Amazon touts its low-cost cloud computing in generative AI race

A key way that Amazon's cloud division aims to set itself apart from rivals perceived to have a leg up on artificial intelligence is by competing on price, an executive said on Tuesday. The AI models behind a viral chatbot like ChatGPT require immense computing power to train and operate, the kinds of costs Amazon Web Services is good at lowering, said Dilip Kumar, vice president overseeing its applications group. On quality, Kumar did not answer how Amazon's own family of AI models known as Titan stacks up against its more famous counterparts such as the GPT series from Microsoft-backed OpenAI or Google's PaLM. He instead pointed to other Amazon traits, such as "our specific way of dealing with privacy, our specific way of dealing with accuracy," at a time when concerns abound about what happens to confidential data given to AI and the technology's tendency to generate incorrect information. In addition, as the cloud industry's biggest player, "more companies of all sizes have data already in AWS," he said, making it a reason to use its AI.