50 万行代码不敢交给 AI？TypeScript 之父直言：它就像是个“高级复读机”

Core Insights - The existing programming languages are more suitable for AI programming not due to their advancement but because they possess the largest training datasets [1] - Current large models are essentially advanced "repeaters" that regenerate previously done tasks with simple deductions [2] TypeScript 7.0 Core Upgrade - TypeScript 7.0 introduces a native compiler, which is currently in the preview stage, promising a performance increase of 10 times [5] - The original TypeScript compiler was written in TypeScript itself and ran on the V8 JavaScript engine, but performance issues arose as project scale and usage scenarios expanded [5][6] Compiler Migration and Language Choice - The TypeScript team faced controversy when choosing the implementation language for the native compiler, ultimately selecting Go over Rust and C due to technical requirements [7][8] - Hejlsberg emphasized that the decision was based on practical needs rather than personal preference, asserting that the choice of tools was correct [8] AI in Code Migration - Initial attempts to use AI for migrating TypeScript to Go were unsuccessful, as AI-generated outputs often lacked the necessary certainty, leading to the need for manual checks [9] - Hejlsberg suggested that AI should be used to generate auxiliary migration tools rather than directly translating code, which could yield more reliable results [10] Future of TypeScript - TypeScript will continue to follow the standardization process of JavaScript, with no radical changes expected in the language itself, but significant transformations anticipated in the toolchain [11] - AI is expected to play a central role in development tools, potentially changing the traditional IDE landscape [12] Origin of TypeScript - TypeScript was conceived not to create a new language but to address the shortcomings of JavaScript, stemming from the needs of the Outlook Web team [13] - The migration of TypeScript's compiler to Go reflects a recognition of performance limitations, raising questions about the language's perception in the industry [13]