Process Praise

In the AI Story

The premise of process praise is that the process being praised belongs to the learner. When a teacher says "you worked really hard on that essay," the implied causal chain connects the student's effort to the essay's quality. The praise validates the effort by linking it to the outcome, and the student learns that effort produces results — a learning that reinforces growth-mindset orientation and sustains effort through future difficulty.

AI-assisted production breaks this causal chain. If the teacher praises the output, she is praising the system's capability, not the student's growth. If she praises "the work," the praise is diffuse, directed at an undifferentiated blend of human and machine contribution. The third option — praising the direction — requires recognizing the cognitive work of formulating a question, defining what is needed, evaluating the machine's response, and adjusting the request. This is genuine intellectual labor that deserves process praise. But it is also invisible to observers calibrated to identify effort through traditional markers like visible hours spent writing.

The pedagogical innovation described in The Orange Pill — the teacher who stopped grading essays and started grading questions — operationalizes process praise for the AI age. The assignment is not to produce an essay but to produce the five questions the student would need to ask before she could write an essay worth reading. The questions become the object of evaluation and praise, making visible the cognitive labor that AI collaboration demands.

A 2025 study in Learning and Instruction found that AI-driven feedback could serve as a "powerful catalyst for nurturing growth mindsets" — but specifically when the feedback was designed to highlight process over product. Feedback focused on what the student did produced growth-mindset outcomes; feedback focused on output polish produced the fragility characteristic of talent-based praise.

Origin

The process praise finding emerged from a series of studies Dweck conducted in the 1990s with Claudia Mueller, published most influentially as "Praise for Intelligence Can Undermine Children's Motivation and Performance" (Journal of Personality and Social Psychology, 1998). The experimental design was elegant: children were given puzzles, succeeded at them, and then received one of two types of praise — either for their intelligence or for their effort. On subsequent harder puzzles, the intelligence-praised children avoided challenge, lied about their scores, and showed performance declines. The effort-praised children sought harder puzzles, acknowledged mistakes, and maintained performance.

The finding has been replicated, refined, and extended across subsequent decades, becoming one of the most operationalized results in educational psychology.

Key Ideas

Process produces resilience. Praise directed at effort, strategy, and engagement produces psychological resilience; praise directed at innate ability produces fragility.

The causal chain matters. Effective process praise connects the learner's effort to the valued outcome, teaching that effort is the mechanism of development.

AI breaks the traditional chain. When output is AI-assisted, traditional process praise misidentifies the locus of the process being praised.

Direction is the new process. In AI-mediated work, the cognitive labor of formulating questions, evaluating responses, and iteratively refining direction is the process that deserves praise.

Invisible work requires structural support. Because direction-work is invisible, praising it requires new assessment structures that surface the cognitive process rather than only the final product.