ServiceNow's maturity index finds AI leaders 14 times likelier to fund ongoing staff training

The index's figures are internally consistent. Weight the two groups' average scores by their shares of the sample and the result is about 51 ^[1], the same as the index's reported average for this year ^[12]. Apply the same weights to unsupervised multi-step workflows, run by 36% of pacesetters and 2% of the rest ^[9], and you get about 9% ^[2], the share David gave for the whole sample ^[8].

The same weights put about 15% of all surveyed organisations among those funding ongoing AI upskilling, and roughly four in five of those funders are pacesetters ^[3]. That concentration can be read three ways. Training money could be what lifts a firm over the 60-point line ^[5]. Firms already over it report an average ROI of 160% ^[11], and they could simply be the ones with cash to spend on upskilling and AI hires. Or talent could be one of the inputs to the score, so a talent gap is partly how pacesetters get defined. Fortune's account of the report does not say which questions feed the 100-point score ^[5].

"Talent investment. That was the biggest gap," David told Jeremy Kahn, Fortune's AI editor ^[2]. Measured in percentage points (or rather, measured the one way that supports her), the widest gap is hiring and keeping AI talent at 58, with upskilling at 53 and unified data at 50 ^[4]. Ratios favour anything the laggards almost never do. On that measure autonomous workflows lead at 18 to 1, upskilling runs about 14 to 1 and talent hiring about 7 to 1 ^[5]. The data gap is three points narrower than the training gap ^[4].

I think the people gap is real. The survey supports people and data moving together better than it supports training budgets as the single deciding factor. The counter-case is that training is the gap a firm can close soonest: 42% of employees say they are not getting enough AI training, and 59% of organisations lack long-term HR plans for AI ^[10]. David's own framing went wider than training. "It's really operational discipline, right?" she said ^[16].

Kahn described what many companies did instead. They gave employees AI tools and "let a thousand flowers bloom," he said, then decided it was too expensive with no productivity gain, and are now narrowing to specific use cases ^[14]. David's objection was to headline adoption figures. "But it is not translating into productivity, and it's often not with training, and that doesn't equal maturity," she said ^[15].

Settling it would take the same firms tracked over two years. If firms below 60 this year that start funding upskilling cross the line next year faster than firms that put the money into data, training is the lever. If the firms that cross first are the ones already reporting returns, the training spending is following the returns.