A recent survey commissioned by TripleTen indicates that preparing for AI-driven change has become widespread among US workers: 48 percent of full-time American employees have considered a backup career path in case their current trajectory does not deliver expected outcomes. Of those who considered an alternative, 77 percent took concrete steps to prepare for it — a practice the research labels "career cushioning."
Career cushioning does not necessarily mean immediate switching; rather, workers are developing parallel skills, networks and learning paths that reduce dependence on their current role or employer. Motivations are not limited to avoiding job loss: respondents cited better quality of life, lower stress and more enjoyable work as important drivers, while pay ranked lower on the list.
Sector differences: tech and finance move faster
The survey highlights strong industry differences in AI adaptation. In the technology sector, 59 percent of employees reported being asked to learn how to use AI chatbots; in finance the rate was 48 percent. By contrast, 25 percent in education and 18 percent in health and wellness reported similar expectations.
Detailed sector indicators from TripleTen include:
- AI has significantly or substantially changed the job: technology 69%, finance 47%, manufacturing 46%, retail 41%, education 36%, health and wellness 34%.
- Did not learn any new skills: technology 3%, finance 9%, manufacturing 18%, retail 24%, education 25%, health and wellness 31%.
These figures suggest the technology sector can be read as a leading indicator for skill shifts that may later reach more regulated or people-centered fields.
Workers are self-funding training
Self-funded training is widespread: on average workers spent $1,367 of their own money on professional skill development in the past 12 months. Industry averages vary significantly:
- Technology: $1,758 average (13% spent nothing)
- Finance: $1,664 average (24% spent nothing)
- Health care: $990 average (45% spent nothing)
- Education: $883 average (42% spent nothing)
The higher spending in tech and finance aligns with stronger immediate workplace expectations and faster role changes in those sectors.
Beyond time savings: AI expands role boundaries
Corporate AI adoption is often framed as productivity gains — less admin, faster writing, automated routine tasks. Ana Riabova, TripleTen's head of AI strategy and capability development, argues the early payoff is different: AI expands what a worker can accomplish. Employees who use AI to take on tasks they previously handed off gain an advantage, effectively extending their role while their core expertise continues to provide necessary oversight.
This dynamic increases the value of professional experience and domain knowledge, because integrating AI outputs reliably requires understanding context, risks and company-specific data.
The most important — and underappreciated — skill: spotting AI errors
Despite its importance, the ability to judge whether an AI output is correct is undervalued. Only 30 percent of leaders consider this an absolute baseline requirement, and just 22 percent of employees were asked to learn it. Yet detecting erroneous inferences, bad data interpretation or context-free suggestions requires domain expertise, especially in education and health care, where fewer than one-fifth of workers were asked to learn how to evaluate AI outputs.
As generative AI tools spread, the capacity to notice when a system confidently errs becomes increasingly valuable for risk management.
Innovation often comes from the bottom up
The survey also finds that AI adoption frequently begins with employees rather than top-down programs. Seventy-eight percent of workers have started learning AI tools on their own without explicit employer instruction. Among leaders and business owners the share is 96 percent, and among entry-level employees it is 51 percent.
Managers reported concrete impacts: 31 percent could name a skill that a subordinate demonstrated which they would not previously have considered; in 88 percent of those cases the solution later became standard workplace practice. About 40 percent of the skills cited were related to AI or automation.
Employers prefer retraining internal staff
On the employer side, the dominant approach is not immediate replacement. Eighty-two percent of managers and leaders said they would prefer training existing employees rather than hiring new ones, and 87 percent said their organization would contribute in some form to financing an employee's internal retraining. That choice is rational: AI tools deliver value only when a worker can integrate them into existing workflows, and retraining a person familiar with a company's data, customers and risks often carries less risk than hiring externally.
Methodology note
TripleTen's survey was conducted by Talker Research between September 4 and 11, 2026, via an online questionnaire of 2,000 full-time US workers with internet access. The sample included 500 respondents working in the technology sector. The study was commissioned by TripleTen, which is itself active in the training market and therefore represents an interpretive limitation.
The survey's core message: the next labor-market dividing line will not be who uses AI, but who can tell when not to trust its outputs.
