Observed Signal · Mar 26, 2026 · Research Report · Source: https://martech.org/feed/ · Impact: 3/5 · Sentiment: Positive
Study: Top 5% of AI Users Drive Impact
Researchers from KPMG and the University of Texas business school analyzed 1.4 million workplace AI interactions at KPMG and found that, while roughly 90% of 2,500 employees used AI tools, only about 5% met the researchers’ definition of “highly sophisticated” users. Those high-performing users engage in longer, iterative sessions, assign roles and examples to the model, delegate complex multi-step tasks, switch models/tools by task, and treat AI as a collaborative partner. The study concludes these behaviors do not spread organically and recommends organizations codify AI-first best practices, provide hands-on scenario-based training, and set role-specific expectations. The article cites broader workforce data from the World Economic Forum about widespread training needs and includes commentary from researcher Zach Kowaleski and media thinker Douglas Rushkoff on treating AI as a generative, exploratory medium.
Empirical evidence that only a small share of users drive meaningful AI value and that behaviors can be taught; actionable guidance (playbooks, training, role expectations) is relevant to marketing and operations teams but is not a platform-level or regulatory shift.
Track KPMG Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- KPMG and the University of Texas business school analyzed 1.4 million real workplace AI interactions at KPMG.
- Among 2,500 employees studied, ~90% used AI but only 5% were classified as "highly sophisticated" users.
- Highly sophisticated users demonstrate behaviors such as longer interactive sessions, iterative prompting, role assignment, multi-step delegation, model/tool switching, and conversational prompting.
- Researchers recommend codifying AI-first best practices, hands-on scenario-based training, and role-specific expectations to scale high-impact AI use.
- The World Economic Forum survey noted on average 59% of employees will need additional training to meet evolving skill demands by 2030.
Connected Companies & Entities
2 Entities mappedRelated Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
Most Users Don’t Use AI Effectively
A newsletter by The Algorithmic Bridge argues that AI usage is highly concentrated: a tiny minority of power users extract most of the productivity gains while the vast majority treat tools like ChatGPT as toys or use them minimally. Citing an OpenAI chart, the author reports that top power users (above the 95th percentile) employ seven times more advanced "thinking capabilities" than the median paid user. The piece cites raw user figures (≈800M weekly active users in June 2025; ~900M at time of writing; ~50M paying users) and warns of an emerging divide—an "AI-rich" minority and an "AI-poor" majority—affecting firms’ productivity, geographic concentration of advantage, and enterprise adoption strategies. The essay discusses Davos remarks from Anthropic and Microsoft leaders and provides advice for companies and employees on closing the adoption gap.
AI Experience Gap: What Employees Get from AI
AI is touted as a time-saver in the workplace, yet real-world gains often fall short. A study by Pendo in collaboration with Appinio finds that while 85% of respondents view AI as exciting and full of potential, many experience frustration as they must continually adjust prompts and rework queries. Among regular users, 96% report weekly prompt adjustments, and 42% refine inputs until the AI response fits their needs. Although 36% say AI saves more than an hour per week, the benefit is partially offset by repeated corrections. Trust in AI results remains fragile, with only 30% believing the outputs are correct and 36% feeling confident to operate tools effectively. Consequently, 80% of employees desire targeted training or clear guidelines from their employers. The study highlights an AI-experience gap that can erode ROI unless user understanding and AI-agent UX are improved, a point underscored by Pendo’s Todd Olson, who emphasizes studying user behavior to optimize outcomes.
AI's Biggest Opportunity: Creating New Value
The article argues that while most organisations use AI (88% per McKinsey), only a small share (6%) see significant enterprise-wide impact because companies mainly apply AI to existing tasks rather than rethinking business models. Only 23% of generative-AI users have redesigned workflows for the technology. The author contrasts a 'factory' (efficiency) mindset with a 'laboratory' (experimentation and effectiveness) mindset and recommends marketing operations lead experimentation to discover new revenue models. Examples include Pieter Levels' portfolio of experiments generating sizable monthly revenue and IKEA’s chatbot 'Billie', which resolved 47% of inquiries, triggered reskilling of call-centre staff into design advisers and produced €1.3 billion in new revenue in 2022. The piece emphasises deliberate, low-cost experimentation to move organisations into higher-value AI stages.
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.