"AI-generated and approved by engineers" is doing a lot of work there. If accepting a 4-character Gemini autocomplete counts, Copilot users hit >90% last year. The useful metric is % of functions where >50% of the body was AI-written before human edits. just my 2 cents
There is a big difference between just “vibe coding” without reviewing the code and actually knowing what you’re doing and checking everything the AI produces.
In my opinion, humans don’t really need to write code anymore. The main skill is steering AI models.
One possibility is that most of the code that matters is being written by hand, while enormous amounts of code are being generated for other things. People are being evaluated on their AI usage after all.
Our engineers are orchestrating fully autonomous digital task forces, firing off agents and accomplishing incredible things.
Who swallows this stuff? Like all marketing it's full of weasel works and unverifiable claims. I don't doubt 75% of their code started life as the output of a model mind you, it just "forgets" to mention the subsequent hours of humans put into reviewing, discarding the crap, and refining what was left.
Edit: it's slop. How did I not spot it was slop posted by the marketing department immediately? If they want engineers to pay any attention they will have to do better. Or maybe they are targeting engineers any more? blog.google used to be a reliable read.
Big tech is inflating the AI bubble.
Because of that, I find it hard to fully trust these kinds of claims — especially from complex systems like Google.