Companies keep making the same mistake with artificial intelligence: they talk about it as though it were one thing.
That is the central warning from journalist Josh Tyrangiel, who argues that AI is in reality a sprawling collection of distinct scientific techniques, each suited to different problems.
Treating it as a monolith, he says, is where organisations go wrong from the outset.
Precision over platitudes
Leaders who want their teams to make progress need to be specific about which technique applies to which problem, rather than issuing broad directives to "use AI".
That specificity does more than sharpen strategy; it also calms nervous staff who fear a vague, all-encompassing technology bearing down on their jobs.
Tyrangiel notes that the companies genuinely extracting value from AI share a common habit: they ask precise questions about specific business problems, rather than chasing the technology for its own sake.
His practical starting point is deceptively simple.
Businesses should look first at where "code" already exists within their operations, the processes and decisions that follow rules and patterns AI can plausibly learn or automate.
From there, the harder question follows: is the underlying data clean enough to use, or does it need significant preparation before any model can be trusted with it?
The CEO's blind spot
A recurring obstacle, according to Tyrangiel, is the CEO.
Many chief executives lack technical backgrounds, and that gap breeds reluctance to engage with the specifics of what AI can and cannot do.
The consequence is a leadership vacuum at precisely the moment detailed direction is needed.
Tyrangiel's answer is for CEOs to build a far deeper working relationship with their chief technology officers than has traditionally been the norm.
That means treating meetings between the two roles as essential rather than optional, and investing real time in learning to speak each other's language.
Only then, he argues, can a chief executive communicate AI goals clearly enough for the rest of the workforce to act on them.
The reconfiguration of a business around AI is not a one-off project but an ongoing process of experimentation, and it needs sustained, informed leadership rather than a single strategy memo.
Tyrangiel's argument amounts to a rebuke of vagueness at every level of the organisation, from the language executives use in public to the questions they ask in private.
Companies that keep treating AI as an undifferentiated buzzword, rather than a toolkit of specific techniques matched to specific problems, will keep failing to extract value from it.
