The argument that AI adoption is slow due to normal tech diffusion is flawed. If AI models possessed true human-equivalent capabilities, they would be adopted faster than human employees because they could onboard instantly and eliminate hiring risks. The current lack of widespread economic value is direct evidence that today's AI models are not yet capable enough for broad deployment.

A benchmark testing AI agents against paid freelance jobs found the best performers could only autonomously complete 2.5% of the work. This provides a crucial reality check, showing that while AI excels at discrete tasks, full job automation by general-purpose agents is still far from reality.

AI models excel in domains with discrete, quantifiable outcomes like coding or chess. However, they struggle with most knowledge work, which is often "unverifiable" and lacks a single correct answer for reinforcement learning models to train on. This distinction explains AI's current limitations in many professional roles.

Despite marketing hype, current AI agents are not fully autonomous and cannot replace an entire human job. They excel at executing a sequence of defined tasks to achieve a specific goal, like research, but lack the complex reasoning for broader job functions. True job replacement is likely still years away.

Despite hype about full automation, AI's real-world application still has an approximate 80% success rate. The remaining 20% requires human intervention, positioning AI as a tool for human augmentation rather than complete job replacement for most business workflows today.

Despite marketing claims, current AI agents cannot truly learn or improve over time like a human employee. They operate by consulting static knowledge bases, not by gaining experience. This "narrative gap" between public perception and actual capability is a major industry challenge.

AI's value is overestimated because experts view complex jobs as simple, solvable tasks. The real bottleneck is the unproductive effort required to build a custom training pipeline for every company-specific micro-task. Human workers are valuable precisely because they avoid this “schleppy training loop” by learning on the job, a capability current AI lacks.

Alex Karp argues that an AI's high score on a single benchmark is irrelevant for enterprise adoption. Real institutions require passing thousands of consecutive, differentiated tests. An AI model that is brilliant at one task but fails at the 50th in a complex sequence is effectively useless.

Handshake's CEO suggests that while AI safety is important, the constant focus on existential risks conveniently distracts from a more immediate problem: frontier models struggle to deliver tangible ROI in enterprise settings outside of software engineering, leading to high token costs without proportional productivity gains.