[Paper Review] AI-Assisted Goal Setting Improves Goal Progress Through Social Accountability

Michel Schimpf, Julian Voigt+1 more|arXiv (Cornell University)|Mar 18, 2026

Psychological Treatments and Assessments0 citations

TL;DR

An preregistered three-arm RCT shows an AI career coach increases short-term goal progress versus no-support, mainly by enhancing perceived accountability; over a matched written reflection, AI showed higher accountability but not significantly higher progress.

ABSTRACT

Helping people identify and pursue personally meaningful career goals at scale remains a key challenge in applied psychology. Career coaching can improve goal quality and attainment, but its cost and limited availability restrict access. Large language model (LLM)-based chatbots offer a scalable alternative, yet the psychological mechanisms by which they might support goal pursuit remain untested. Here we report a preregistered three-arm randomised controlled trial (N = 517) comparing an AI career coach ("Leon," powered by Claude Sonnet), a matched structured written questionnaire covering closely matched reflective topics, and a no-support control on goal progress at a two-week follow-up. The AI chatbot produced significantly higher goal progress than the control (d = 0.33, p = .016). Compared with the written-reflection condition, the AI did not significantly improve overall goal progress, but it increased perceived social accountability. In the preregistered mediation model, perceived accountability mediated the AI-over-questionnaire effect on goal progress (indirect effect = 0.15, 95% CI [0.04, 0.31]), whereas self-concordance did not. These findings suggest that AI-assisted goal setting can improve short-term goal progress, and that its clearest added value over structured self-reflection lies in increasing felt accountability.

Motivation & Objective

• Identify whether an AI-assisted goal-setting coach improves short-term goal progress in working adults.
• Isolate the psychological mechanisms by which AI coaching affects progress, focusing on accountability and self-concordance.
• Compare AI coaching to a structurally matched written reflection and a no-support control to parse conversational effects.
• Examine whether AI coaching influences goal specificity and perceived satisfaction with the tool.

Proposed method

• Three-arm preregistered randomized controlled trial (N=517) with conditions: AI career coach (Leon), matched written questionnaire, and no-support control.
• Goal progress assessed at two-week follow-up using a 9-item, 7-point scale (alpha=.86).
• Mediators (accountability and self-concordance) measured at T1; mediation tested via parallel mediation with bootstrap (5,000 resamples).
• Manipulation checks showed AI was more interactive and perceived reflection higher than control.
• Exploratory analyses included LLM-based goal specificity coding and goal-domain classification.

Experimental results

Research questions

• RQ1 Does AI-assisted goal setting improve short-term goal progress compared with a no-support control and with a structured written reflection?
• RQ2 Is perceived accountability a mediator of the AI vs. control and AI vs. questionnaire effects on goal progress?
• RQ3 Does self-concordance mediate AI-driven progress, and how does goal specificity relate to outcomes?
• RQ4 Do AI coaching and structured reflection differ in perceived interactivity and perceived structured reflection?
• RQ5 Do AI goals include more non-career content and are goals more specific under AI coaching?

Key findings

• AI coaching produced higher goal progress at two weeks than the control (mean 3.45 vs 3.02; d=0.33; p=.016).
• AI coaching increased perceived accountability compared with both the questionnaire and control (d=0.43 and d=0.68 respectively; p=.002 and p<.001).
• Compared with the questionnaire, accountability mediated the AI-vs-questionnaire effect on goal progress (indirect effect = 0.15; 95% CI [0.04,0.31]).
• Self-concordance did not differ across conditions and did not mediate AI effects.
• Exploratory analyses suggested AI goals were more specific than those in other conditions, and AI surfaced more non-career goals (non-career content was 40.3% in AI vs 13.2% in questionnaire and 10.6% in control).
• NPS (user satisfaction) was substantially higher for AI at T1 than both comparison conditions (d=0.86 vs control; d=0.52 vs questionnaire; p<.001).