Some workers worry that generative AI will take their jobs, but before that happens, it looks like it could make some software engineering tasks a lot easier.
Amazon CEO Andy Jassy said in a LinkedIn post on Thursday that the company has integrated its AI-generated assistant, Amazon Q, into its internal systems and updated the underlying software.
Jassy said the results were a “game changer.”
“The average time to upgrade an application to Java 17 has been dramatically reduced from the typical 50 developer-days to just a few hours,” he wrote. “We estimate this has saved the equivalent of 4,500 developer-years (yes, that's a crazy number, but it's true).”
According to his post, the AI is not only fast, it also appears to be pretty accurate: Amazon developers shipped 79% of AI-generated code reviews without any additional changes, Jassy wrote.
“Beyond the amount of developer effort saved,” he said, “the upgrades will improve security, reduce infrastructure costs, and generate estimated efficiencies of $260 million per year.”
Jassy's comments seem to tie into a common argument about artificial intelligence: that it can help free up time for boring but necessary work.
“One of the most tedious (but important) tasks for software development teams is updating the foundational software,” he writes. “This task is either loathed, or put off for more exciting work, or both.”
But while Amazon enjoys productivity gains, developers may worry that AI's efficiency will reduce the need for human workers.
Matt Gurman, CEO of Amazon Web Services, said in a recent company meeting that software engineers may need to develop other skills given the pervasiveness of AI in coding.
“If we look 24 months from now, or even further out – I can't predict exactly when that will be – it's possible that most developers won't be coding,” he said.
And Gurman isn't the first top executive to voice this warning: Nvidia CEO Jensen Huang said last year that thanks to AI coding tools, “everyone's a programmer now.” Former Stability AI CEO Emad Mostakeh even predicted that “in five years, there will be no programmers.”
Developers aren't the only ones facing extinction: In a now-deleted tweet, Klarna CEO Sebastian Siemiatkowski said fintech companies could save $10 million this year by using generative AI to handle marketing tasks previously performed by human employees.
“We are spending less on photographers, image banks and marketing agencies,” he wrote. “Our in-house marketing team is half the size it was last year, yet producing more!”
While Siemiatkowski faced some fierce backlash, the press release backed up his enthusiasm for AI, stating that Klarna “used genAI to generate over 1,000 images in the first three months of 2024, reducing image development cycles from six weeks to just seven days.”
U.S. Bank's chief marketing officer, Michael Lacolazza, previously told Business Insider that using generative AI enabled the bank to “shave about two and a half months off” the development cycle of a new brand campaign.
Despite the impressive efficiency gains, Lacolazza reassured employees that he sees generative AI as “an aid to humanity, not a replacement for humanity.”
Meanwhile, Jassy said Amazon will continue to use Amazon Q.
“Not only will the Amazon team be leveraging this conversion capability further, but the Q team will also be adding more conversion capabilities for developers to leverage,” he said.
