plan for the dip: why your AI investment hasn't paid off yet.
4 mistakes that deepen the dip that stands in the way of AI gains and how to design your way through it
Is your AI investment stuck? Learn about the productivity J-curve, the essential dip that happens when adopting generative AI tools. This article explains why organizations expect steady AI gains but instead hit a learning curve and a temporary decline in workforce productivity. Discover the four common mistakes that deepen this dip and get actionable steps for intentional work design and workflow redesign to shorten the transition. Find out how successfully navigating this curve prepares you for future growth and addresses looming talent scarcity.
Something doesn't add up. Ask individuals whether AI has made them more productive, and most will say “yes.” Ask organizations whether they've seen a measurable uptick in productivity, and the answer is far less convincing. Across the global economy, the massive gains that generative AI promises haven't materialized — at least, not yet.
This isn't a failure; it's a pattern. And understanding it starts with confronting a mistaken assumption.
what most leaders expect
When organizations invest in AI, they expect a relatively smooth upward trajectory. New tools are introduced, people start using them, and productivity steadily climbs. The adjustment period is assumed to be short. Returns are expected to follow investment quickly.
This is the picture most leadership teams have in mind.
It's a reasonable assumption, but it's wrong.
what actually happens
Every transformative technology follows a different arc. Productivity doesn't climb steadily from the moment of adoption — it dips first. Organizations invest in new tools, workflows slow as people learn, old processes break before new ones are built, and for a period, output actually declines. Only after the learning phase does productivity rise above its original baseline, often significantly.
Economists call this the productivity J-curve. And it explains the disconnect we're seeing right now.
Today’s employers are in the installation phase of generative AI. The tools are deployed, the licenses are purchased, the pilots are running, but the productivity gains haven't followed. They can't — not yet — because the dip comes first. The decisions organizations make during this period of disruption will determine whether they emerge stronger or weaker on the other side of this transformation.
4 mistakes that deepen the productivity dip
- Expecting immediate returns: Too many organizations have introduced AI tools without planning for the learning curve. They expect productivity to rise from day one and grow frustrated when it doesn't. The result is premature disappointment — and premature decisions about what's working and what isn't.
- Cutting too early: A significant number of current layoffs are being "AI washed" — attributed to automation when the actual drivers are broader business conditions. Organizations are reducing headcount before they've achieved any real ROI from AI. This is the equivalent of dismantling the team during the dip, before the gains arrive.
- Organizational impatience: There is a gap between what C-suite leaders expect and the realistic timeline for AI productivity gains. Customer service functions may see returns sooner, but for most areas of the business, meaningful productivity growth is still one to two years away. Most organizations are not having honest conversations with their teams about this timeline.
- Technology without workflow redesign: Dropping AI tools into existing processes doesn't work; the process itself needs to change. Success requires identifying which steps can be handled by AI, which need human-AI collaboration, and which must remain entirely human. Organizations that focus on technology adoption without investing in workflow redesign will be, quite simply, stuck in the dip longer than they need to be.
the confidence gap
Randstad's 2026 Workmonitor research reveals a striking disconnect: 95% of employers believe their business is on track for growth in the next year, but only 51% of talent share that confidence. That gap matters — not just as a sentiment indicator, but as a real barrier to navigating the J-curve.
Getting through the dip requires people to experiment, adapt and tolerate temporary discomfort. That doesn't happen when the workforce is anxious, and leadership is projecting confidence without acknowledging the reality of the transition. Psychological safety — creating the conditions where people can flag what's not working without fear — is where value is either created or eroded during this phase.
Organizations that rush AI implementation without bringing their people along are not accelerating through the dip. They're deepening it.
what's on the other side
Here's what the job-replacement narrative gets wrong: The J-curve doesn't end with fewer people — it ends with more demand for them.
When AI productivity gains do materialize, they create capacity. Capacity enables growth, and growth requires people. Organizations that successfully increase workforce productivity through AI won't be looking to shrink their teams; they'll have the incentive and the means to expand them. The return on investment from AI doesn't replace the need for talent; it amplifies it.
Now, layer in demographics. Across developed economies, the working-age population is shrinking. The moment organizations emerge from the dip and need to hire, they'll be doing so in a labor market with structurally fewer people available. Talent scarcity — which many assume AI will solve — is poised to return with a vengeance. The organizations cutting headcount today without a long-term talent strategy may be walking toward a wall they can't yet see.
design your way through
The good news: The J-curve is not fixed. Organizations can't eliminate the dip altogether, but they can make it shallower and shorter.
That minimized curve is what intentional work design looks like. It means redesigning workflows rather than just deploying technology. Providing psychological safety so people are unafraid to flag challenges. Setting realistic expectations with teams. And positioning AI as a way to create capacity and enable work — not purely as a cost-cutting measure.
The organizations that emerge from the dip first will be hiring in a market that hasn't fully tightened yet, while the ones that wait will be competing for scarce talent at a premium.
For CHROs, this is the moment. You’re not just managing the adoption of another tool, but architecting the transition. You will need to align work, people and technology so the organization comes out of the dip ready to grow.
Your company’s J-curve is a temporary part of its transition. What you build during that dip is not.