An application-specific integrated circuit (ASIC) is a chip designed for one particular task rather than general-purpose computing like a CPU. In AI, ASICs are tailored to the arithmetic of neural networks – Google's Tensor Processing Units (TPUs) are one example. For that task they run faster and more energy-efficiently than general-purpose chips, but they are expensive to develop and hard to adapt to new workloads.
Neuromorphic computing is an approach to hardware and software modelled on the structure of the brain. Instead of moving data between separate memory and processor in clock cycles, artificial neurons and synapses work in parallel and respond only to events known as spikes. This promises very low energy consumption, for example in sensors or robots. Well-known research chips include Intel Loihi and IBM TrueNorth; broad commercial use is still at an early stage.
Deep learning is a branch of machine learning that uses artificial neural networks with many layers. Such networks learn from large amounts of data which features matter – from edges in images to relationships of meaning in texts. Deep learning underpins today's language models, image and speech recognition and machine translation, and usually requires considerable computing power and training data.
Artificial neural networks (ANNs) are computational models made up of many connected units (“neurons”) arranged in layers that process inputs via weighted connections. During training, the weights are adjusted step by step using examples, usually by backpropagation. ANNs are the foundation of deep learning and power image recognition, speech recognition and large language models. Despite their biological inspiration, they work far more simply than the human brain.
