The Productivity Panic
We are living through the underlying fears of job security, and one example of this global situation is decision-makers changing their minds when switching web design agencies, for instance, only to decide to drop one because their marketing team has started using coding agents to update the…

We are living through the underlying fears of job security, and one example of this global situation is decision-makers changing their minds when it comes to switching web design agencies, for instance, but deciding to drop one because their marketing team has started using coding agents to update the company website on its own.
This signal shows us that it is not only senior management that, armed with AI agents, is stepping out of its traditional lanes. One example was reported by Issie Lapowsky in Bloomberg, writing about Intuit, where she found that product managers and designers are being encouraged to code their own features for QuickBooks, the company's flagship software.
Intuit has not yet released those prototypes to customers, but they are building the conversational bridge between the product manager and the engineer, resolving the eternal dilemma: “"I want something like this"”.
Lapowsky's article documents how AI coding agents are generating a productivity panic among executives and engineers, but the sharpest finding comes from UC Berkeley's research: employees who used AI tools increased both the amount of work they could complete and the variety of tasks they could take on, even when they were not required to adopt the technology.
The paradox is brutal: as employee productivity rose, so did the amount of work they took on, partly because AI made it easy to start new tasks.
A quote from one worker interviewed captures it all: you thought that being more productive with AI would let you work less, but in reality you don't work less. Breaks shrank — workers picked up the habit of sending one last prompt before lunch or after working hours.
Meanwhile, OpenAI president Greg Brockman said that every moment your agents aren't running feels like a wasted opportunity. That phrase is the perfect distillation of what we could call cognitive extractivism: the pressure is no longer only on the human who produces, but on the human who is not delegating enough to the machines.
That mindset — every second without agents running is a wasted second — is not new. It is the same mechanism economics identified more than a century ago, and it now returns with a chat interface.
The Digital Jevons Paradox
Researchers at UC Berkeley and Harvard Business Review warn that the initial «productivity boom» can result in sloppy work and burnout — and as frustration builds, workers may quit.
The same source draws the exact historical parallel: a more efficient steam engine did not reduce coal consumption — it increased it. The invention of the mechanical loom multiplied textile productivity — workers did not work less; they worked harder in factories that could now demand much higher production quotas.
It is the Jevons Paradox applied to software: more efficiency does not produce more free time, but more demand for output.
What this means for us in the Caribbean
This is where the analytical frameworks I have developed over two decades take on new urgency:
- The invisibles under crossfire. If Bloomberg reports a 16% drop in job postings for young software workers, our Dominican professionals who work remotely for foreign markets are on the front line of this adjustment. They have no local labor protections, they do not take part in the national productive fabric, and they now face the added pressure of competing not only against other humans but against autonomous agents. The domestic techno-colonialism I denounced is intensifying: the same machinery that promised to democratize access is concentrating downward pressure.
- The Pokémon effect turns into a crisis. Product managers started writing code. Designers took on engineering tasks. What Silicon Valley describes as «empowerment» is exactly what I have warned about: the illusion that collecting superficial capabilities is equivalent to real mastery. When everyone can generate code but debugging AI-generated code takes 3 times longer and governance frameworks are almost nonexistent, quality becomes the first casualty. In a labor market like the Dominican one, where there is already a gap between certifications and real competence, this deepens the problem.
- Accelerated Corporate amnesia. AI reduces cognitive friction in areas such as remembering syntax, searching documentation and generating scaffolding, creating a strong perception of speed. But delivery metrics such as lead time, defect rate and deployment frequency often remain unchanged because the bottlenecks migrate to the review and validation stages. In other words: speed is illusory if there is no institutional capacity to absorb what has been produced. This is, in essence, the Corporate amnesia I documented since 2018 — now amplified by agents that produce more than organizations can process, understand and retain.
- Symbolic industrialization in a new disguise. The country's decline in global innovation indices — which I have already documented — does not stop while the consumption of imported AI solutions grows at a scale thirty times larger than investment in local development. The productivity panic that Bloomberg narrates in producer economies translates, in consumer economies, into a double vulnerability: we depend on tools whose evolution we do not control, and we absorb the labor pressures without the benefits of the value chain. It is symbolic industrialization with a chat interface.
The uncomfortable question
Current economic frameworks are ill-equipped to handle the scale and speed of the transformation under way. If this is true for the U.S. and Europe, it is exponentially more true for the Caribbean.
In its publication, Bloomberg documents the panic. But panic is a luxury of those who produce the tools. For those of us who consume them without taking part in their design, the risk is not panic — it is the naturalization of dependence.
The real question is not whether AI agents will make us more productive, but: productive for whom, under what terms, and with what bargaining power?
Between centers of excellence with NVIDIA, rare earths that will transform the economy, a spaceport for 2028, the expansion of digital education and the calls for responsible AI, the country is piling up announcements that place it in the dance of the technological future. But without clear answers to those questions — productive for whom, under what terms — those announcements risk becoming narrative. The answers can help establish cross-cutting capabilities that contribute to human development, not just to the next day's headline.
I reiterate what I have said before about the augmented intelligence framework: AI should amplify human talent, not replace understanding. Productivity without cognitive sovereignty is simply speed in the service of others.