critique

When Everyone Has AI and the Company Still Learns Nothing

By / Blog / , ,

Robert Glaser discusses the complex “messy middle” phase of AI adoption in organizations, where widespread AI use does not necessarily translate into organizational learning or improved capabilities. He emphasizes the need for companies to develop systems—like Loop Intelligence Hubs—that track and harness AI-driven learning from real work loops to enhance decision-making, distribute useful agent capabilities, and avoid treating AI use as mere token consumption, highlighting that operational control and learning velocity will become key competitive advantages.

https://www.robert-glaser.de/when-everyone-has-ai-and-the-company-still-learns-nothing/

Shooting Down Ideas Is Not a Skill

By / Blog / ,

The article discusses how easily proposed ideas in meetings are often dismissed due to immediate criticism, which requires little effort compared to the imagination and courage needed to create them. It highlights that while identifying flaws is important for preservation, it does not create value, and encourages adopting a mindset that first explores an idea's potential before critiquing it, promoting constructive contributions that build up ideas rather than quickly tearing them down.

https://scottlawsonbc.com/post/shooting-down-ideas

Here’s a Thing – What if Shadow AI Is Actually Telling Us Something Useful?

By / Blog / , , , ,

Dana Louise Simberkoff of AvePoint suggests that shadow AI, like shadow IT before it, signals a cultural stress test within enterprises rather than simply being a technological failure, reflecting a gap between business needs and governance. She advocates for a shift in organizational mindset where employees are treated as stewards of AI, emphasizing trust, clear controls, and distributed judgment to manage AI safely and effectively, rather than imposing restrictive bans that drive usage underground.

https://diginomica.com/heres-thing-what-if-shadow-ai-actually-telling-us-something-useful

EnshittifAIcation

By / Blog / , , ,

In the article “EnshittifAIcation,” Stefano Marinelli describes challenges he faces dealing with AI-driven customer service bots and automated systems in managing e-commerce servers, highlighting issues such as rigid AI responses, misunderstandings about technical configurations, and inaccurate recommendations that ignore expert human input. He argues that overreliance on AI systems without proper human oversight leads to inefficiencies, confusion, and erosion of reliability, emphasizing that current AI lacks the ability to learn or understand context like experienced professionals do.

https://it-notes.dragas.net/2026/03/20/enshittifaication/

AI Still Doesn’t Work Very Well, Businesses Are Faking It, and a Reckoning Is Coming

By / Blog / , ,

Experts from AI advisory firm Codestrap warn that enterprise AI applications often fail to deliver expected benefits due to underlying model limitations and lack of proper metrics to assess AI-generated code quality and business content. They predict a reckoning in 8-9 months as AI misuse leads to failures, lawsuits, pricing pressures, and insurance challenges, urging businesses to adopt clearer strategies, measure true outcomes, and address the hype around AI capabilities.

https://www.theregister.com/2026/03/17/ai_businesses_faking_it_reckoning_coming_codestrap/

Every Layer of Review Makes You 10x Slower

By / Blog / , ,

The article argues that each additional layer of review in a process slows progress by a factor of ten, primarily due to waiting time rather than effort, and this bottleneck is not alleviated by AI coding tools. While reviews are necessary to maintain quality and reduce costly mistakes as organizations grow, excessive layers can degrade efficiency and mask root causes of problems, leading to a culture that values checks over genuine quality improvement. The author suggests adopting a Deming-inspired approach emphasizing trust, continuous systemic improvements, and modular small teams that produce high-quality components to reduce reliance on slow review cycles and create a more effective, scalable engineering culture.

https://apenwarr.ca/log/20260316

Push to Replace Workers With AI Faces Backlash — Even From Management

By / Blog / , ,

Survey shows most workers prefer human collaboration over AI, citing limitations in innovation and customer relations. Companies may face resistance to replacing employees with AI, as many executives believe in human value for critical thinking and relationship building. Concerns also exist about the impact on entry-level hiring and organizational culture. Despite predictions of increased automation in white-collar jobs, experts suggest a cautious approach, highlighting the potential for new economic opportunities alongside AI adoption.

https://www.cio.com/article/4138743/push-to-replace-workers-with-ai-faces-backlash-even-from-management.html

How “95%” Escaped Into the World

By / Blog / ,

“95 percent” of organizations claim no measurable P&L impact from generative AI, a statistic gaining traction but lacking robust academic backing. This claim originates from an MIT report with significant methodological issues: it provides no confidence intervals, and its sample is potentially unrepresentative, mixing various types of data without consistent definitions. The broad applicability of “95 percent” is questionable, as evidence suggests success rates may be closer to 25%. Moreover, the research lacks the scholarly rigor of peer-reviewed work, leading to potential misinterpretations. Overall, this figure should be regarded as unreliable and oversimplified.

https://www.exponentialview.co/p/how-95-escaped-into-the-world

Scroll to Top