Artificial Intelligence (AI) continues to experience considerable growth. According to the Gartner 2019 CIO survey, enterprise adoption grew 270 percent over four years. Artificial intelligence is primarily used for research and development but has practical applications across multiple sectors.
A 2019 O’Reilly survey reveals AI is used for information technology (IT), customer service, marketing/advertising, facilities/operations management, and more. A high percentage of respondents (85%) report they’re using AI in production. While adoption rates are incredible, they could be improved if the AI skills gap were narrowed.
Organizations experience a range of barriers to adopting artificial intelligence. Among the many obstacles is a lack of AI-skilled staff. Retraining the existing workforce is one way of closing the skills gap.
Some companies are already investing in AI training for existing employees, and workers without access to a company-sponsored training program can still learn the necessary AI-specific skills required to succeed in the field. (Read also: Tech Career Shift: 5 Factors to Consider.)
Jumping into artificial intelligence requires a strong background in mathematics and computer science. Tom Taulli, author of Artificial Intelligence Basics: A Non-Technical Introduction, says this about getting into the field,
“In terms of getting started with AI, it is important to get a refresher on basic statistics. Keep in mind that the technology is generally about probabilities and predictions. This means understanding standard deviations, bell curves, Bayes’ theorem, and so on.”
Strong mathematics and computer science skills are a must, but having a goal will also affect your learning path. Andrea Beckman, Director of Product Management at Relativity, recommends choosing a problem you would like to solve:
“AI is a broad field, so to get started learning I suggest identifying an initial problem or two that you want to tackle. Taking this first step will help narrow down the AI technology solutions you’ll want to investigate further. Ideally, you can see demos and potentially trial those solutions to get a deeper understanding of the AI options available to solve your unique challenges. Starting small and having hands-on experience with AI will accelerate your learning and enable you to understand how you can use this powerful technology to augment and amplify your human efforts and expertise.”
A Very Brief History of Artificial Intelligence
Artificial Intelligence is a broad field. Contained within are multiple areas of study in both practical and theoretical applications. Understanding the origins of the field and how it evolved over time is an important aspect of learning how to wield it.
Artificial intelligence was born of a fundamental question: “Can machines think?” The concept of autonomous machines, robots even, was covered in works of science fiction prior to the 1950s.
Artificial Intelligence dates back to the 1950s, a time when computing technology had advanced enough to allow computers to store commands. Early computers could execute commands but were unable to store data. Computing time was also quite pricey in those days.
The 1960s brought about the creation of ELIZA, a natural language communications program. Originally programmed to mimic casual conversation, ELIZA was an early chatbot. The program could respond to typed messages from a pre-programmed script based on words provided by the participant. AI development slowed through the 1970s, a time known as the first AI winter.
In the 1980s, early knowledge systems known as expert systems grew in popularity. These programs could provide answers to questions by sifting through saved information based on an operator’s query. They were intended to provide information as a human expert might.
By the late 1990s and into the early 2000s, artificial intelligence had already met important goals envisioned by its creators. (Read also: AI: Older Than Pizza)
Michael S. Gashler Ph.D. from the University of Arkansas, Department of Computer Science and Computer Engineering sees where the future for AI applications lie:
“As AI applications move toward mobile devices, people with skills in sensors, wearable computing, and human-computer-interfaces will be needed. Since AI is immensely compute-intensive, people who know how to parallelize using general-purpose graphical processing units and cloud systems are also needed. And people who are familiar with the domains in which artificial intelligence is being applied play an important role in helping with the transition to building more fully-automated systems.”
Three Kinds of Artificial Intelligence
There are three types of Artificial Intelligence:
Contained with artificial intelligence are Machine Learning, Neural Networks, and Deep Learning. The three are described as a subset of the previous, creating a neatly nested grouping of technologies. Eda Kavlakoglu, Program Manager at IBM Cloud, explains
“Deep learning is a subfield of machine learning, and neural networks make up the backbone of deep learning algorithms.”
Programming Languages for Artificial Intelligence
Artificial intelligence practitioners must be skilled computer programmers. A number of programming languages are used for artificial intelligence. Python and R are among the most popular languages for Artificial Intelligence practitioners. Other popular languages include Java, Prolog, Lisp, Haskell, and Julia.
Framework Libraries and Toolsets for Artificial Intelligence
A growing number of libraries and toolsets are being used in artificial intelligence. TensorFlow and Py Torch are among the most popular. TensorFlow and PyTorch are machine learning frameworks that are used for deep learning algorithms in neural networks. (Read also: Artificial Neural Networks: 5 Use Cases to Better Understand.)
Google AI tools include hands-on labs, toolsets, and a machine learning kit for Android. While Facebook AI offers a range of tools, frameworks, and libraries for a variety of machine learning and natural language processing.Amazon Web Services (AWS) Machine
Learning Artificial Intelligence
University-level courses can provide a great foundation for learning about AI. Many universities offer degrees in artificial intelligence and related fields. Program formats will vary. For example, some universities offer online courses but still require in-person classroom attendance. (Read: Computer Science: Top 5 Online Undergraduate Degree Programs.)
However, university degree programs may not be suitable for students who work full time or are unable to attend university in person. This is where online courses and certificate programs excel. There are also plenty of free courses available.
While this isn’t an exhaustive list, it can serve as a starting point for exploring artificial intelligence coursework.
Kaggle Learn offers micro courses in Python, machine learning, deep learning, natural language processing and more.
EdX offers artificial intelligence programs and individual courses on the applied and theoretical uses. This material covers introductory and more advanced topics.
Google AI offers courses, labs, and datasets to teach machine learning.
Udacity offers technical and business courses on artificial intelligence. Students can choose from a list of specializations.
Coursera offers a wide variety of courses, specializations, and professional certificate programs. IBM is among the artificial intelligence professional certification content providers.
Google Developers Machine Learning is an introductory course on machine learning with an emphasis on fast paced learning modules. This course requires mathematics (trigonometry, calculus, statistics) and Python programming skills.
Practice with Open Datasets and Compete in AI Challenges
As with any new skill, practice is essential to reach proficiency. Access to heaps of data is as well. Fortunately, there are a number of datasets and AI challenges available for skill-building.
Data.gov Datasets offers more than 200,000 open federal and state datasets.
Samasource, an AI data training firm, published a list of 13 Open Source Datasets for Machine Learning.
IBM Data eXchange (DAX) is a collection of open data sources suitable for audio, language processing, image, and more.
Data Science Central also has a list of deep learning data sets that may be of interest, with a variety of fields represented.
Data science competitions are a great way for data scientists to participate in interesting challenges and build essential skills. Participants compete for recognition and prize money (in some cases).
The opportunity for artificial intelligence to profoundly affect everyday life is immense. The future will need hard-working practitioners to solve simple and complex problems across industries. Now is an excellent time to learn AI. The sheer volume of resources available to anyone with an Internet connection is great and the opportunities are approaching endless!