90% of businesses report zero AI productivity gains, here's why
TL;DR
A 6,000-executive NBER study across the US, UK, Germany, and Australia found that nearly 90% of firms have seen zero measurable AI impact on employment or productivity over the past three years. Despite two-thirds of executives claiming they use AI, average actual usage amounts to just 1.5 hours per week. BCG's research identifies "AI brain fry", cognitive overload from managing too many disconnected tools, as a primary reason more AI is leading to less productivity. The 10% of businesses winning with AI treat it as a fundamental business transformation, not a software purchase.
Is the AI productivity revolution actually happening?
The headline from a major new National Bureau of Economic Research (NBER) study is blunt: nearly 90% of firms surveyed reported that AI has had zero measurable impact on employment or productivity over the past three years. The study surveyed 6,000 CEOs and executives across the US, the UK, Germany, and Australia.
Despite two-thirds of those executives claiming they use AI, average actual usage amounts to just 1.5 hours per week.
"You can see the computer age everywhere but in the productivity statistics.", Nobel laureate Robert Solow, 1987
Apollo chief economist Torsten Slok invoked that exact quote to describe what is happening with AI today. Outside of the massive tech giants building the models, he argues there are no signs of AI in profit margins or earnings expectations.
What is "AI brain fry" and why is it destroying your team's output?
Boston Consulting Group (BCG) recently identified a phenomenon they are calling "AI brain fry." When workers use four or more AI tools simultaneously, productivity does not just stall, it actively plummets. Workers report mental fog, headaches, slower decision-making, and a crowded sense of thought.
The reason is counterintuitive. Oversight of AI, checking its work, refining its outputs, validating its accuracy, is actually increasing mental effort rather than reducing it.
If AI enables someone to complete 20 tasks in a day instead of two, the natural temptation is to load 20 more tasks onto the plate. But managing, reviewing, and correcting 40 tasks instead of 20 doubles the cognitive load. Constant context-switching, second-guessing outputs, and fixing mistakes is not making teams superhuman. It is burning them out.
The technology designed to free people from drudgery is creating an entirely new form of it. Your best people are spending their days as AI babysitters, proofreading chatbot outputs and correcting hallucinated data instead of doing the creative, strategic, high-value work you actually hired them for.
ManpowerGroup data confirms the pattern:
- AI usage has increased 13% over the past year
- Worker confidence in AI has fallen 18% over the same period
People are using it more, trusting it less, and exhausting themselves in the process.
How is Australia actually performing on AI productivity?
KPMG reports that while Australia is a global leader in responsible AI governance, it is lagging severely in productivity gains. Only 25% of Australian organisations are successfully converting AI experiments into measurable business value. The rest are stuck in what is being called "pilot purgatory", generating outputs that require so much human intervention to fix they barely qualify as automation.
What does the Singapore data reveal?
The picture in Singapore is equally stark:
- 64% of financial institutions are deploying AI in live operations
- Only 3% have achieved what is considered "AI leadership status"
- 60% of regional respondents see less than a 5% impact on EBIT
- 18% report absolutely no financial impact at all
The barrier is not the technology. It is internal resistance to changing legacy infrastructure, and the proliferation of "Franken-Stacks", fragmented, disconnected AI tools that are impossible to govern and exhausting to use.
See where AI fits in your business. Free.
A 45-minute audit. We map the highest-value automations and what they're worth in time and money. No pitch, no pressure.
What is the UK doing with its £500 million AI investment?
The UK government has committed £500 million through its Sovereign AI Fund and speaks ambitiously about making Britain an AI superpower. The reality on the ground is that most British businesses face the same fundamental challenges as their counterparts in Sydney and Singapore: buying tools without strategy, deploying agents without governance, and measuring success by the number of AI subscriptions held rather than what those subscriptions actually deliver.
What separates the 10% of businesses actually winning with AI?
A major PwC study identified a stark divide. The top 20% of companies are capturing 74% of all AI-driven economic returns. What sets them apart:
- 2.6x more likely to use AI to reinvent their entire business model, not just tweak existing processes
- 2.5x more revenue invested in AI initiatives
- 80% more likely to systematically track the business impact of those investments
- Up to 4x more invested in data and analytics foundations (Gartner)
They are not buying ChatGPT licences and hoping for the best. They understand that if underlying data is disorganised, AI simply generates disorganised results faster. They focus on one or two high-impact use cases, provide targeted training, and establish clear governance before expanding.
What is the J-curve, and where are most businesses sitting on it?
Economists call it the J-curve. When a major new technology is introduced, there is an initial productivity slowdown as companies invest time and money learning how to use it properly. The line on the graph dips before it surges exponentially upward.
The 1980s IT revolution is the most instructive parallel. Computers were widespread through the 1970s and 1980s, yet productivity growth actually slowed during that period. It was not until the mid-1990s, once companies had finally reorganised their workflows around the technology, that the productivity surge arrived. The firms that invested in training, process redesign, and data infrastructure captured the enormous gains of the internet era. The ones that just bought computers and hoped for the best were left behind.
Right now, 90% of businesses are stuck at the bottom of the J-curve. They are absorbing the friction, cost, and cognitive overload without yet reaping the exponential rewards. The question is not whether AI will deliver on its promises, it will. The question is whether your business will be positioned to capture those gains when the curve turns upward.
What to do this week
1. Audit your AI tools. Count exactly how many AI tools your team is actively using and what for. If staff are juggling more than three or four disconnected applications, consolidate. Pick one or two that integrate seamlessly into existing workflows and retire the rest.
2. Assess your data foundations. You cannot build on disorganised data. If your internal information is siloed, inconsistent, or inaccurate, any AI initiative will fail regardless of the tool. Prioritise cleaning and structuring your data before deploying advanced AI agents.
3. Measure actual impact, not vague efficiency. If an AI tool is supposed to save your team ten hours a week, track exactly where those ten hours go. Are they redirected to higher-value work, or consumed by fixing AI errors and managing cognitive overload? If you cannot answer that question, you do not yet know whether the tool is working.
4. Put humans first. Ask your team what problems they actually need solved before imposing a top-down AI mandate. Create safe spaces for experimentation. Celebrate the small wins, ten minutes saved on a report, one fewer meeting per week, because those compound into measurable gains over time. The businesses that fail are the ones that announce an "AI-first strategy" by company-wide email, hand out tool subscriptions, and wonder why nobody is using them six months later.
Where to from here
Book a free 60-minute AI audit, we'll explore exactly what workflows are worth augmenting with AI.
Live with passion & AI,
Brett
Running an event? Put practical AI on your stage.
Keynotes and workshops that send business owners home with a plan they can use Monday morning. No hype.
Frequently asked questions
Why are 90% of businesses seeing zero AI productivity gains?
+
According to a 6,000-executive NBER study, most firms are using AI for an average of just 1.5 hours per week and deploying tools without the data infrastructure, training, or governance needed to extract real value. They are treating AI like a light switch rather than a fundamental business transformation.
What is AI brain fry and how does it destroy productivity?
+
BCG identified AI brain fry as the productivity decline that occurs when workers use four or more AI tools simultaneously. Workers report mental fog, slower decision-making, and cognitive overload from constantly checking, correcting, and managing AI outputs, which increases mental effort rather than reducing it.
What are the top companies doing differently with AI?
+
PwC data shows the top 20% of companies capture 74% of all AI-driven economic returns. They are 2.6 times more likely to use AI to reinvent their entire business model, invest 2.5 times more revenue into AI initiatives, and are 80% more likely to systematically track business impact.
How much more do successful AI companies invest in data foundations?
+
Gartner research found that organisations with successful AI initiatives invest up to four times more in data and analytics foundations than those that fail. Clean, structured data is a prerequisite, not an afterthought.
What is the J-curve in AI adoption?
+
The J-curve describes the initial productivity dip businesses experience when adopting major new technology before the exponential gains materialise. The 1980s IT revolution followed the same pattern, productivity slowed while computers spread everywhere, then surged in the mid-1990s once companies redesigned their workflows around the technology.
How is Australia performing on AI productivity?
+
KPMG reports that while Australia leads globally on responsible AI governance, it lags severely on productivity gains. Only 25% of Australian organisations are successfully converting AI experiments into measurable business value, the rest are stuck in pilot purgatory.
What percentage of Singaporean financial institutions have achieved AI leadership?
+
A study found that while 64% of Singaporean financial institutions are deploying AI in live operations, only 3% have achieved AI leadership status. Sixty percent of regional respondents see less than a 5% EBIT impact, and 18% report no financial impact at all.
Brett is a four-time founder (Darra Tyres, Gladfish, EzyTrac, Anaboo) and the operator behind AIOS, Anaboo's AI Operating System. He writes from inside the build, installing AI in his own businesses first and reporting back what actually moves the numbers. Based between Singapore, the UK and Australia.
