We are on the cusp of an AI revolution. All smart machines, cars and TVs are AI-assisted. AI is basically based on learning specific patterns from the data on hand. To illustrate, there is an image of a bus. AI can interpret the image in terms of its various attributes — the length, the height, shape and features such as wheels, doors and steering. In traditional softwares, a machine is given a set of instructions as a programme and data. On this basis, the machine generates output. In AI, things differ. Here both the inputs and outputs are provided enabling the machine to master-learn the programme used originally to generate the output. It is called training the computers in the dialect of AI. The whole programme is treated as a model. Post-training, one can give just input to the machine and ask it to generate the output.
AI models are broadly of two types. Discriminative models accept certain data as input and provide certain output, say image of two animals, such as donkey and horse and provide the class of the image as an output whether its donkey or horse. The machine is here trained to discriminate between images. Generative models learn the process of generation itself. A random sound of a roar and the machine generates a class of objects, say a tiger or lion.
Theses days generative AI models have gained traction. There are image generation models or music generation models. Dall E generates image from a descriptive text. It is from OpenAI.
Models are used to develop text-based applications such as chatbots, text generators, jokes, poems and essays . These models can be made responsive to queries which would be answered. There are models which generate computer codes, accepting natural language as the input.
There are transformer-based language models. They are based on natural language description of the problem, e.g. ChatGPT.
New models are being created in areas of text, speech and vision. Your engineering or technology background helps to learn AI but one can upskill by taking certification courses too.