Почему работать программистом больше не интересно — Влад Тен — Мы обречены

Summary

The episode explores how generative AI is changing software development—and why some experienced programmers find the change demoralizing. The speakers argue that the concern is not simply job loss: AI can take over the parts of programming they enjoyed, while leaving people to coordinate tools, review output, test it, and manage the work.

AI coding tools and the changing developer role

• The speakers distinguish between a coding assistant, which suggests or explains code, and an agent, which can take action across a project and carry out tasks. They imagine that a modern coding agent could have greatly reduced the time spent mentoring interns, answering routine questions, and maintaining a website.
• They discuss GigaCode and GitVerse as an example of an AI assistant used alongside a shared code repository. The conversation describes connecting an assistant through a plugin, using different available models, and keeping a team’s code and collaboration in one place. They also discuss the possibility of running such tools locally for organizations that need their data kept on their own infrastructure.
• Their central caution is that faster code generation does not necessarily mean a better development process. Developers may have to hold more generated code in mind, wait on agents, and inspect large changes to find errors. They worry that the balance is shifting from engineering quality toward producing more features more quickly.
• The discussion suggests that system design and the ability to direct and coordinate agents may become more important. ML engineering is not automatically protected: routine scripting and data-processing work may be automatable, while research and work requiring deep domain understanding are described as harder to replace. The speakers also note that AI can perform poorly in specialized domains with limited training data.
• One speaker sees this as a loss of professional identity as well as a change in tasks. Programming used to offer opportunities to build tools, invent frameworks, and improve systems directly; they fear that AI-assisted work may leave less room for that kind of craft.

AI agents for customers and business use

• A guest from Avito describes experiments with agents in customer support and seller workflows. Examples include answering common questions, helping a seller respond while they are busy, and completing simple service bookings without a customer having to wait for a person.
• The guest says users are becoming more willing to use agents when they can resolve a request directly. This differs from older support bots, which often frustrated users and could not actually solve the problem.
• The economics remain a constraint. Powerful models can be expensive at high volume, but lighter models are increasingly capable of handling narrower, repetitive tasks. Businesses must also account for privacy, data storage, and infrastructure requirements.

AI-generated content, search, and open source

• The speakers are concerned about AI-generated “slop”: networks of videos produced with generated scripts, visuals, and narration to maximize views with minimal effort. They discuss the risk that such channels will copy the topics, scripts, and packaging of human creators, and that platforms may become saturated with mass-produced material.
• They also describe AI summaries changing how people find information. Instead of searching, visiting sites, and following links, users may increasingly get answers directly from AI. One speaker remarks that programming’s traditional search-and-learning habits—including relying on Stack Overflow—are already changing.
• The claim that “open source is dead” is raised provocatively, not established as fact. The concerns are that AI-generated contributions could overwhelm project maintainers and that platforms or model developers may benefit from code hosted by developers, while reducing the value of conventional open-source contribution and review.

Broader reflections

The conversation also touches on the wider effects of automation: the emotional difficulty of having a valued skill become less central, uncertainty about what meaningful work will look like, and the difference between creating something oneself and simply prompting a system to produce it. The later discussion moves into online culture, AI companions, hobbies, and questions of ambition and self-realization.

Reviews, guides, and tutorials

• There is no step-by-step coding tutorial or formal product review.
• GigaCode and GitVerse are discussed as examples of an AI coding assistant and collaborative repository, with the speakers encouraging interested viewers to consult the links in the description.
• The Avito guest shares practical observations about customer-facing agents, but does not present a setup guide.

Main speakers and sources

• Vlad Ten — programmer and creator, discussing his changing relationship with software development.
• Phil and Artyom (“Tyoma”) — podcast hosts participating in the discussion.
• Victor from Avito — guest describing customer- and seller-facing AI agent use cases.
• Products discussed: GigaCode and GitVerse.
• Other sources mentioned: Stack Overflow, GitHub, YouTube, and VK.