Brief history and overview of AI today

- AI is garnering a lot of attention these days. What exactly is it? And the question that you're likely asking, is it destined to make humans obsolete? But first a definition. I like this one from Merriam-Webster dictionary. AI is the capability of a machine to imitate intelligent human behavior. Most other definitions are a variation of this. The idea that human activities can be simulated by a machine, is at the core of AI research and development. It would be easy to believe based on its recent high profile that AI emerged in just the last few years. In fact, AI has been an area of serious research since the 1950s, meaning we've been working on it for almost 70 years. In recent years, improved computing performance and networks, software languages and operating systems, algorithms, and big data have all contributed to an acceleration in AI research and breakthroughs. Popular culture would have us imagine AI in the form of C-3PO in Star Wars or HAL 9000 in 2001 a Space Odyssey. In reality, AI is software that is embedded in devices from our smartphones, to dishwashers, and self driving cars. Yes, and the occasional human looking robot. While AI at some level can appear magical, today it performs a particular set of tasks well. As a result, the most common functions are appropriately called Narrow AI. This differs from other forms of AI, such as AGI or Artificial General Intelligence which is the notion of reasoning and thinking machines, the distinction will be important later on. Fundamentally AI takes existing data, lots of it the more the better, and use a statistical techniques to help software learn how to become progressively better at a relatively narrow task. This type of approach is called machine learning, or ML. It's great at performing activities such as returning results in online search, image recognition such as identifying an object in a digital photograph, recognizing our voice commands made to Siri and Alexa, and enabling self driving cars to avoid collisions. A subset of ML called Deep learning uses advanced algorithms and lots of data to support the concept of learning by example. Specifically show a Deep learning program many pictures of bicycles, and it will soon know what a bicycle is in almost any picture presented to it. Every outcome with ML can be improved and be more accurate through the use of deep learning. Today, AI research and applications are expanding to include areas such as problem solving, perception, planning, and the control of systems for robots. While many organizations are still building their own AI solutions and many will continue to, access to sophisticated capabilities is being simplified by big vendors providing AI as a service. Players such as Microsoft, IBM, Amazon, Google and others are making it easier for AI functions to be embedded in all types of software solutions. As an example, any organization wanting to include a Chatbot and interface between a human and machine that simulates a basic conversation, can simply copy a few lines of code onto their website and get the power of a massive AI supercomputer at their disposal. AI is enhancing many aspects of the human experience. It's often obvious for example, when we use a map to get directions. They can also be less apparent when AI optimizes for fuel and weather in order to route a commercial aircraft to its destination. Despite how remarkable AI appears today, it's important to recognize that we're in the early stages of its potential. In the years ahead, and as you'll see particularly with the emergence of GPT-3, it's real impact is still to be felt. The biggest challenges of AI in the future may not be the technology, but questions that involve ethics, philosophy, culture, regulation, and economics. Will be able to evaluate some of these questions and the potential business and career opportunities of AI through the lens of understanding where it's headed. That's where we'll go next to the story of open AI and the creation of the next leap in AI GPT-3.