Context on state of AI coding:

• AI writes majority (~90%? see Dario pod with Dwarkesh for context / the timeline) of code

• If we say that coding encompasses all knowledge tasks, it’s generally a fair assessment to say AI is roughly at the 90th percentile of most fields

• The main reason AI writes the majority of the code now is that it’s much faster

• Recent model launches on 2/5/26 (Opus 4.6, Codex 5.3) seem to represent a step change where they’re offering Free usage for a couple months (Codex is Free until 4/2/2026, Anthropic is giving $50 in extra usage credits for Claude largely due to Opus 4.6), additionally they’ve launched Opus 4.6 Fast (2.5x the speed, 6x the cost), and Codex 5.3 Spark

• OpenClaw also represents a large step change in AI coding agents with a new level of autonomy and one person project possibly being worth a billion leading to OpenAI hiring Steinberger

So for this market, it’s asking when autonomous vehicle driving reaches a similar level of advancement where

• majority of drivers don’t manually drive

• There’s probably a term similar to “vibe driving” that takes advantage of the increased speed

• One sign of this may be increased speed limits or AI only driving zones?

  Critical benchmarks include:

• % of miles driven with minimal / no human oversight

• New vehicles that are meant for minimal user input (similar to AI first IDEs)

• Whether large players (e.g. Apple, NVIDIA) get into the space since currently it seems like Tesla owns majority of market share, not too sure about Waymo and GM, does Google still work on this?

  additionally, similar to how people think AI progress can reach a singularity / takeover, the same discussions should likely be occurring in the self driving space:

  e.g.

  https://youtu.be/040ejWnFkj0?si=yABcyNeCGye_-Wo2

  "How Self-Driving Cars will Destroy Cities (and what to do about it)" by the channel Not Just Bikes.

  It matches your description perfectly:

  • "How fast they are": It discusses how AV companies will likely lobby to remove speed limits to increase "throughput," making streets dangerous and loud.

  • "Consume everything": The video explicitly argues that autonomous vehicles (AVs) will "consume every street in the city," pushing out pedestrians, cyclists, and public transit.

  • "Roads will be bad / Noise": It details how faster, heavier electric AVs will actually generate more tire noise than current cars (above 30mph/50kmh) and make cities hostile, fenced-off environments similar to highways.

  ---

  Resolution criteria

  This market resolves YES when autonomous vehicles reach a level of adoption and capability comparable to AI coding as of February 2026. Specifically, the market resolves YES when:

  1. Majority of miles driven autonomously: Autonomous miles driven reach a 100% CAGR through 2032, with the majority of total vehicle miles in developed markets driven with minimal or no human oversight (defined as Level 4+ autonomy or equivalent).

  2. Mainstream autonomous-first vehicles: New vehicle models are designed primarily for autonomous operation with minimal user input, analogous to AI-first IDEs in software development.

  3. Market consolidation and new entrants: Tesla has over 400,000 vehicles equipped with FSD in North America, while Ford and General Motors are investing billions to launch self-driving taxi services by 2026. Resolution requires evidence that major tech companies (Apple, NVIDIA, or others) have meaningfully entered the autonomous driving market alongside Tesla and Waymo.

  Resolution sources: Goldman Sachs projects about 35,000 robotaxis across the US in 2030, capturing roughly 8% of the rideshare market. Track progress via Tesla FSD deployment data, Waymo service expansion, and NHTSA autonomous vehicle testing reports.

  The market resolves NO if autonomous driving adoption stalls significantly below these benchmarks or if regulatory/safety barriers prevent widespread deployment by a reasonable future date.

  Background

  Anthropic reports that 70% to 90% of the company's code is now AI-generated, and AI models are materially helping to build more advanced iterations of themselves. 4% of GitHub public commits are currently authored by Claude Code, with projections reaching 20% or more by the end of 2026. By the end of 2025, roughly 85% of developers regularly use AI tools for coding.

  In contrast, autonomous vehicle adoption remains nascent. Only 1-2% of total global vehicle sales in 2026 are expected to have Level 3 features. L4 robo-taxis are now available in the first cities in the United States and China, with global rollout expected at large scale in 2030 rather than 2029, and L4 urban pilots for private passenger cars expected in 2032.

  Considerations

  The comparison assumes "as good as" means functional parity in adoption and capability, not identical use cases. AI coding operates in a controlled digital environment with clear success metrics (code compiles, tests pass), while autonomous driving must handle unpredictable real-world conditions, liability concerns, and regulatory fragmentation. As of January 2026, regulators are increasingly enforcing "proven safety" metrics rather than self-certification, which may slow deployment compared to software adoption. Additionally, mass adoption timelines differ significantly between robotaxis, autonomous trucking, and consumer-owned passenger vehicles, making a single resolution threshold challenging.

  This description was generated by AI.

• Update 2026-02-16 (PST) (AI summary of creator comment): Rounding rules for resolution year:

  • Late 2031 (second half) resolves to 2032

  • Early 2031 (first half) resolves to 2030

  • The 2050 bucket includes 2050 or later

  • Similar rounding applies to other years