AI Forum

🔗AI Use at Work Has Nearly Doubled in Two Years The use of artificial intelligence (AI) at work in the U.S. has seen rapid growth over the past two years. The share of employees who have used AI at least a few times a year rose from 21 % in 2023 to 40 % in 2025. Frequent use (a few times per week or more) nearly doubled from 11 % to 19 %, while daily usage doubled in the past 12 months alone, going from 4 % to 8 %. This increase is most prominent in white-collar roles. Twenty-seven percent of office-based employees now frequently use AI at work—a 12-point rise since 2024. Adoption is highest in the technology sector (50 %), followed by professional services (34 %) and finance (32 %). Meanwhile, frontline and production workers have seen almost no increase, with usage hovering at 9–11 %. Leaders managing other managers are the most frequent users: 33 % use AI a few times a week or more, compared to only 16 % of individual contributors. The higher adoption among leaders reflects the growing strategic pressure to drive efficiency and innovation at the managerial level. Interestingly, fear of job loss due to AI has not increased. Just 15 % of employees say it’s likely that automation or AI will eliminate their job in the next five years, consistent with past years. However, this fear is slightly more pronounced in industries like tech, retail, and finance (20–21 %). Despite organizational AI integration, clear strategies are lacking. While 44 % of employees report that their organization is integrating AI, only 22 % say there’s a communicated strategy, and just 30 % acknowledge formal guidelines or policies. That leaves a critical 14-point gap in governance. The main challenge to AI adoption is the lack of clear use cases. Only 16 % of AI users strongly agree the tools provided are useful. Employees who have used AI with customers are far more likely to see its value (68 %) than those who haven’t (13 %). Gallup concludes that leaders must demonstrate value and clearly communicate AI plans to drive adoption effectively.

🔗GenAI paradox: exploring AI use cases | McKinsey The “Gen AI paradox”: nearly 80 % of firms use generative AI, yet most report little impact . McKinsey argues for shifting from reactive gen AI to proactive, agentic AI—equipped with autonomy, planning, memory, and integration. They introduce the “agentic AI mesh” architecture and stress that success requires addressing both technical and human challenges: trust, adoption, and governance. Organizational transformation must be driven by the CEO, with a move from siloed pilots to strategic programs, workforce upskilling, infrastructure adaptation, data productization, and agent-specific regulation. The “Gen AI paradox”: the widespread adoption of generative AI has yielded minimal financial impact. This is because organizations deploy horizontal solutions—chatbots, copilots—without reimagining core processes . AI remains reactive; to unlock value, it must evolve to proactive, goal-driven systems. “Agentic AI mesh" implies systems that plan, act autonomously, remember, and integrate across workflows . It distinguishes single-agent from multiagent orchestration. Real-world examples include call centers, HR, finance, and legacy system modernization. McKinsey emphasizes that only the CEO can lead this strategic shift . Businesses must move from isolated pilots to strategic programs, from siloed AI efforts to cross-functional units. This entails redesigning operating models, infrastructure, governance (costs, ethics, accountability), roles, metrics, and corporate culture. Generative AI was a prelude; true competitive advantage lies in agentic AI's ability to automate complex processes and unlock new revenue opportunities. Realizing this potential requires holistic transformation—process redesign, infrastructure, talent, governance—led by the CEO.

🔗https://newsletter.pragmaticengineer.com/p/vibe-coding-as-a-software-engineer by Gergely Orosz and Elin Nilsson 1. What is Vibe Coding? Vibe coding refers to the emerging trend where developers lean heavily on AI tools to generate code—sometimes without fully understanding the details or verifying the results. It blurs the line between AI "taking the wheel" and developers using AI as a support. The difference? In vibe coding, the developer lets the AI run with the task and only intervenes if something looks off. It's like AI pair programming, but with the human taking a backseat—at least initially. 2. Vibe Coding Tools Several tools are powering this movement. GitHub Copilot continues to be a staple, but tools like Claude Code are gaining traction for their ability to handle longer contexts and write coherent code snippets. ChatGPT is still a reliable option for many. These tools allow developers to move faster in early stages of development or exploration. Choice of tool often depends on the developer’s workflow, task complexity, and the model’s coding strength. 3. Use Cases Vibe coding shines brightest in prototyping—when speed and creativity matter more than clean, production-grade code. Developers use it to brainstorm ideas, spin up minimum viable products (MVPs), or even build better developer tools. It offers a low-friction way to test concepts and see what might work—without worrying too much about perfection. 4. Vibe Coding Examples The article highlights two telling stories: one, of an experienced iOS developer who built a working app in just three hours using AI; and another, of a product manager who, stuck with their vibe-coded app, decided to learn Swift to complete it. These stories illustrate both the creative boost and the limitations of the approach: vibe coding can get you surprisingly far—but not always all the way. 5. Reminder: It’s Not Production Ready! A crucial caution: code generated by AI, especially in a vibe coding session, is not ready for production. It may contain security flaws, inefficient logic, costly API calls, or bugs. Developers must review, test, and harden the AI-generated code before shipping it. Skipping this step risks serious consequences in live environments. 6. What Will Vibe Coding Change? Vibe coding is a glimpse into the future of software development. With agentic LLMs becoming more autonomous and capable, engineers will be able to build faster than ever. But far from replacing software engineers, these tools will increase the demand for skilled developers who combine technical depth, product sense, and architectural thinking. Vibe coding won’t remove the need for engineers—it will amplify the value of the best ones.