---
title: "AI isn't making you faster"
description: "5 source-backed AI claims from AI isn't making you faster, with key statistics, context, and the original AI Adopters Club source."
url: "https://kbanc.com/claims-library/ai-isnt-making-you-faster"
source: "https://aiadopters.club/p/ai-isnt-making-you-faster"
date: "2026-03-21"
topics: ["measurement", "implementation", "strategy"]
generated: "2026-08-31"
---

# AI isn't making you faster

By Kamil Banc | March 21, 2026

## Claims

1. **METR Trial: Slower, Not Faster** (source summary): In a METR controlled trial, experienced developers predicted a 24% speedup but were actually 19% slower.
2. **Perceived vs. Measured Savings** (source summary): Ninety percent of AI users say it saves time, but measured savings are 2.8% of work hours.
3. **ROI Remains Elusive** (source summary): Ninety-five percent of organizations see no measurable ROI from AI, and only 21% replaced a single task.
4. **Training Drives Productivity Gains** (source summary): Workers with over 81 hours of deliberate AI training reported 75% higher productivity than peers.
5. **Full Delegation Separates the 5%** (Kamil's interpretation): The 5% who see real gains pick one task and hand it to AI completely.

## Evidence

### Quote
> "Fast.ai calls this dark flow, a state where AI tools generate the feeling of productivity without the output." - Kamil Banc

### Key Statistics
- **24% predicted speedup vs. 19% measured slowdown**: METR's controlled trial gave 16 experienced developers the best AI tools for 246 real tasks; their predictions diverged sharply from measured outcomes.
- **2.8% measured savings**: While 90% of AI users claim the technology saves them time, measured savings amount to only 2.8% of work hours.
- **95% no measurable ROI**: Despite widespread AI adoption, 95% of organizations report no measurable return on investment, and only 21% have replaced a single task with AI.
- **75% higher productivity**: EY found workers who invested 81+ hours in deliberate AI training reported 75% higher productivity than colleagues using the same tools without training.

## Context
The article synthesizes findings from multiple controlled studies and surveys, including METR's randomized trial of experienced developers and large-scale workforce surveys from Tech Policy Press, HBR, and Verasight. The consistent pattern across sources is a wide gap between perceived productivity gains and measured outcomes, which Fast.ai terms 'dark flow.' For practitioners, Kamil Banc's recommendation is a low-risk experiment: select one repetitive task, delegate it entirely to AI for five days, and measure both time saved and whether the output survives without human intervention. This approach treats AI as a skill requiring deliberate investment, consistent with EY's finding that extensive training correlates with substantially higher productivity.

## Source
- Original: [AI isn't making you faster](https://aiadopters.club/p/ai-isnt-making-you-faster)
- Cite: kbanc.com/claims-library/ai-isnt-making-you-faster

## Primary Evidence
- [controlled trial by METR](https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/) (metr.org; supports claim 1)
- [90% of AI users](https://www.techpolicy.press/generative-ais-productivity-myth/) (techpolicy.press; supports claim 2)
- [95% of organisations](https://hbr.org/2025/09/ai-generated-workslop-is-destroying-productivity) (hbr.org; supports claim 3)
- [Only 21%](https://www.verasight.io/reports/ai-adoption-in-2026) (verasight.io; supports claim 3)
- [do see real gains](https://www.ey.com/en_gl/newsroom/2025/11/ey-survey-reveals-companies-are-missing-out-on-up-to-40-percent-of-ai-productivity-gains-due-to-gaps-in-talent-strategy) (ey.com; supports claim 4)