Episode One

The first episode establishes the 2 Giga Later format: Alexander Polomodov and Anton Kosterin discuss technology stories at a measured pace. They begin with Y Combinator's video on vibe coding through voice commands and regeneration. It lowers the barrier to prototyping, but a product still needs engineering and product judgment.

The hosts connect critical infrastructure's reliance on foreign products with rebadged open source. Citi crediting $81 trillion instead of $280 shows why financial systems need independent controls. Their C++ discussion separates memory-safety concerns from the language's trouble communicating its value to a new generation.

A long section follows Perplexity and Aravind Srinivas's path through Google and OpenAI. The hosts examine AI search's fortunate timing, the difference between a product and its base model, and the surrounding ecosystem. An electrification analogy separates AI that augments people, assists an action, or completes a task autonomously.

The final study links a candidate's appearance with interview outcomes. Models learn from human decisions and inherit hidden biases in their data. In medicine and recruitment this creates a risk of harmful recommendations, so data provenance and human verification remain part of system quality.