Anthropic just released what they’re calling the largest qualitative AI user study ever conducted. It includes almost 81,000 interviews across 159 countries and 70 languages, asking people what they actually want and fear from AI.
It won’t come as a surprise that economic concerns are the top predictor of how people feel about AI. Among respondents, 22.3% cited worries about jobs and the economy, which includes job displacement, unemployment, and wage stagnation, among other concerns.
Job displacement and unemployment fears make sense. Every day, the headlines sound new alarms about companies downsizing to drive greater efficiency with AI. Wage stagnation, however, is a bit different.
Just last week, I spoke with several Fortune 100 leaders who shared that they are “flipping the pyramid.” They plan to hire more senior-level employees with a “builder-mindset” who can master multiple roles, spin up supporting AI agents, and supercharge their productivity. These people draw higher compensation packages, but companies require fewer of them.
This dynamic accelerates wage stagnation in most enterprises. Stagnation comes from thousands of unconnected decisions compounding over time. New-hire offers that outpace internal pay. Merit budgets that don’t keep up. Retention counters that create new gaps instead of solving old ones. The people who stayed loyal earn less than the people who just walked in.
For example, “Total Rewards” refers to the full cost of employee compensation, including salary and benefits, and is a major organizational cost. Yet most pay decisions are still made in isolation, so small differences add up to costly mistakes over time
Compensation executives know these patterns. As companies replace larger teams with fewer, higher-paid roles, the compression only accelerates.
The problem
In the current, non-AI landscape, humans are already making biased, inconsistent pay decisions every day. In my work with compensation leaders, I’ve seen how hundreds of millions of pay decisions are made across merit cycles, promotions, lateral moves and retention adjustments.
These decisions happen under pressure, in isolation, and are rarely revisited unless something goes wrong. No single person is truly governing the decisions. Instead, they’re shaped by fragmented systems, local manager discretion and the urgency of the moment.
Human oversight does exist, but it occurs at discrete checkpoints rather than continuously. Managers make individual pay decisions during merit cycles, promotions and hiring, often with limited visibility into how those choices compare across teams.
HR and compensation leaders step in later, during annual reviews or pay equity audits, to assess outcomes in aggregate. By then, the decisions themselves have already been made, making oversight more reactive than proactive.
As a result, many organizations invest heavily in pay equity audits and remediation, adjusting salaries, issuing back pay and addressing gaps after the fact. In my experience, that spending is largely corrective, addressing decisions made without consistent oversight at the time.
If prematurely labeled high risk, organizations may avoid the use of AI for pay when, in fact, it’s one of the most powerful and cost-effective use cases enterprises can tap into today.
The solution
AI can be of tremendous value to organizations if deployed with transparency and a human as the final decision-maker. AI can analyze large swaths of data, including:
- Market and competitive benchmarks
- Internal employee data
- Performance scores for promotion decisions
- Recent offers
- Pay gap risks and implications
AI can provide in-the-moment recommendations that are competitive, compliant and explainable. Each recommendation is presented to a human, who can either accept, deny or escalate. All of the logic is captured to further improve a company’s practices to drive better business alignment with pay.
All of this changes the overall cost of pay. For example, governance—how pay decisions are guided, reviewed and kept consistent—can reduce waste because of wide discretion. It avoids costly remediation, often applied months later when looking at the decisions in aggregate.
These systems can generate thousands of transparent pay recommendations in real time, for managers to review and approve. As a result, companies gain insight and the opportunity to look at patterns across many decisions, making traditional, static, backward looking adjustments a thing of the past.
In my work with compensation leaders, I’m seeing this play out across many organizations today.
The path forward
Avoiding AI in pay decisions forgoes unlocked cost savings, productivity gains, and efficiencies. The solution is to govern AI in pay decisions.
AI that surfaces a recommendation grounded in market data, internal equity (how pay compares across similar roles at the company), and compliance guardrails (rules that keep decisions in line with laws and policies), with a human reviewing and making the final call, is how companies can stop the pattern rather than accelerate it.
That’s the only version of AI for pay that actually closes gaps rather than compounds them.
Maria Colacurcio is the CEO of Syndio.