A recent study by KPMG, conducted in partnership with the University of Melbourne, reveals that 57% of employees admit to concealing their use of artificial intelligence tools from their managers and colleagues. This trend not only raises concerns about transparency and trust within organizations but also poses significant risks related to data management and compliance. The report, titled “Trust, Attitudes, and Use of Artificial Intelligence: A Global Study 2025,” surveyed over 48,000 individuals across 47 countries. It found that 58% of employees actively use AI in their work, with a third doing so at least weekly. Alarmingly, 66% of these users do not verify the accuracy of the AI-generated results, and nearly half have shared sensitive business information on public AI platforms.
A recent study has revealed that 42% of consumers believe artificial intelligence and ChatGPT can manage complex customer service inquiries as effectively as humans. However, despite the growing acceptance of AI in customer interactions, 51% of U.S. customers reported receiving incorrect information from AI self-service bots, highlighting the technology’s limitations. The study, conducted by Shep Hyken, surveyed over 1,000 U.S. consumers and found that 50% successfully resolved customer service issues using AI without human assistance, a significant increase from 32% in the previous year. Additionally, 63% of customers anticipate that AI technologies will become the primary mode of customer support. While younger generations, like Gen-Z, are more comfortable with AI, privacy and security concerns remain prevalent, with 70% of consumers expressing worries about these issues when interacting with AI.
The role of prompt engineer, once considered one of the hottest jobs in artificial intelligence, is now largely obsolete due to rapid advancements in AI technology. According to Jared Spataro, Chief Marketing Officer of AI at Work at Microsoft, the necessity for specialized prompt engineers has diminished as AI models have evolved to better understand user intent and ask clarifying questions. A recent survey by Microsoft revealed that among 31,000 workers across 31 countries, prompt engineering ranked second from the bottom in terms of new roles companies are considering adding. Meanwhile, roles such as AI trainer, AI data specialist, and AI security specialist topped the list. Job postings for prompt engineers have significantly declined, with searches on the job platform Indeed peaking shortly after the introduction of ChatGPT but now averaging only 20 to 30 searches per million.
LinkedIn has revealed significant insights into the adoption of artificial intelligence tools among businesses, emphasizing that companies integrating these technologies have seen notable improvements in productivity and revenue. According to LinkedIn’s latest economic impact report, 75% of businesses utilizing generative artificial intelligence reported time savings and revenue increases exceeding 10%. Despite these advantages, the report highlights a concerning trend: small and medium-sized businesses are lagging behind larger corporations in AI adoption. The report indicates that while 62% of companies in India and 51% in the United States are adopting AI at high rates, small and medium businesses struggle to keep pace, potentially missing out on major economic benefits.
Why do we care?
We care because these insights collectively highlight critical inflection points in the enterprise adoption of AI—where hype gives way to operational and ethical complexity, and where productivity gains are now tempered by trust issues, uneven access, and growing risks.
Ungoverned AI adoption is a double-edged sword: while it fosters grassroots innovation, it also poses serious risks related to data leakage, compliance (especially with regulations like GDPR), and model reliability. More critically, the fact that 66% of users do not verify outputs illustrates how companies may unknowingly base decisions on flawed information—creating downstream operational and reputational risks.
The failure to launch of the prompt engineer role, as Microsoft reports, underscores the rapid commodification of AI interfaces. As models become more conversational and context-aware, specialized prompting becomes less necessary. This serves as a useful signal for workforce planners and service offerings: AI fluency is becoming more general-purpose, not niche. If larger companies are realizing revenue gains of 10% or more, smaller firms lacking access to capital, talent, or infrastructure risk falling permanently behind. Providers have a significant opportunity—and responsibility—to develop AI “starter kits” or industry-tailored offerings for SMBs, thus closing this adoption gap affordably.