Field Notes
ReportNov 11, 2026 · 11 min read

What companies actually intend when they “implement AI”

Read a company's AI announcement and you will usually hear about transformation and value creation. Look at what companies actually did in 2026, and my reading is that three other…

Greg JeffersonGreg JeffersonFounder, Everyday Creative Works

Exploring how AI is changing real work.

Read a company's AI announcement and you will usually hear about transformation and value creation. Look at what companies actually did in 2026, and my reading is that three other intentions do most of the work: funding infrastructure, cutting ahead of results, and bolting tools onto old workflows.

The short version

Experimentation is widespread, but payoff is not: in PwC's survey of 4,454 CEOs, fielded from September to November 2025, 56% said they had seen no significant financial benefit from AI to date, and only 12% said it had delivered both cost and revenue benefits [1]. Workforce decisions are running ahead of the technology: in a Return on AI Institute survey of 1,006 executives, released in March 2026, only 2% had made large headcount cuts tied to real AI implementation, yet nearly 90% had already reduced or frozen hiring in anticipation of future AI productivity gains [2]. Companies that cut staff did not report better returns: Gartner (May 2026) found workforce reduction rates were nearly equal for companies with higher and lower AI returns [3], and Wharton and GBK Collective (October 2026) found organizations reporting strong returns were no more likely than others to report staff reductions [4]. The few AI high performers are also the ones redesigning work: in McKinsey's 2026 survey, just 6% of respondents qualify as AI high performers, and nearly three-quarters of them report substantially redesigning workflows, compared with one-quarter of other organizations [5].

What companies say

Ask CEOs about their priorities and AI comes first. In a SHRM survey of CEOs reported by HR Dive in December 2025, AI adoption was the most frequently cited primary objective for 2026, ahead of growing revenue [6]. James Atkinson, SHRM's vice president of thought leadership, said executives increasingly believe AI can create value instead of solely being used to cut costs [6].

Then Wharton and GBK asked 1,416 U.S. leaders and employees, in a study published in October 2026, what AI's primary job is over the next 12 to 24 months, and 52% of leaders said improving productivity or reducing costs [4]. Only 22% pointed to growing the business through new revenue, new products, or better customer experience [4]. My reading is that the stated ambition is transformation and the working assignment is efficiency.

Money is flowing either way. The National Bureau of Economic Research working paper "Firm Data on AI" (revised October 2026, based on a summer 2026 survey of nearly 9,000 executives in seven countries) reports that US firms expect to spend $2,000 per employee on AI in 2026, implying around $300 billion in aggregate [7]. That spending is highly concentrated: just 2% of firms account for over 50% of the total [7].

Four intents behind the word

Read the behavior rather than the press release and four distinct intentions show up: funding the AI buildout, cutting ahead of results, bolting tools onto old work, and redesigning the work. My reading is that most companies run more than one at once.

Intent one: fund the buildout

For the largest tech companies, my reading is that layoffs and AI spending are tied together by budget math. Meta said it was laying off about 8,000 workers, about 10% of its workforce, for the sake of efficiency and to allow new investments in parts of its business, the Associated Press reported [8]. Meta had warned investors that its 2026 expenses would reach $162 billion to $169 billion, driven by infrastructure costs and employee compensation, particularly for AI experts [8]. Bloomberg reported that Meta would also leave about 6,000 jobs unfilled [8].

Reuters' reporting, covered below, points to a replacement motive at Meta as well, so my reading is that Meta sits in more than one intent at once. On October 26, 2026, Fortune reported that Microsoft was offering buyouts to roughly 60,000 employees, about 27% of its workforce, and that the company has been cutting headcount to trim costs while investing billions to build out AI data centers [9]. In these cases, my reading is that "AI" mostly names where the money is going.

Intent two: cut ahead of results

Thomas Davenport and Laks Srinivasan documented this pattern in Harvard Business Review in January 2026: in a December 2025 survey of 1,006 executives, which appears to be the same survey behind the Return on AI Institute figures above, they found AI-linked layoffs were almost completely in anticipation of AI's impact [10]. Gartner's survey of 350 executives at companies with at least $1 billion in annual revenue, reported by Fortune in May 2026, found that 80% of those that had piloted AI or autonomous technology reported workforce reductions, made regardless of whether the technology was generating returns [3]. Gartner analyst Helen Poitevin told Fortune the companies reporting high ROI were not the same ones reporting AI-related workforce reductions [3].

Challenger, Gray & Christmas counts show how often employers named AI. Employers cited AI in 121,401 announced job cuts through October 2026, approximately 20% of all cuts [11], more than double the 54,836 cited for all of 2025 [12]. By October, AI-cited cuts fell to 1,265, the lowest monthly total of the year, and market and economic conditions had overtaken AI as the leading reason for job cuts in 2026 [11].

Intent three: bolt tools onto old work

My reading is that the most common intent is also the least dramatic: give everyone a license, count usage, and hope the gains add up. In McKinsey's 2026 survey, 80% of respondents said AI has improved their individual productivity, while only 37% reported positive EBIT impact [5]. Executives in the NBER survey report that AI raised annual labor productivity growth by 0.2% over the last three years and had essentially no net effect on firm employment [7]. The authors note that this small net effect masks considerable churn: more than 40% of firms anticipate positive or negative employment effects that roughly offset only in the aggregate [7].

Wharton and GBK found that 67% of leaders report using AI agents in some form, but only 28% say AI is fully embedded in core business processes, and just 17% let agents act without a person approving each step [4]. GBK's Brian Smith described most companies as paying for autonomy and then reviewing every action, which he said is where returns are weakest [4].

Intent four: redesign the work

McKinsey's high performers behave differently. McKinsey's 2026 survey of 1,719 participants, in the field from May 4 to June 8, 2026, found its high performers pursue growth or innovation alongside efficiency, redesign workflows around AI rather than insert AI into existing ones, and back deployments with leadership commitment [13]. Wharton and GBK found companies measuring AI against growth outcomes were far more likely to report strong ROI than those focused on efficiency alone, 35% vs. 13% [4]. Among departments that met the study's workflow transformation measure, 67% reported strongly positive ROI, compared with 28% of enterprise leaders overall [4].

The reversals

My reading is that some companies that cut with replacement in mind have since walked it back. Reuters reported on August 26, 2026 that, to make its workforce "AI native," Meta explored slashing many teams by as much as 60% in two waves [14]. Meta laid off 10% of its employees on May 20, 2026, but called off planning for the November cuts, according to an internal document Reuters reviewed, and in early July Mark Zuckerberg conceded miscalculations on the reorganization's timing at an internal town hall [14].

Robert Half research, published in its labor market outlook through year-end 2026, found that 32% of U.S. hiring managers who eliminated positions after adopting AI later had to reinstate the same or similar roles [15]. Robert Half said the findings reinforce the risks of treating AI as a substitute for employees in roles that still depend on context, relationships and judgment [15]. In SHL's survey of 2,036 employed adults in the United States and the United Kingdom, conducted in June and July 2026, nearly one in five American respondents (19%) said they had firsthand experience of being replaced by AI and then asked back [16][12]. Fortune reported in November 2026 that Ford hired and promoted 350 experienced engineers in June after AI-powered quality control tools failed to perform as expected, and that Klarna began recruiting human customer service representatives after relying heavily on AI chatbots [12].

What to do with this

If your company has implemented AI, five questions can show which intent is in charge.

  • What is it measured against? Efficiency measures alone, such as time saved or license usage, are the warning sign; growth outcomes such as new revenue, new products or better customer experience go with stronger reported returns.
  • Did the workflow change, or just the tool? If the same steps happen in the same order with a chatbot added, it is a bolt-on.
  • Did headcount move before results did? A hiring freeze announced with the rollout, not after measured gains, is a bet on future gains.
  • Who owns the outcome? High performers back their deployments with leadership commitment and operational rigor.
  • Who is paying for it? If the AI budget grows while unrelated teams shrink, the cuts may be funding infrastructure rather than reflecting work AI now does.

My reading of the evidence is that announcements talk about transformation, most leaders describe AI's working job as efficiency, and the largest tech companies are also using cuts to fund infrastructure. The strongest returns reported so far come from the small group that rebuilt the work itself.

Limitations

Almost every figure here is self-reported by executives or employees in surveys, not measured financial results. The surveys use different samples, countries and definitions of return, so their numbers are not directly comparable. Several sources have a commercial interest in their findings: McKinsey, Gartner and GBK Collective sell advice, the Return on AI Institute is a partner of Scaled Agile, and SHL and Robert Half sell hiring and assessment services. Challenger counts the reasons employers give in layoff announcements, which is not proof that AI replaced anyone. The motives assigned to Meta and Microsoft are my interpretation of their public statements and reported spending; neither company has stated them.

How this was researched

Sources were opened and checked on November 10, 2026. The piece draws on 16 sources: survey reports and press releases from PwC, the Return on AI Institute, Wharton and GBK Collective, McKinsey, SHRM, NBER, Challenger, Robert Half and SHL, an article in Harvard Business Review, and news reporting from Reuters, Fortune and the Associated Press. Every figure was compared against text saved from its source and re-checked by a separate reviewer; figures that could not be matched were removed or narrowed.

Sources

Work with ECW