June 21, 2026
AI gold rush, buyer’s remorse
CTOs Agree: Cognitive Debt Is the New Technical Debt
Bosses splurged on AI, now the internet says they bought a bigger mess
TLDR: Tech leaders now say the real problem isn’t just messy software, but the mental burden of handling huge amounts of AI-made work while proving it was worth the cost. Commenters were harsher: many said executives chased a trend, and AI is mostly exposing bigger management and workflow problems.
A room full of tech bosses in Toronto basically admitted the AI spending party is over. After two years of buying shiny tools with few questions asked, the mood has shifted to one painfully simple question: what did we actually get for all this money? The article says companies are struggling to measure whether these tools are helping, because costs can balloon unpredictably and the payoff is still fuzzy. In plain English: management signed the checks, and now finance wants receipts.
But the real fireworks came from the comments, where readers were absolutely not in a forgiving mood. One poster translated the whole thing as: AI is “next to useless,” slows teams down, and burns cash while throwing away the one thing companies already had — workers who actually knew their own systems. Another went straight for leadership’s throat, calling recent executive decision-making pure bandwagon behavior. Ouch. Others argued the new buzzword, “cognitive debt,” might just be old-fashioned technical mess with a trendier label, while some said the real problem is human overload: too much code, too much to review, too much to keep in one person’s head.
There was even a darkly funny vibe running through the thread: congrats, AI may help produce code faster, but now it reveals all the other bottlenecks nobody wanted to talk about — paperwork, approvals, and the soul-crushing task of checking machine-made mistakes at every level. The crowd’s verdict? The robots didn’t kill software jobs — they may have just created a bigger headache for everyone in charge.
Key Points
- •Engineering leaders at a CTO Craft dinner in Toronto said organizations still have not fully figured out AI adoption in engineering.
- •The article says the AI spending phase with minimal scrutiny has ended and has been replaced by demands for measurable ROI.
- •Participants described token-based AI usage as difficult to forecast and manage, making the investment side of ROI unclear.
- •Several companies reportedly bought AI tools before developing internal financial models, a situation compared to early cloud adoption before FinOps.
- •Some teams are responding by standardizing AI tool usage, while hiring discussions are increasingly separating AI product work from system design responsibilities.