Although Alexa can sound like a miracle to children, AI is limited currently and in the foreseeable future to weak AI:
-a system that can be trained to accomplish specific tasks through the execution of an algorithm. And so let us show you how this works.
An algorithm is a rule. A bit like when you teach your child multiplication rules. 3 times 9 is 27. Input this into a calculator and you’ll get the same result.
To create an algorithm:
-You take a whole lot of data that is labelled (supervised learning), pass it through the algorithm and compare it to the correct answers. (The algorithm will learn from its mistakes (“loss function”)
-Progressively trained models figure out the optimal set of rules for determining what features match the data with the answers.
-The model is then tested by taking new data (test set) that the model hadn’t seen before and asking it to make predictions by applying what it learns from the dataset to see how it performs. And that’s it.
Given an appropriate dataset, we can create a model using this for any tasks that you can imagine.
Reference: Justin Ko, MD. AAD Position Statement on Augmented Intelligence. Fusing technology with human Expertise to enhance Dermatological Care. 8th World Congress of Teledermatology, Skin Imaging and AI in Skin diseases – November 2020