The missing step between hype and profit

Despite massive investment and bold predictions, AI systems outside of coding still struggle to deliver economic value in real workplaces. A study by Mercor found that AI agents powered by top models from OpenAI, Anthropic, and Google DeepMind failed to complete most of 480 real-world banking, consulting, and legal tasks. The gap between AI hype (Step 1) and promised transformation (Step 3) remains unfilled — the missing Step 2. The author argues that closing this gap requires transparency from model makers, better real-world evaluation methods, and honest coordination between researchers and businesses rather than relying on wild weekly claims.