Skip to content
Article Smashing Magazine Oct 2026

Smashing Magazine: The organizational mistakes slowing down AI adoption

Paul Boag, author of Digital Adaptation (2014), published this piece in Smashing Magazine on October 16, 2026. It revisits the same mistakes organizations made during digital transformation and finds them being repeated, almost unchanged, with AI.

What the article is about

Boag’s thesis is that AI adoption is a cultural and people problem, not primarily a technology problem. Most organizations treat it as a technology rollout — IT deploys licenses, issues generic training materials, and marks the project done. The result is low adoption, high anxiety, and staff experimenting through personal accounts rather than approved tools.

The article draws on Boag’s consulting work with design and product teams and opens with an example from a UX team that wanted to build a research repository and AI personas. Their official tool couldn’t do it. Getting approval for a different tool meant joining a steering group process. Their enthusiasm was fading.

Key mistakes identified

Ignoring what staff gain. AI primarily increases productivity, which means employees do more in the same time for the same pay. Without a concrete benefit for the individual — career growth, time back, less tedious work — adoption stays low.

Underestimating the disruption. Replacing years of established routines takes time, dedicated training calibrated to specific roles, and protected space to experiment and fail. Organizations that treat this as a one-week training exercise find the skills don’t stick.

Punishing experimentation. If KPIs and performance reviews stay unchanged, failing at an AI experiment has a real cost. One example in the article involves a staff member reprimanded for using too many API tokens while learning. Learning requires mistakes; systems that penalize mistakes block learning.

Governance by prohibition. Long lists of what AI cannot be used for create fear and make inaction the path of least resistance. This pushes staff toward “shadow AI” — personal accounts used to do work-related tasks — which creates exactly the data risk the governance was meant to prevent.

Locking approved tools too early. Different roles need different tools. Defining the approved stack before real use patterns emerge means the right tools are blocked and the wrong ones are mandated. Boag recommends letting teams experiment broadly first and tightening controls once genuine needs are understood.

What Boag recommends

Start by asking what parts of a designer’s or writer’s job are genuinely tedious, and whether AI could take those over. Build the case for adoption around that answer. Gather the staff already experimenting on personal accounts and let them demonstrate what works to peers — peer examples persuade more than management directives. Adjust review criteria so experimentation is explicitly not penalized during a defined learning period.

Who it is useful for

Design team leads, operations managers, and anyone responsible for rolling out AI tools to creative or editorial teams. The diagnosis — that the human side of change matters more than the technology — is not specific to design, but the examples and framing are drawn from UX and product work.