO and founder of artificial intelligence company Flybits, offered another example of AI versatility:

“Using this technology, if you have a mortgage with the bank and it’s up for renewal in 90 days or less … if you’re walking by a branch, you get a personalized message inviting you to go to the branch and renew purchase. If you’re looking at a property for sale and you spend more than 10 minutes there, it will send you a possible mortgage offer.”

Image used under license from Shutterstock.com