Not All Artificial Intelligence Is Created Equal

Artificial Intelligence - March 25 2019

Worldwide spending on cognitive and artificial intelligence systems is expected to triple over the next four years, as organizations invest in projects that use cognitive/AI software capabilities, according to a recent report from International Data Corporation (IDC). Spending on cognitive and AI systems will reach $77.6 billion in 2022. That’s up more than three times compared with the $24.0 billion spending forecast for 2018.

Cognitive/AI capabilities can significantly empower organizations across multiple industries and verticals to rethink age-old processes and business methods. Harnessing the power of AI has proven to reduce costs, improve staff productivity and increase profitability. However, with different types of Artificial Intelligence (AI) and each type having its unique capabilities, it can be challenging to select the right AI solution that meets corporate goals and seamlessly integrates with existing infrastructures.

 Often Artificial Intelligence can be confused with other technologies such as Robotic Processing Automation (RPA) which is a rules-based engine. Whereas, Artificial intelligence is an algorithmic method of evaluating data and making predictions based on past results. And unlike AI, RPA does not learn or get better over time the rules are hard coded into the software.

So not only are there different categories of AI, but there are similar technologies available which can make selecting the right solution difficult. Lastly, not all AI is created equal. There are different types of AI suitable for different business applications.

Seven Categories of Artificial Intelligence Explained

  • Machine Learning is the academic endeavour of studying and building predictive models that can learn over time and make “decisions” based on past data sets. There are many different types of training methodologies that are used in machine learning including supervised, semi-supervised and unsupervised learning.
  • Supervised learning uses labeled data with determined inputs to outputs. By leveraging supervised models, the training data set can be significantly smaller to achieve high levels of accuracy.
  • Semi-supervised learning falls between supervised and unsupervised learning. It uses a small amount of labeled data and large amounts of unlabelled data and can produce considerable improvement in learning accuracy over unsupervised learning when the training data set is small.
  • Unsupervised learning uses only unlabelled data to train the model leveraging large amounts of data points to identify patterns and similarities. The benefit of unsupervised learning is that given a large enough dataset it is usually the most accurate.
  • Natural Language Processing (NLP) is a machine’s ability to read language as humans do, pulling out relevant pieces of information, assigning value to those words, and intelligently analyzing structured and unstructured text. Semantic analysis reads all the words to capture the true meaning of any text. It relates syntactic structures, from phrases, clauses, sentences and paragraphs to the level of the writing as a whole to their language-independent meanings. Basically, it analyzes context in the surrounding text and it analyzes the text structure to accurately under the relationships the words.

There are multiple applications for NLP such as:

  • Content extraction which is the ability to read unstructured data stored in data lakes and data stores faster than a human and with greater accuracy. By unlocking the potential to access this data, organizations can make better business decisions and feed downstream systems enabling them to automate traditional high volume, repetitive tasks.
  • Named Entity Recognition is the ability to identify data types such as a person, names, or organizations, locations, quantities, monetary values, and percentages by applying classification techniques.
  • Questions and Answers is the ability to provide information relevant to a specific question and context. For instance, Google is a perfect example where you can ask and receive automated answers on any device. Chatbots are software that take over repetitive, tedious business tasks over text conversation. They are frequently used by sales and marketing professionals on websites to automatically respond to frequently asked questions or to act as virtual sales representatives who can engage with prospective customers 7 days a week, 24 hours a day.
  • Machine translation is the automated translation of languages like English to French. A good example of this application is Google Translate.
  • Automatic text generation is the act of creating a readable and coherent text based on a specific writing style or topic. Automatic text generation can be used for:
    • Automatic document summary generation
    • Automatic article and blog post generation
    • Automatic technical documentation generation
    • Automatic weather reports from raw data
  • Expert Systems are recommendation engines that provide recommendations based on actions. Amazon and other ecommerce sites use this technology to provide shoppers with alternative like-product recommendations based on their browsing history and which items other customers have viewed or bought. Companies like Toronto-based Blue J Legal provides tax and employment foresight to the legal community using artificial intelligence and machine learning to predict legal outcomes. Robo advisors, another type of expert system, a class of financial adviser that provide financial advice or Investment management online with moderate to minimal human intervention. Companies like Wealthsimple provide a service that uses highly specialized software to do the job of wealth managers or investment advisors.
  • Machine Vision and/or image recognition is the ability to automatically extract information from an image. The information extracted can be a simple good-part/bad-part signal, or a complex set of data such as the identity, position and orientation of each object in an image. The information can be used for such applications as automatic inspection and robot and process guidance in industry, for security monitoring and vehicle guidance. Hand writing recognition and OCR fall in this category.
  • Speech includes speech to text and text to speech. These solutions can understand human voices and digitize it. Good examples of this application are GPS navigation systems, Siri and Alexa.
  • Planning AI solutions enable automated prediction planning and scheduling. It involves a sequence of actions that can be delivered to business analytics. These solutions can be uses for Just In Time (JIT) manufacturing planning or for industries that have a high volume of transient or seasonal staff requirements such as during the holiday season.
  • Robotics is the actual application of robots who can replicate human actions. A combination of hardware and software is used to control the robots. Real-world examples of how robotics is being used today is warehouse distribution. Companies are deploying robots to stock pick items stored in warehouses.

What is a Purpose-Built AI Solution?

Purpose-built or better known as point solutions cater to the specific needs of an industry. For example, purpose-built solutions include a niche lexicon, workflows, models and pre-existing algorithms straight out-of-the-box eliminating the need for extensive customization. Implementing a purpose-built solution can be significantly less expensive as several vertical specific AI solutions address distinct use cases enabling companies to meet business objectives quickly and efficiently.

In specific verticals like insurance that is heavily regulated, purpose-built AI solutions for insurance will consume less resources, produce immediate results and provide peace of mind.

For insurers looking to pilot AI for underwriting, claims, sales and service, choosing a purpose-built solution makes it easy to address pain points and implement with existing systems.

AI Reads Like a Human

The benefit of AI is it gets better overtime and a large data set machine can process tens of millions of data points. AI solutions read and break down the data into digestible, understandable information that can be used for to automate high volume, repetitive tasks. Basically, it can read like a human, but faster with greater accuracy.

 

Browse different topics

Recent Posts