AI Profits Drought: The J-Curve

The A.I.-Profits Drought: Why 95% of Projects Fail

Connecting economic history to the lack of human judgment in today's AI implementations.

The AI Productivity Paradox

This section highlights the surprising gap between AI adoption and measurable financial returns, reminiscent of the 1980s "computer productivity paradox." We're seeing massive investment and worker usage, but almost no impact on the bottom line for the majority of firms.

45.6%

of workers are using AI tools

95%

of organizations are getting **zero measurable return** (P&L impact) on GenAI investment after six months.

Executive Sentiment: Confidence Drop

The rapid drop in CEO confidence suggests a recognition that the initial hype is not translating into results. The Akkodis survey shows a sharp decline in C-suite certainty about their AI strategies.

The Historical Lesson: General Purpose Technology

Like the steam engine, electricity, and computers, **Generative AI is a General-Purpose Technology**. These transformative inventions *always* take longer than expected to show economy-wide productivity gains because they require the development of new infrastructure, complementary skills, and management restructure.

The lag is normal, but the cost of waiting (and failing) is not. Click the chart points to explore the stages of economic transformation.

Reasons for Failure: Why Intervention is Key

The 95% failure rate is not a technical problem; it's a **Decision Science problem**. AI projects fail when human oversight is absent and the tool is expected to act as an oracle. The key is intervention, customization, and purpose-driven deployment.

The Solution: Re-Engage Human Judgment

The article clearly demonstrates that the value isn't in the AI tool itself, but in the **human's approach** to using it. AI's core limitation--its inability to retain context, judge values, or own consequences--is exactly why the **Four Leadership Anchors** are critical.

Authority Check

Don't let the draft be the final.

Purpose Check

Optimize for values, not probability.

Accountability Check

Own the risk, design the loop.

Truth Check

Fight bias and fabrication.

AI projects without user intervention fail because they lack these human anchors. The 95% failure rate represents companies that delegated thinking to the tool. The 5% success rate represents companies that used the tool to *augment* their judgment in narrow, customized, and high-value ways.