Do you understand the difference between Artificial Intelligence and Machine learning? Interested to learn? Great, sit down or stand up and take a moment to read this short blog and we will provide our explanation of the difference between Artificial Intelligence and Machine learning. And, while we on the journey, we will take a moment to explain the TonkaBI approach to Ai?
Artificial Intelligence is two words which have two separate meanings, artificial meaning made by people, often as a copy of something natural or something that was created unintentionally. Intelligence meaning the ability to learn understands and makes judgments or have opinions that are based on reason. Sometimes, Artificial Intelligence in the computer programming world is misunderstood as a system or an application but this is not true, Ai is the study of how to train computers to perform tasks that share characteristics of human intelligence. Ai code can be embedded into applications, systems or devices this does not mean theyre Ai it means theyre augmented by AI.
Machine Learning, lets again look at the words separately. Machine – a piece of equipment to perform do a particular type of work. Learning – the activity of obtaining knowledge by studying it or by experience. ML in computer programming is when the machine can learn on its own without being explicitly programmed. Machine learning algorithms are used in sorting high volumes of data. Machine learning also supports Ai development by reducing the amount of hard rule based code, without ML you would have to write hundreds maybe thousands of lines of code within the Ai algorithm.
A computer (program) can gain the ability of Ai with little training and with a small amount of data, but this Ai will lack precision and intelligence in analyzing contrasting data e.g. computer vision. The Ai could identify a car door from selected images but not a car door from various angles or what side the car door is, and, it would not be able to identify a car door from random never seen before image data.
Ai can further lack intelligence through the wrong or weak algorithm and engineering ingenuity. A true Ai algorithm should perform its deigned task with high accuracy and scalability and outperform human counterpart at that given task, example AI image processing and classification in analyzing vehicle damage faster, more accurate, multiple cases, and 24 hours per day better than a human can manage.
TonkaBI uses two examples for the types of Ai we develop and use. These explanations and definitions may not be academically correct, but these are the terms we use in-house and when explaining Ai and the type of Ai we use when communicating to our partners and clients.
Narrow Ai For TonkaBI this means the Ai is a decentralized standalone series of code that is embedded into client customer systems and processes. Narrow Ai does not have the ability to learn from data or its mistakes or its correct choices. It just performs a set task, day in day out to an acceptable standard. The Ai does not need internet access to work; neither does it require APIs. This Ai provides many businesses with the ability to have Ai technology in remote situation, embedded on hardware or environments where access to the internet is unachievable. The Ai also provides many businesses with a “Good Enough” approach to many situations. TonkaBI can update its Narrow Ai, which maybe embedded into client system, through “swapping” new code for the old. The new code would have been further trained with better understanding accuracy features etc. This Ai grows in steps as required for clients.
Active Ai Similar to Narrow Ai but with a big difference, the Active Ai has a parent (a teaching model) that corrects mistakes and supports change and growth. This means the Active Ai can learn from data and gain knowledge that would be otherwise be forgotten or lost on the fly, in real time. So, the more data the Ai consumes the better it gets. Active Ai needs access through the internet or a network to the TonkaBI parent teaching module.
TonkaBI fully supports both Ai models and agrees there are advantages and disadvantages to both approaches. Our way of working and providing modern solutions to businesses has come from understanding our clients what to have the code embedded in their own systems, ones they control and manage, our businesses model comes from listening to the market.