Josh Tyrangiel, the journalist and former technology executive, has identified a common failing among companies pursuing artificial intelligence: treating it as a single, undifferentiated initiative.
He argues that organisations should stop referring to AI as one thing.
He describes it instead as a complex collection of distinct scientific techniques, each suited to different problems.
Leaders who fail to make that distinction risk confusing their own teams about what they are actually trying to build.
Tyrangiel advises executives to be specific with employees about which AI techniques apply to which business problems.
He suggests this specificity can also ease worker anxiety, since vague talk of "AI transformation" tends to fuel uncertainty rather than clarity.
Precise questions, not broad ambitions
According to the commentator, companies that successfully extract value from AI share a common habit: they ask precise questions about specific business problems, rather than pursuing AI adoption as an end in itself.
He suggests a practical starting point is for businesses to identify where "code" already exists within their operations, areas where processes are structured enough that AI techniques can be meaningfully applied.
Data quality is another prerequisite, Tyrangiel says.
Organisations need to assess whether their data is clean enough to use immediately or requires significant preparation before any AI effort can proceed.
The CEO knowledge gap
Tyrangiel points to a structural problem sitting above these technical questions: many chief executives lack technical backgrounds.
That gap, he argues, can make CEOs reluctant to engage with the specifics of AI technology, leaving strategy vague by default.
His solution is for CEOs to build a much closer working relationship with their chief technology officers.
Rather than treating meetings with CTOs as optional or occasional, Tyrangiel argues that chief executives should invest real time in learning to communicate with them effectively.
Two languages, one strategy
He frames this as CEOs and CTOs needing to learn to speak each other's languages.
That shared vocabulary, Tyrangiel suggests, is what allows leadership to manage the experimentation AI requires and to plan for how their businesses may need to be reconfigured as a result.
Without it, he warns, companies risk repeating the same mistake: chasing "AI" as a broad ambition, rather than applying specific techniques to specific problems where they can demonstrably work.
