We know the recipe for costly IT fiascos: now we're applying it to AI

The source for this piece is only a short teaser, so the summary can cover just what it states. It is an opinion column (klumme) titled "We know the recipe for costly IT fiascos: now we're applying it to AI."

The column draws on research by Bent Flyvbjerg and colleagues. They have documented that IT projects carry the most extreme cost risk of 23 project types they examined. In other words, when IT projects go wrong on budget, they go wrong more severely than projects in any of the other categories studied.

The column's central argument is that the most important lesson from this research is not about technology. It is about how management teams make decisions. The cost overruns and failures are presented as a product of decision-making habits rather than of the tools involved.

From that, the column draws a warning about AI. It says the outlook is not good for the AI projects now lining up to be launched, suggesting that organisations are about to repeat the same decision-making patterns that have produced costly IT failures in the past.

The teaser does not say which specific decision-making mistakes are involved, which organisations or AI projects are meant, or what the author recommends doing about it. Readers who want those details, or the full reasoning behind the comparison between past IT projects and current AI initiatives, would need the full column.