iCIMS survey finds only half of purchased AI capabilities reach production

Hiring teams’ confidence in AI is running ahead of deployment. In research released October 28, 2026, iCIMS and Kyle & Co. found that 45% of surveyed HR and talent leaders rated their AI implementation highly successful. Yet the average organization put only about half of its purchased AI capabilities into production.

The 2026 Talent Transformation Yearbook: AI Progress Report surveyed 500 HR and talent leaders and 1,000 U.S. job seekers. Published by hiring-platform provider iCIMS with Kyle & Co., it compares employers’ reported hiring performance with their AI preparation and use, alongside candidates’ experiences of the technology.

More deployment, stronger reported results

The researchers grouped leaders by self-reported performance across six measures: hiring speed, cost, quality, new-hire retention, hiring-manager satisfaction and contribution to business outcomes. The highest group, representing 37% of respondents, rated itself above average on all six. The lower group, representing 34%, rated itself above average on three or fewer. These were respondents’ assessments, not independently measured hiring results.

The seven capabilities covered screening and ranking résumés, candidate chatbots, interview scheduling, AI-led assessments, AI-assisted writing, skills matching and agentic AI—systems handling multistep hiring workflows. Full-scale deployment of that last category was reported by 42% of the higher-performing group, compared with 9% of the lower-performing group.

Preparation differed, too. Completion rates for five foundational practices ranged from 51% to 62% in the higher-performing group, versus 17% to 24% in the lower group. Those practices covered AI literacy training, data ownership, AI contract terms, manager enablement and redeployment. The findings show an association between readiness, deployment and reported outcomes; they do not establish that deploying more AI caused better hiring.

Candidates see benefits, but not always the process

For job seekers, AI screening was far from universal. Forty-four percent said they had never encountered it during their 2026 search, and 17% encountered it only once. Another 32% encountered it several times or in most applications. Those answers describe what candidates noticed, rather than a direct count of employers’ screening systems.

Candidates who noticed AI were more likely to describe its effect positively than negatively: 45% versus 21%. Fairness drew a less decisive response. Among candidates who went through automated screening, 42% believed they were evaluated fairly, while 33% were concerned they were not.

Disclosure was another weak spot. Among candidates who believed AI evaluated their applications, only 32% said employers clearly disclosed its use. Another 41% suspected AI had been used without being told. Positive experiences with the technology therefore sat alongside uncertainty about where it entered the decision.

Trent Cotton, iCIMS’ head of talent insights, urged employers to explain when AI evaluates candidates’ responses, its role in the decision and where human review occurs. In HRD’s coverage of the report, he said that need not require a lengthy explanation. Whether greater disclosure improves candidate confidence remains a question to examine, he added.

Scaling AI is not the only 2027 priority

Leaders’ plans put AI growth behind two other hiring priorities. The leading choices for 2027 were:

• High-volume and frontline hiring capacity: 28%.
• Skills-based hiring or internal mobility: 27%.
• Scaling AI: 24%.

Yet 46% described their approach to high-volume hiring as mostly reactive. Building candidate pools ahead of demand was more common in the higher-performing group: 42% reported proactive talent pipelines, compared with 25% of the lower-performing group.

The leading concerns were closely matched: candidate experience at 30%, maintaining human judgment at 29%, and responsible, accurate and compliant AI use at 28%. Employers’ planning challenge includes not just putting purchased tools to work, but preserving the human judgment and candidate experience they also rank highly.