The Rise of AI-Generated Performance Reviews and the Employee Feedback Gap
A new survey shows 78% of managers use AI for performance reviews, but many employees find the feedback generic. Meanwhile, workers are using AI to practice tough conversations.
The integration of artificial intelligence into the workplace has reached a critical touchpoint: the manager-employee feedback loop. According to a recent survey of 1,034 corporate workers conducted by professional development company Highwire, managers are increasingly relying on AI to write and refine performance reviews, though the practice remains largely hidden from the employees receiving the feedback.
The study reveals that 78% of managers used AI to help draft, edit, summarize, or inform performance evaluations over the last year. While the technology successfully accelerates the preparation process for managers, transparency is lagging behind. Only 16% of non-managerial staff were notified that AI influenced their evaluations. This hesitation to disclose AI usage may be linked to employee concerns about professional perception; a June study by Atlassian indicated that admitting to using AI in the workplace can backfire on employees.
The Paradox of AI-Generated Performance Reviews
The impact of AI-assisted feedback on employees is highly polarized. On one hand, 54% of those surveyed reported that their feedback has become more specific and actionable since AI entered the workflow. On the other hand, the shift has left a substantial portion of the workforce feeling alienated. Specifically, 34% of employees found the feedback to be more generic, and 32% stated it was less useful.
The quality of the output depends heavily on how managers implement the technology. There is a vast difference between using AI to format or summarize a manager's original observations and delegating the entire evaluation process to an LLM based on raw performance metrics. When over-relied upon, the resulting "workslop" can diminish the value of the review. Highwire, whose business model centers on human-led professional training, emphasizes that certain interpersonal capabilities cannot be automated, stating that their service "conditions the human skills that can’t be automated."
AI as a Training Simulator
Beyond performance reviews, AI is stepping in to fill a significant support vacuum in employee development. The survey found that nearly one in four employees now rehearse difficult workplace conversations with an AI tool. This trend highlights a disparity between employee needs and manager availability. While 85% of workers agree that practicing high-stakes conversations is crucial, only 44% report receiving the necessary support from their managers to do so.
To address this training gap, companies are launching specialized products. AI video platform Synthesia recently introduced a feature called "Sessions," which allows employees to roleplay challenging scenarios with AI avatars and receive feedback from an automated "AI coach." Developed in response to enterprise client demand, the feature aims to scale managerial touchpoints without overextending human managers, though Synthesia clarifies that the tool is not intended to replace human mentorship.
What it means for developers
For software engineers and system architects, this shift highlights a growing market for specialized, context-aware HR tools. Standard, out-of-the-box language models often produce generic, unhelpful feedback that employees dismiss as automated filler. Developers have a unique opportunity to build highly customized, fine-tuned models that integrate specific organizational guidelines, historical performance data, and communication styles to generate truly personalized feedback.
Furthermore, the rising demand for interactive training simulators, like Synthesia's conversational avatars, indicates that developers should focus on low-latency, multi-modal systems that combine voice, text, and video. Building and testing these complex pipelines requires access to a variety of state-of-the-art LLMs. Developers looking to experiment with these architectures can try top AI models cheaply through one API at https://apixoai.online. By leveraging unified API access, engineering teams can rapidly prototype, compare how different models handle sensitive performance data, and deploy tools that offer genuine value to both managers and employees.
Source: Managers are using AI to write performance reviews – and it shows — ZDNet AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
