Artificial intelligence (AI) might be the next big thing in learning. Seriously. Our excitement about video discs, Second Life, and Google Glass may have been fleeting, but AI is already making inroads. If you’re like me, you want to learn more.

AI sits at the intersection of powerful computer processing, data, sensor technology, and the human desire for more efficient and effective outcomes. At its foundation, AI requires math. A lot of math. Really complex math. And statistics. It requires fast, expensive processing. Graphic processing units (GPUs). But none of this should scare you. I’m convinced there’s a role for each of us in the future AI landscape. You don’t need to be a data scientist to play a part.

A lot of attention in the AI conversation focuses on the kind of AI that would power the Matrix, HAL 9000, or Data (from Star Trek: The Next Generation). But this kind of “strong AI,” approaching general human intelligence, is monumental in complexity, and for the foreseeable future, probably not profitable. While there are some very real opportunities and risks that individuals and organizations should be talking about relative to strong AI, it’s unlikely that you will be adding it to your list of learning offerings any time soon.

Fortunately, there is another kind of AI that does hold considerable promise for learning and performance—and it may be closer than you think. “Weak,” or “specialist AI,” focuses on a single, specific task. Instead of relying on hyper-complex systems to derive the intelligence to perform a whole spectrum of activities, specialist AIs are “trained.” The “rules” for the specialist AI are programmatically created and then applied to a data set. In some cases, repetition is used to improve accuracy. Some pretty smart people think this is where we’ll realize the opportunities of AI in the short-term.

Examples of specialist AIs include facial recognition, object recognition in general, voice recognition, speech-to-text, and intelligent search-results sorting. If you’ve used a modern search engine, with predictive text, you’ve used a simple application of AI. Even prompting your smart speaker or phone to tell you a bad joke is a kind of AI. In general, most AI we directly encounter in 2018 is performed by one or more weak AIs chained together.

Watching the trends

It’s a convergence of technologies and technological advances that enable the continued growth of AI. These currently include, but are not limited to:

  • The Internet of Things: Specifically, sensor technologies that let us collect real-time data on everything from traffic and weather to the temperature on the third rack of our refrigerators are making more and better data available for AI use.
  • Advances in computing and processing power: As we get better with the underlying math of computers, we can apply AI to solve increasingly complex problems. In parallel, computers themselves are getting faster and more accurate, improving the speed of AI.
  • Machine learning: While this term is often used in place of “AI,” it really refers to autonomous learning by machines—one kind of AI. Advances in machine learning also advance AI in general.
  • Natural language processing: The ability to detect, translate, and parse human speech into commands and input has given rise to an entirely new era of computing, enabling us to achieve a level of laziness that doesn’t even require hefting our cellphones to find out what the temperature is just outside our windows.
  • Advances in data science and deep learning theory: Deep learning is getting more efficient and thus faster. Deep Reinforcement Learning is being applied to solve business problems through the same kind of pattern recognition that let AI master gaming strategies and emerge the reigning champion for Chess and Go.
  • Visual processing: Capsule networks, a new type of neural network, process and categorize information in ways similar to the human visual system, leading to a huge reduction in errors.

What does AI look like in learning?

To get to AI, we must get to the point where the computer is making decisions—even simple decisions. In the learning and performance context, AI might be useful for:

  • Suggesting new learning resources based on past learning behaviors and other context clues.
  • Automatically curating resources based on utilization and ratings by specific groups.
  • Presenting curated resources based on user patterns in either learning or work contexts.
  • Driving virtual agents for specific domains, such as on-boarding, customer service, or policy. (Chatbot technology such as Alexa, Google, Skype, or Cortana may be a good vehicle for these.)
  • Voice note transcription, especially for self-paced learning. Once learners get used to taking notes by voice, it may be faster and less distracting than typing.
  • Using recognition technology to verify learner identity and queuing up appropriate content based on learning history or preferences.

You don’t need a robot to take advantage of AI

Anyone can learn more about AI and how to integrate it into solutions. Your involvement might range from suggesting an enhancement to integrating learning solutions using APIs (application programming interfaces). Here are five key questions to ask yourself as you plan your foray into learning more about AI:

  1. What will your role be in relation to AI and learning?

    A good way to start mapping out your own AI learning path is to think about your potential role with AI, in relation to learning. Do you do the hands-on development of learning solutions? Do you define curriculum objectives and hand off the development to someone else? Or do you function more strategically, looking at the big picture of learning and performance and tracking trends?

  2. Is your organization or industry already involved in AI?

    If so, how; what tools are being used? Is there a centralized nature to enterprise involvement with AI? Is there a stated direction? If your organization hasn’t ventured into AI yet, what other areas will be interested in AI? How would they potentially use it? Who would likely lead the adoption and integration of AI technology? Is there a standard emerging in your industry?

  3. Do you want to build AI solutions yourself (hands-on with technologies)?

    There are a couple of ways to approach this. At least a little programming will be required (at this point in the evolution of these technologies). Check out the resources at the end of this article for some starting points.

  4. What learning problems could you address with AI? What modalities might you want to pursue?

    Look at the top trends and buzz in the AI space; then think about your specific learners. Some ideas are sure to surface. Use your ideas as a proving ground for your learning journey.

  5. Are there specific, data-intensive human tasks in your organization that could be performed better if part of the information processing was outsourced to computers? (decision support)

    While this may seem like a huge question, it’s worth considering. The answer can lead you to understand where the greatest impacts for AI might be for your organization. As well, the answers may lead to clues about others in your organization who might also be thinking about AI and may be willing to partner with you.

So, You Want to Build Stuff?

If you want to create your own AIs, chatbots are a good place to start because each of the major platform vendors (focused on capturing market share) makes available an assortment of tools, tutorials, resources, and frameworks to get you started. Each has their strengths and weaknesses.

Here are some learning resources about artificial intelligence worth considering. Some are geared toward the more tech-minded, but just about anyone could benefit from the content in the AI Basics group. We’d love to hear about your own suggestions in the Comments as well.

AI basics

  • Machine Learning: Andrew Ng (Stanford University) course on Coursera. Topics include broad intro to machine learning, datamining, statistical pattern recognition, supervised and unsupervised learning, and best practices.
  • Machine Learning for Humans”: Vishal Maini article on Medium.com. An overview of machine learning with a nice set of learning references in the appendix.
  • Artificial Intelligence: Ansaf Salleb-Aouissi (Columbia University) course on edX. Topics include intro and history, intelligent agents, machine learning algorithms, applications of AI, and Python.
  • Introduction to Artificial Intelligence (AI): Microsoft course on edX. Topics Include high-level overview, Microsoft Azure, Python, MS Bot framework.

Programming AI and machine learning

  • Python for Data Science and Machine Learning Bootcamp: Udemy course focusing on data science through the Python programming language and related tools.
  • Learn with Google AI: a collection of tools and resources from machine learning experts at Google. Emphasis on the TensorFlow API, an open-source machine learning framework.
  • Amazon Rekognition: Deep Learning-based image and video analysis. The service can identify objects, people, scenes, activities, and more. Amazon claims it can even identify inappropriate content. Includes facial recognition capabilities for user verification and more.
  • H2O: Open-source machine learning for enterprises.
  • Skytree: Machine learning platform.

Chatbot resources

As you dive into artificial intelligence, focus in on one area at a time. It’s too big a domain to master in a day, a week, or a month. But with a little persistence and focus, you can identify and pursue a small slice of the AI space, and use it to produce meaningful solutions for learners.