I’m a CTO. AI still needs your human expertise | Madhu Shashanka

AI is advancing rapidly and can perform complex tasks with impressive accuracy, but it still lacks the human touch required for judgment-driven work. Autonomous vehicles, for example, have proven to be safer than human drivers, yet they still struggle with basic parking and navigation tasks. This phenomenon, known as "jagged" AI, demonstrates the system's strength in specific areas while struggling in others.

As a result, the nature of human work is shifting towards requiring more judgment and critical thinking. While AI is capable of providing confident answers to many tasks, it often fails to indicate when it is incorrect, leaving people to determine the accuracy of its output. The biggest challenge lies in the situations where there is no clear right answer, as these require human judgment to determine what constitutes a good outcome.

As AI continues to evolve, new roles are emerging where humans oversee and manage AI agents, ensuring the system operates within predetermined limits. However, the most crucial role in this new landscape is one that combines an understanding of the real-world task with the ability to recognize when AI has failed to provide the correct answer.

This demand for human expertise is not limited to experienced professionals but also extends to newcomers, who must learn to develop judgment through hands-on experience. Despite the fear of job losses in entry-level, routine positions, these roles are essential for building judgment over time. As AI takes over repetitive tasks, the need for human judgment becomes more valuable than ever.