Everything You Need to Know About Artificial Intelligence, According to Chinmaya Madan of Toronto
Chinmaya Madan of Toronto Explains the Ins and Outs of Artificial Intelligence
Starting back in the 1950s, the fathers of the artificial intelligence (AI) field, McCarthy, and Minsky described innovation as any task conducted by a machine that would previously require human intelligence to execute. Since this is such a broad definition of the term, arguments have erupted over what is AI and what is not.
Modern definitions of creating intelligence are more specific, citing that intelligence is linked to a system’s ability to improvise and adapt to a new environment while applying its knowledge to unfamiliar scenarios. Chinmaya Madan of Toronto notes that intelligence is the efficiency by which new skills at tasks are acquired.
Generally, AI systems present the following behaviors that are like human intelligence:
- Knowledge representation
- Learning
- Manipulation
- Motion
- Perception
- Planning
- Problem-solving
- Reasoning
Today, AI is used by society daily in simple tasks like recommending what you should purchase next online and understanding your comments to virtual assistants. It’s also recognizing what and who is in a photograph, detecting credit card fraud, and identifying spam.
Chinmaya Madan of Toronto explains that there are two high-level types of AI: narrow AI and general AI. Narrow AI represents the common tasks society experiences daily. These intelligence systems have learned how to perform specific tasks without being programmed how to do them.
Narrow AI machine intelligence is common in speech and language recognition software, vision-recognition systems, and recommendation engines that display products you might enjoy. Unlike humans, these systems can be taught or learn how to efficiently conduct defined tasks, hence the name narrow AI.
The other AI type is general AI which is like the adaptable intellect humans possess. This flexible form of intelligence is capable of learning how to perform a variety of tasks ranging from building spreadsheets to haircutting to reasoning about different topics based on experience.
Chinmaya Madan of Toronto notes this is the type of AI you have seen in movies, like Skynet in Terminator, that do not yet exist today. AI experts are extremely divided on when general AI will become more of a reality.
While there have been countless breakthroughs in AI, some of the highlights include:
- 2009 – Google first released the self-driving Toyota Prius that completed 10 journeys of 100 miles each. This kickstarted the push for driverless vehicles.
- 2011 – IBM Watson won Jeopardy! By beating two of the best players on Earth. To win, Watson utilized natural language analytics and processing across endless repositories of data used to answer human-posed questions, within a fraction of a second.
- 2012 – AlexNet ruled the ImageNet Large Scale Visual Recognition Challenge with an accuracy that halved the error rate of rival systems during an image-recognition contest.
- More recently, Google further refined AlphaGo Zero, a system that played random games against itself and learn from each game.
Chinmaya Madan of Toronto highlights that each year AI continues to improve exponentially eventually reaching the point where they will become smarter than humans.