HR's Perspective on Three Years of ChatGPT at Work

The conversation about how AI is transforming the landscape of work is far from straightforward. Since ChatGPT’s launch in 2022, AI has disrupted the workplace, but questions about AI's productivity and profitability remain open. On the flip side, AI enthusiasts are pushing for even wider adoption. All the while, the question of AI regulation.

What does it all mean for your organization, and your teams?

To answer these and other burning questions, we compiled a new editorial research report grounded in conversations with HR leaders, founders, and practitioners across industries. We asked how AI is really being used, where it’s helping, and where gaps in skills, policy, and trust are still creating friction.

The leaders responding most effectively aren’t chasing every new tool. Instead, they’re setting clear AI policies, offering targeted training, redesigning workflows, and defining how AI fits into hiring and performance decisions. Along the way, they're tracking a small set of KPIs for adoption, quality, and trust.

In this editorial report, you’ll get a practical roadmap for HR teams navigating the next phase of AI at work. Keep reading for what’s stayed true, what’s changed, and what to do differently with ChatGPT (and other AI platforms) in 2026.

Three “AI-isms” from 2022 that are still true

1. AI use is high but uneven

In 2022, most employees met ChatGPT on their personal devices. By 2025, many are using a mix of copilots, niche tools, and homegrown scripts across almost every function—whether IT has rolled out an approved platform or not.

Today, AI touches many of the everyday documents and tasks that shape internal collaboration and external relationships. In the past year, almost half of all employees used AI to:

• Automate repetitive tasks and retrieve information (48%)
• Prepare presentations (42%)
• Edit and polish work (42%)
• Draft client emails (41%)

The data shows that daily AI use is common and growing, especially among senior leaders and in large companies. But adoption isn't uniform across the workforce. AI use remains much lower among individual contributors and in smaller firms.

Who is using AI daily?

The leaders we interviewed see the same pattern in practice.

• Recruiting: Prem Kumar, Co-Founder of Humanly, described high-volume hiring where an AI interviewer speaks with thousands of applicants and routes the best matches to human interviewers, instead of forcing recruiters to “compare two pieces of paper” and skim the first 200 résumés.
• Learning, talent, and HR: Leaders like EY Partner Darryl Wright and former Charles Schwab Managing Director of Employee Experience Clayton Newman described copilots summarizing long policy documents, drafting manager communications, and turning raw data into first-pass dashboards across their HR tech stacks.
• Coaching: Coaches and outplacement firms use AI companions to help people structure job searches and prepare for interviews. Respiris Founder and CEO Angel Cruzado is building AI “companion coaches” so human coaches can focus on the emotional and strategic work only they can do.

In 2026, the question is less “Are we using AI?” and more “Who is using it, on what, and with what support?” For HR, the goal is to set clear guardrails, build shared literacy, and place AI where it improves real workflows—especially in hiring, development, and leadership work.

2. Policy and training lag behind enthusiasm

Most employers allow AI, and most employees are not waiting for permission to try it. They do, however, look to leaders for guidance and training. Those expectations often outpace what organizations currently offer.

• 77% of employees say their employer either encourages AI use or does not restrict it.
• 72% of employees want to improve their AI skills, but only 32% have received formal AI training from their employer.
• 45% of employees at larger companies receive employer-provided AI training, compared with 21% at companies at smaller companies (defined as 500+ employees vs. < 50 employees respectively).

The research also points to a communication gap. Leaders believe they are sending a clear “use AI” message. Many employees do not hear it:

• 56% of executives say their organization actively encourages AI use.
• 41% of managers say that the message reaches them.
• 17% of individual contributors say the same.

At the same time, when AI halves the time to draft, analyze, or research, several interviewees said the response tends to be, “Great—now do more.” Economic pressure, return-to-office pushes, and "9-9-6"-style narratives then layer on top, raising expectations further. As a result, employees operate in a confusing intersection of:

• Permission or tolerance for AI use
• Strong interest in getting better with AI
• Limited access to structured learning
• Rising expectations for productivity and efficiency

The result is an AI divide. Those closest to power and strategy make heavy daily use of AI. Those closer to the work feel more uncertainty and receive less formal support.

HR leaders tend to see the downsides first:

• Burnout and higher turnover in high-intensity teams
• Uneven adoption, where some groups gain from AI while others fall behind
• Constant “urgency inflation,” where everything feels like a fire drill

For 2026, AI literacy, clear guardrails, and clear boundaries on time and urgency belong at the center of HR’s AI strategy. Employees now see AI as both a valuable productivity tool and a potential driver of overwork. HR needs to set standards for training, working hours, right-to-disconnect practices, and what truly counts as urgent—not treat those as separate wellness topics.

3. AI is reshaping work, quality, and trust

In almost every interview, leaders returned to the same point: The most important skill is no longer, “Can you do this task end to end?” The real test is, “Can you break the work down, decide what AI should handle, and judge the result?”

Recruiting shows this shift clearly. A recruiter who only knows how to run a standard requisition will struggle. A recruiter who can design a flow—AI interview first, human conversation second, clear transparency for candidates, and defined human checkpoints, for example—can handle far more volume without losing quality.

Other HR domains look similar. Interviewees described HR moving from “AI as another reporting tool” to “AI as a digital assistant that sits across the tech stack and orchestrates the experience.” HR analysts now need to design prompts, interpret model output, and push back when AI suggestions do not match the business reality.

When employees use AI in real work, leaders often like what they see. But individual contributors are more cautious. That gap shows up in the data:

• 63% of employees say AI-generated work is high quality, but only 28% of individual contributors agree.
• 61% of executives say they value employees who use AI more, yet only 13% of individual contributors believe leaders actually value AI use that way.
• 23% of individual contributors do not disclose AI use, compared with 6% of executives.

Security and detection concerns also add to this tension:

• Only 30% of employees can correctly identify AI-written content, even though 54% believe they can.
• Only 49% of women see AI as secure, compared with 67% of men.
• Only 27% of individual contributors see AI as secure, compared with 74% of VP/C-suite executives.

Employees use AI, yet many are unsure whether they can trust its output or their employer’s stance on it. Leaders say they reward good AI use, but most individual contributors do not see that recognition. Meanwhile, quiet, undisclosed AI use is common, which raises both ethical and operational risks.

Often, these issues are most visible in hiring and recruiting:

• 48% of hiring managers use AI to manage high application volumes.
• 58% use AI to help predict a candidate’s long-term success.
• 29% of HR managers have rescinded offers because candidates used AI during interviews.
• 25% of HR leaders worry AI could reinforce bias in hiring, while 44% see it as a way to reduce bias if it is designed and monitored well.

AI is already screening, scoring, and shaping hiring decisions. HR leaders know they must balance speed and scale with fairness, quality, and candidate trust. They are also managing AI-related anxiety as employees watch layoffs and wonder how automation will affect their own roles.

To improve quality and trust while AI changes how work gets done, HR needs to turn this new reality into an actionable practice. That means:

• Defining when AI use is encouraged and when it is off-limits
• Making expectations about disclosure explicit for employees and candidates
• Spelling out what “good” AI-assisted work looks like in each function
• Requiring human review on high-stakes decisions
• Updating skill models so every critical role includes three layers: core functional skills, baseline AI literacy, and “system skills” for orchestrating AI, data, and people

AI now sits in the middle of real work, quality, and trust. In response, HR must design the systems around it so employees know when and how to use AI tools, leaders know how to judge AI-assisted work, and candidates and employees can trust how it shows up in decisions. Ultimately, the choices HR makes in 2026 will shape how work feels long after this first wave of AI tools.

Two AI misconceptions to leave in 2025

1. “Everyone is already using AI, so the gap is closing on its own”

Many leaders feel late because they lack agents embedded in every HR process. Our interviews suggest something different: most organizations are still in phase one or two—governance, literacy, and targeted pilots.

The most effective leaders are focusing on:

• Clear principles for where AI is appropriate
• A safe sandbox for teams to experiment
• Prompt libraries and shared patterns by role
• A short list of high-impact workflows to redesign

That approach beats a rushed, top-down rollout of AI “in everything” that nobody trusts. For 2026, the useful question is not “how many processes have AI in them.” It is “where does AI measurably improve the experience or outcome, and what did we stop doing as a result.”

2. “AI is a tech project, and HR's job is to keep it safe on the sidelines"

Technical architecture and guardrails are non-negotiable for AI.

But every leader we spoke with eventually came back to people issues as the real bottleneck:

• Supervisors who don't know how to evaluate AI-assisted work
• Employees who worry AI will be used against them, not for them
• Bias baked into training data and processes
• Talent systems that still assess people as if AI did not exist

The real differentiator is how quickly people can learn to work with AI and how fairly organizations create that opportunity. That is core HR work—it touches workforce planning, skills, job design, performance, inclusion, and wellbeing. For 2026, HR cannot treat AI as something “owned by IT” with a few policies bolted on. HR has to co-own the strategy, design the human side, and keep an eye on who benefits and who gets left out.

Four practical questions to shape AI at work in 2026

Planning for AI in 2026 doesn’t require a massive strategy document. It does require a clear stance, a few focused decisions, and a plan you can explain in simple terms. The questions below keep AI planning grounded in real work and real people.

1. Where will AI show up in the employee experience?

Start with a small set of journeys to redesign with AI support end-to-end. For example:

• High-volume hiring for frontline or early-career roles
• Onboarding for critical roles and teams (engineers, managers, sales)
• Performance and development for people managers

For each journey, you should map:

• Tasks AI will handle (screening, summarizing, drafting, routing)
• Tasks that stay with humans
• How you will explain AI’s role to candidates and employees
• How you will check for bias, errors, and drift over time

Then decide what work you will remove or simplify, so AI doesn't just add more volume to the same people.

2. What AI skills should every employee have by the end of 2026?

Define a baseline for AI literacy across the organization. For example:

• Basic understanding of generative AI and its limits
• Ability to write a clear, task-focused prompt in your main tools
• Ability to review AI output for accuracy, bias, tone, and data risk

Layer role-specific expectations on top of that baseline. A recruiter, plant supervisor, customer success manager, and HRBP will each use AI in different ways.

Use the data to support your people:

• Employees want to improve their AI skills but don't have access to formal training.
• Employees in smaller companies receive less employer-provided training.
• Individual contributors and older workers have lower daily use and lower trust in AI.

Those groups are strong candidates for structured learning and hands-on practice, led by HR and reinforced by managers. Tie AI skills to real tasks, not abstract concepts.

3. How will we handle equity, trust, policy, and time?

Employees want clarity on how to use AI well and safely. AI policy needs more than a simple “allowed / not allowed” line.

Build a plan that covers:

Align recognition and promotion criteria with your AI stance. If leaders say they value effective AI use, employees should see that value reflected in performance and growth conversations.

Then watch the communication gap carefully. Executives say they encourage AI use. Many individual contributors say they do not hear that message. Managers sit in the middle. HR can support them with talking points, FAQs, and real examples they can share with teams.

4. How will we govern AI in high-stakes decisions and adjust over time?

AI already plays a central role in recruiting and will likely expand into performance, promotion, and succession planning. Treat these areas as priority zones for design and oversight. In hiring, for example:

• Decide where AI can assist (e.g., screening large applicant pools, highlighting patterns in assessments, drafting interview questions).
• Decide where humans must lead (e.g., final hiring decisions, evaluating culture add, interpreting nuanced answers).
• Set expectations for candidates about acceptable AI use (e.g., prep, note-taking) and where it crosses into misrepresentation.
• Plan for regular bias audits to check real outcomes, not just intentions.

Apply the same structure wherever AI touches ratings, pay, or advancement. Keep humans in the loop for high-stakes people decisions.

Treat 2026 as a learning year. Set up simple loops to listen and adjust:

• Create a cross-functional group for AI and work-design decisions (HR, IT, legal, operations, and frontline managers).
• Use regular surveys and focus groups to track employee comfort, training needs, and trust.
• Track a small set of metrics: AI usage by level and function, training coverage, disclosure rates, and employee perceptions of security and fairness.

Use those signals to refine use cases, training, policies, and workload expectations. Adjust early when something creates more friction than value.

Where HR goes next

AI in the workplace is no longer a distant trend. Executives use it daily. Many employees do too. Many others want to use it more effectively and securely. At the same time, economic uncertainty, AI-linked layoffs, and shifting regulations sit in the background of every adoption conversation. Employees weigh opportunity and risk simultaneously, and leaders do the same.

For HR, the task in 2026 is straightforward: Make AI useful, fair, and understandable.

Close the training and communication gaps. Set simple rules for quality, trust, disclosure, and time. Design around the people who are not yet confident with AI, not just the early adopters.

The tools will keep changing. But the foundations HR builds now—skills, policies, guardrails, and work-design practices—will shape how AI functions at work for the next several years.

How this mini-report came together

This report is a high-level overview of AI at work, building on BambooHR News coverage throughout 2025. It pulls together themes from our first two mini-reports—one on 9-9-6 culture and overwork, and one on AI in the workplace—and steps back to ask a broader question: what has actually stuck since late 2022, and what should HR leaders carry into 2026?

First, we conducted dozens of interviews with people leaders, founders, and consultants—most of which also inform standalone feature stories in trade news publications and social media posts.

BambooHR Data at Work research supplied the quantitative backbone of this report, including: Clarity over Chaos: Bridging the AI Divide; Workforce Insights—Hiring Edition (July 2025); Employee Happiness Index Q3 2025; Welcome to the Eggshell Economy: America's Workforce in Survival Mode; The Great Grin-and-Bear-It: Q2 2025 Employee Satisfaction; and Executives Aren't as Happy with RTO Mandates as You Think: Q3 2025 Data Stories.

Finally, the editorial team compared themes across the two earlier mini-reports, the interview transcripts, and the research to identify repeating patterns by role, industry, and company size. The result is a 360-degree snapshot of AI at work in 2025 and a practical stance on what to leave behind, what to keep, and what to build for 2026.