The revolution depends on choices humans have not yet made
AI development has accelerated dramatically since 2019, with transformers and generative models like ChatGPT exceeding expectations at an astonishing pace. Healthcare shows the most concrete progress, with AI improving cancer detection and drug discovery, though clinical implementation remains limited.
- Spanish AI expert Yolanda Gil predicted in October 2019 that AI would revolutionize daily behavior and improve health
- Google Health published research in 2020 showing AI improved breast cancer detection through mammography analysis
- Transformer models and generative AI tools like ChatGPT have accelerated development far beyond 2019 expectations
- AI systems remain largely confined to research settings despite proven capability in healthcare diagnostics
Spanish AI expert Yolanda Gil predicted in 2019 that AI would revolutionize daily behavior and improve health, but experts debate whether expectations have been met as rapid advances in generative AI reshape the landscape.
Four years ago, in the autumn of 2019, Spanish AI researcher Yolanda Gil stood before her peers and made a prediction. The advances coming in artificial intelligence, she said, would protect our environment, cure diseases, and fundamentally reshape how we live day to day. It was an optimistic read on a technology that most people still struggled to define. The debate among experts split cleanly then, as it does now: some saw existential danger ahead, others saw transformation of a benevolent kind. Gil landed firmly in the latter camp.
Today, that prediction reads differently. ChatGPT arrived. Midjourney followed. DALL-E, Bard, deepfakes, and a cascade of other tools have flooded into public consciousness with a speed that caught even seasoned technologists off guard. The question now is whether Gil was right—and if so, whether the revolution she foresaw is actually arriving, or merely beginning to cast its shadow.
When the first transformer models emerged in 2019, they promised to improve systems that already existed. The vision was modest: automate repetitive tasks, reduce human labor where possible, let machines handle the grunt work. AI lived mostly in the background then—in mortgage risk assessments, in recommendation algorithms that shaped what you saw on social media, in systems designed to predict whether a released prisoner might reoffend. It was, as one expert put it, a hidden intelligence. That same year, Google unveiled BERT and Duplex. OpenAI released ChatGPT-2. Microsoft and Nvidia announced healthcare tools for medical diagnosis. Tesla pushed forward with autonomous driving. The pieces were moving into place, but the public barely noticed.
The acceleration since then has been staggering. What existed in 2019 as scattered, primitive attempts at image generation and text creation has exploded into something almost unrecognizable. The tools arriving now don't merely improve existing processes—they generate entirely new content, write code, compose music, produce video. Experts describe it as a paradigm shift, a moment when the pace of development outstripped even the most bullish forecasts. Yet some caution remains. We are still far from the kind of artificial intelligence that Stanley Kubrick imagined for the year 2001—the thinking machine that rebels against its creator. Machines have passed the Turing Test since the 1950s, but the truly spectacular breakthroughs have been sparse. Progress, measured against the hype, may not be as swift as it feels.
In healthcare, the most concrete gains have materialized. AI now helps detect diseases earlier, design new molecules for potential medicines, and analyze medical images with precision that can catch what the human eye might miss. In 2020, Google Health published research showing an algorithm that significantly improved breast cancer detection by analyzing mammography images alone. Yet these systems remain largely confined to research settings. They have not yet woven themselves into clinical practice at scale, partly because implementation requires careful oversight, partly because institutions move slowly. The technology exists to save lives; the infrastructure to deploy it does not yet fully exist.
Beyond medicine, the applications are multiplying. In education, AI personalizes learning and identifies where students struggle. In agriculture, researchers envision robots that diagnose plant disease and harvest ripe fruit without bruising it. Manufacturing could optimize production lines further, cutting waste and improving efficiency. Autonomous vehicles could communicate with one another to smooth traffic and prevent collisions. Finance could detect fraud faster. Entertainment could deliver recommendations tailored to individual taste, or generate music and art on demand. The list grows longer each month.
But the revolution Gil predicted depends on choices humans have not yet made. Some experts worry that AI could weaponize misinformation at scale, flooding the information landscape with falsehoods. Others point to the risk of mass job displacement without adequate social planning. There is broad agreement among researchers that regulation matters—that the technology's impact on truth, employment, and social stability cannot be left to market forces alone. The revolution, in other words, is not predetermined. It will be shaped by how we choose to build it, govern it, and deploy it. For now, AI is arriving so quietly, so subtly woven into the tools we already use, that most people barely notice it is there at all.
Notable Quotes
We are still far from the kind of artificial intelligence that Kubrick imagined for 2001— AI experts quoted in analysis
The revolution of daily human behavior depends on humans—we can use it to deceive or to create a better society— Juan Ignacio Rouyet, President of We The Humans think tank
The Hearth Conversation Another angle on the story
When Yolanda Gil made her prediction in 2019, did she have any sense of how fast things would actually move?
Almost certainly not. Nobody did. The consensus then was that these tools would take years to reach ordinary people. The infrastructure didn't exist. The computing power was expensive. And then suddenly, in the span of a few months, ChatGPT arrived and changed the entire conversation.
So the speed was the surprise, not the direction?
Exactly. The direction—that AI would become more capable, more useful, more integrated into daily life—that part was predictable. What nobody anticipated was that it would happen so fast that we'd barely have time to think about the implications before the next breakthrough arrived.
In healthcare, you mentioned that AI can detect breast cancer better than humans in some cases. Why hasn't that been deployed everywhere?
Because hospitals and clinics move slowly. There's liability to consider, regulatory approval to obtain, staff retraining required. And honestly, there's still skepticism. Doctors want to understand how the algorithm works before they trust it with patient care. It's not enough for a tool to be accurate—it has to be trustworthy, explainable, integrated into existing workflows.
That sounds like a gap between what's technically possible and what's actually happening.
It is. And that gap might be the most important story right now. We have tools that could improve lives, but the institutions that would deploy them aren't moving fast enough. Or maybe they're moving at exactly the right speed, and the technology is just moving faster than it should.
What worries the experts most?
Misinformation, mostly. If AI can generate convincing text and images at scale, what happens when bad actors use it to flood the information space with lies? And employment—if machines can do more kinds of work, what happens to the people whose jobs disappear? Those aren't technical problems. They're human problems. And they require human solutions.