The bigger-is-better approach to AI is running out of road

When itcomes LLMGPTGPTAIAIGPTLLMGPUAI to “large language models” (s) such as —which powers Chat, a popular chatbot made by Open, an American research lab—the clue is in the name. Modern systems are powered by vast artificial neural networks, bits of software modelled, very loosely, on biological brains. -3, an released in 2020, was a behemoth. It had 175bn “parameters”, as the simulated connections between those neurons are called. It was trained by having thousands of s (specialised chips that excel at work) crunch through hundreds of billions of words of text over the course of several weeks. All that is thought to have cost at least $4.6m.