The goal for AI isn't just to match human accuracy, but to exceed it. In tasks like insurance claims QA, a human reviewing a 300-page document against 100+ rules is prone to error. An AI can apply every rule consistently, every time, leading to higher quality and reliability.
Related Insights
AI's core strength is hyper-sophisticated pattern recognition. If your daily tasks—from filing insurance claims to diagnosing patients—can be broken down into a data set of repeatable patterns, AI can learn to perform them faster and more accurately than a human.
As AI agents become reliable for complex, multi-step tasks, the critical human role will shift from execution to verification. New jobs will emerge focused on overseeing agent processes, analyzing their chain-of-thought, and validating their outputs for accuracy and quality.
Don't wait for AI to be perfect. The correct strategy is to apply current AI models—which are roughly 60-80% accurate—to business processes where that level of performance is sufficient for a human to then review and bring to 100%. Chasing perfection in-house is a waste of resources given the pace of model improvement.
The benchmark for AI performance shouldn't be perfection, but the existing human alternative. In many contexts, like medical reporting or driving, imperfect AI can still be vastly superior to error-prone humans. The choice is often between a flawed AI and an even more flawed human system, or no system at all.
AI delivers the most value when applied to mature, well-understood processes, not chaotic ones. Pharma's MLR (Medical, Legal, Regulatory) review is a prime candidate for AI disruption precisely because its established, structured nature provides the necessary guardrails and historical data for AI to be effective.
To ensure product quality, Fixer pitted its AI against 10 of its own human executive assistants on the same tasks. They refused to launch features until the AI could consistently outperform the humans on accuracy, using their service business as a direct training and validation engine.
Unlike deterministic SaaS software that works consistently, AI is probabilistic and doesn't work perfectly out of the box. Achieving 'human-grade' performance (e.g., 99.9% reliability) requires continuous tuning and expert guidance, countering the hype that AI is an immediate, hands-off solution.
AI tools can be rapidly deployed in areas like regulatory submissions and medical affairs because they augment human work on documents using public data, avoiding the need for massive IT infrastructure projects like data lakes.
A key argument for getting large companies to trust AI agents with critical tasks is that human-led processes are already error-prone. Bret Taylor argues that AI agents, while not perfect, are often more reliable and consistent than the fallible human operations they replace.
The benchmark for AI reliability isn't 100% perfection. It's simply being better than the inconsistent, error-prone humans it augments. Since human error is the root cause of most critical failures (like cyber breaches), this is an achievable and highly valuable standard.
