Summary of "Как хоронили программистов и порвали два баяна"

Overview

The video argues that predictions of programmers being “replaced” by AI are already partially true in practice—but also deeply paradoxical.

Key Claims

AI will automate coding tasks soon: Dario Amadei (Antropik) reportedly told the World Economic Forum in Davos that within 6–12 months AI could handle most (or all) tasks developers do, and that the share of machine-written code in companies could eventually reach 90%.
In Antropik, developers already write less code by hand: Team leads reportedly shifted from manual coding to an “editor” workflow—reviewing, validating, and refining AI-generated code. One engineer is quoted as saying they no longer write code and only edit what the model produces.
Concrete example—Cowork: A product (referred to as “Cowork”) is described as turning a neural network into an assistant for non-technical tasks, such as organizing files, preparing reports, and analyzing documents. Its creator says the neural network generated most of the code, and the product reportedly went from idea to a working prototype in about 1.5 weeks, far faster than before.
The contradiction: fewer developers needed vs. high hiring demand: Despite Amadei’s claim that the future requires fewer developers, the video cites that Antropik has hundreds of open positions (about 450 total, ~130 engineering) and offers extremely high salaries—up to $400k, potentially $500k with bonuses. The video frames this as ironic: the company most loudly predicting the death of programming is also hiring and paying programmers heavily.
Why programmers still matter: The video explains that building and improving AI systems isn’t fully automatable yet. Someone must still handle task definition, architecture decisions, security, and integration, even if day-to-day coding becomes more AI-assisted.
Skepticism about AI replacing true junior programmers: The creator of Ruby on Rails, David Heinemeier Hansson, is cited for arguing that current AI coding tools don’t reliably match the consistency of real junior developers. Junior roles are portrayed as important for training—learning to think like engineers, understanding systems internally, and learning from mistakes. If that stage disappears, fewer people may exist to review and validate AI-written code in the future.
Conclusion: The market is moving in two opposing directions: AI threatens to reduce manual programming labor, while demand and compensation for developers are rising. The video concludes it’s too early to “bury” programmers; their roles will likely shift toward oversight and orchestration—more like conductors than assembly-line workers.

Presenters / Contributors

Dario Amadei (Antropik)
Boris Cherny (creator of the CLD-code system; described as confirming neural net wrote most of Cowork’s code)
Felix Riziberg (product manager; stated the timeline to prototype)
David Heinemeier Hansson (creator of Ruby on Rails; cited regarding AI tool reliability vs. junior programmers)
Ai Venga (video narrator/host)