---
title: Survey reveals humans remain central to workplace decision-making despite AI advancements
url: https://www.dataloco.com/en/survey-reveals-humans-remain-central-to-workplace-decision-making-despite-ai-advancements
published: 2026-09-23T21:20:45+00:00
language: en
section: AI
source: https://japan.zdnet.com/article/35252783/
organizations: Resume Now, Pearl, Anthropic, Perplexity, Massachusetts Institute of Technology
publisher: Dataloco
---

# Survey reveals humans remain central to workplace decision-making despite AI advancements

A recent survey conducted by Resume Now indicates that artificial intelligence is not currently competing with human workers in professional environments. The study found that 97 percent of surveyed employees rely on their own judgment or that of other humans when making decisions. Furthermore, 72 percent of respondents stated that they trust the judgment of a colleague over an AI model when the two disagree. These findings suggest that human oversight remains the dominant factor in workplace operations.

While AI capabilities have improved significantly, experts note that widespread adoption of the technology for core business functions is still in its early stages. A panel discussion on higher education in New York highlighted that for most professionals, AI serves an auxiliary role. Strategic planning and high-level tasks continue to be handled by human employees. One investment banker with data science knowledge mentioned that if permitted, they could replace most of their team members with the Claude model, though this view is not representative of the broader workforce.

Workers are generally willing to use AI to speed up routine information gathering tasks. However, they become cautious when the task requires strategic thinking or judgment. According to the Resume Now report, 93 percent of respondents rely on themselves, colleagues, or managers when they must make important and rapid decisions. This indicates that AI is not yet viewed as a reliable substitute for critical human decision-making in high-stakes scenarios.

The survey also detailed specific use cases for AI in the workplace. Nearly a quarter of workers use the technology for research purposes. Fourteen percent use it for creating emails and other messages, while 13 percent use it to summarize information. Only 9 percent utilize AI for data analysis, and a mere 3 percent seek advice on strategy or planning from these systems. These figures underscore that AI is primarily used for administrative and informational support rather than strategic leadership.

Productivity gains have been a major selling point for AI, but some metrics suggest it may not be improving team efficiency as expected. Instead, the technology can sometimes create work slop, which refers to wasted time. The Resume Now findings confirm that employees are not forced to rely on AI for easy solutions but choose to use it selectively based on the nature of the task.

Pearl, a professional services firm, operates a leaderboard that ranks the performance of frontier AI models against human experts. The company uses non-public datasets created by experts that the models have never seen during training. Pearl notes that while public benchmarks report AI performance at 80 to 95 percent of human levels, these datasets are often part of the AI training process. There is a significant gap between benchmark performance and real-world professional capability, a gap that has not narrowed with newer models.

As of September, the top-ranked model on the leaderboard is Claude Fable 5 from Anthropic, which scored 80 percent in fields such as business, technology, law, and healthcare. Pearl equates this score to a B-minus grade for a student. Other models, including Opus 5, Fable 5.1, and Perplexity's Sonar Deep Research, follow with scores slightly below 80 percent. The Massachusetts Institute of Technology predicted in April 2026 that AI would be able to perform some economically important tasks at a minimally acceptable level by 2029. The current 80 percent score aligns broadly with this prediction.