AI algorithmic progress multiplier reaches 4× (~2× overall R&D)

What AI 2027 Predicted

The scenario describes the culmination of AI-driven R&D acceleration: by March 2027, coding has been “fully automated” and 200,000 Agent-3 instances run in parallel, equivalent to 50,000 elite human coders working at 30× speed. Despite this massive labor force, overall algorithmic progress is “only” 4× faster due to bottlenecks: research taste remains difficult to train, feedback loops in research are longer than in coding, and there are diminishing returns to throwing more coding labor at fundamental research problems.

This 4× multiplier represents the ceiling of the “coding automation” phase — further acceleration would require breakthroughs in automating research direction-setting and judgment, not just implementation.

How We Track This

We monitor:

• All indicators tracked for the 3× multiplier prediction (n29)
• Evidence of full coding automation at frontier labs
• Scale of parallel agent deployment for internal AI R&D
• Reports of diminishing returns to additional AI coding labor
• Progress on automating research taste / experiment design (the stated bottleneck)

Current Evidence

Current state of AI R&D acceleration:

• Estimated current multiplier: ~1.1-1.3× for full R&D pipeline (see n29-rd-multiplier-3x for detailed evidence)
• METR’s February 2026 update acknowledges developers are “more sped up” in early 2026 vs. early 2025, but doesn’t quantify the improvement (METR, February 2026)
• Individual developer productivity gains of 20-40% reported in favorable settings, but full R&D pipeline speedup is much lower

Coding automation trajectory:

• No frontier lab has reported full coding automation internally
• Claude Code and similar tools are powerful assistants but still require human oversight for complex systems
• The AI Futures authors’ updated estimates for full coding automation: Daniel Kokotajlo median 2029, Eli Lifland early 2030s
• SWE-bench-Verified scores remain below the scenario’s projected trajectory (74.5% actual vs. 85% predicted by mid-2025)

Parallel agent deployment:

• No public reports of anything approaching 200,000 parallel agent instances for internal R&D
• Frontier labs are running parallel agent evaluations at scale (thousands of concurrent runs), but this is for testing, not production research

The “only 4×” framing:

• The scenario acknowledges diminishing returns explicitly — 200,000 superhuman coders produce “only” 4× speedup. This reflects a real insight: most AI research bottlenecks are in problem formulation, not implementation
• Current evidence supports this insight: coding gains don’t translate proportionally to research breakthroughs

Sources:

• METR: Experiment redesign update (Feb 2026)
• METR: Early-2025 AI developer productivity study
• Grading AI 2027’s 2025 Predictions — AI Futures Project

Counterevidence & Limitations

• All counterevidence from the 3× prediction applies with greater force here
• Going from ~1.1-1.3× today to 4× in 12 months would require multiple discontinuous capability jumps
• The scenario’s own authors now estimate the timeline is 35-40% slower than depicted — pushing this milestone to 2029-2030+
• “Full coding automation” as described in the scenario remains far from current capabilities
• The 4× figure assumes breakthroughs that cascade: superhuman coding → scaled parallel deployment → automated experiment design → automated research taste. Each step faces separate bottlenecks
• Even optimistic industry insiders don’t publicly claim 4× R&D multiplier is imminent

What Would Change Our Assessment

• Upgrade to “emerging”: A frontier lab credibly reports 2×+ R&D multiplier; evidence of large-scale parallel agent deployment for internal research
• Upgrade to “on-track”: Reports of 3×+ multiplier with coding approaching full automation
• Maintain “not-yet-testable”: Prediction date is March 2027, ~12 months away
• Preemptive downgrade to “behind”: If the 3× multiplier (n29) is clearly missed by January 2027

Update History

Date	Update
2026-03	Prediction timeframe not yet reached. Current multiplier at ~1.1-1.3×. The scenario authors’ own updated estimates push 4× multiplier to 2029+, acknowledging the original March 2027 target was too aggressive.