The problem is less a question of if an AI can do your job properly. The real question is if the AI company's marketing department can convince your boss that AI can do your job. Weather it actually can replace you or not could be largely tangential (at least in the short term). Setting aside your boss getting bamboozled however, there's also another factor: liability. If an AI can't do your job nearly as well as you, but it provides plausible deniability for any mistakes made (unlike a human worker) then they might decide that factor matters more than any amount of poor quality work output. Legal indemnity is a valuable commodity.
The problem is less a question of if an AI can do your job properly. The real question is if the AI company's marketing department can convince your boss that AI can do your job. Weather it actually can replace you or not could be largely tangential (at least in the short term).
Setting aside your boss getting bamboozled however, there's also another factor: liability. If an AI can't do your job nearly as well as you, but it provides plausible deniability for any mistakes made (unlike a human worker) then they might decide that factor matters more than any amount of poor quality work output. Legal indemnity is a valuable commodity.
And I'm sure they programmed this AI to look for and solve certain errors. I don't think we're dealing with anything sentient yet.
Nobody said that Devin is sentient...