We are still in prehistory with artificial intelligence
From within Brazil's most respected research institution, a voice of measured academic authority has offered a quietly radical reframing: artificial intelligence, for all its recent spectacle, remains in its prehistory. The University of São Paulo professor's choice of word is precise — prehistory does not mean early, it means before the tools existed to understand what was happening. In placing this moment before the beginning of the real story, the assessment invites not alarm, but a deeper humility about what we think we know of what we are building.
- A USP professor's single word — 'prehistoric' — cuts against months of breathless AI headlines, suggesting the field lacks even the frameworks to comprehend what it is creating.
- The tension is unsettling: if today's chatbots, image generators, and coding assistants are merely the opening act, the truly disruptive transformations have not yet arrived.
- Regulatory debates, safety concerns, and labor displacement fears are dominating public discourse — yet they may be urgent questions aimed at a technology still finding its shape.
- The academic perspective, grounded in both theory and application, navigates between genuine breakthrough and hype, pointing toward a longer arc that current predictions consistently fail to map.
- The field is landing not at a summit but at a foundation — one whose structure, and the civilization it might support, remains largely unwritten.
A professor of artificial intelligence at the University of São Paulo has offered a stark and deliberate assessment: we are still in prehistory. The characterization cuts against the breathless coverage that surrounds every new AI model or capability. This is not a metaphor for "early." Prehistory describes a time before the tools existed to document or comprehend what was happening — applied to AI, it suggests we lack the frameworks to fully understand what we are building or where it leads.
The weight of the observation comes from its source. USP is Brazil's leading research institution, and its faculty occupy the intersection of theory and application. They see both genuine breakthroughs and the hype surrounding them. They understand the gap between what AI can do today and what it may be capable of in a generation.
The implication is quietly unsettling. If we are truly in prehistory, the major developments have not yet arrived. The current wave of applications may be less like the invention of the internet and more like the printing press — significant, but not yet the full story of what becomes possible when the tool matures.
This framing also calls for humility. AI's trajectory has resisted reliable prediction, cycling through optimism and stagnation. What looks like a plateau may be a foundation being laid. What appears as exponential progress may be the easy gains before harder problems emerge.
Meanwhile, public discourse is consumed by regulation, safety, and labor displacement — urgent questions, but perhaps questions asked about a technology still finding its shape. The real challenges, the ones that will define how AI integrates into human life, may be waiting in chapters not yet written.
A professor of artificial intelligence at the University of São Paulo has offered a stark assessment of where the field stands: we are still in prehistory. The characterization, coming from one of Brazil's most respected research institutions, cuts against the breathless coverage that surrounds AI breakthroughs. Every few months brings headlines about new capabilities, new models, new applications. Yet this academic perspective suggests something more fundamental—that what we are witnessing now is merely the opening chapter of a much longer story.
The professor's framing is deliberate. Prehistory is not a metaphor for "early." It is a statement about the absence of recorded history, about a time before we had developed the tools to even document what was happening. Applied to artificial intelligence, it suggests that the current moment lacks the frameworks, the understanding, the maturity to fully comprehend what is being built or where it is heading. We are in the phase of discovery without yet having discovered what we are discovering.
This assessment carries weight because it comes from within the research ecosystem itself, not from venture capitalists or technology evangelists. The University of São Paulo is Brazil's leading research institution, and its faculty work at the intersection of theory and application. They see both the genuine breakthroughs and the hype. They understand the gap between what AI can do today and what it will be capable of doing in five years, in ten years, in a generation.
The implication is unsettling in its own way. If we are truly in prehistory, then the major developments have not yet arrived. The transformations that will reshape how we work, learn, create, and think are still ahead. The current wave of AI applications—the chatbots, the image generators, the coding assistants—may be less like the invention of the internet and more like the invention of the printing press: a significant tool, but not yet the full story of what becomes possible when that tool matures.
This perspective also suggests humility is warranted. Predictions about AI's trajectory have been notoriously unreliable. The field has experienced multiple cycles of optimism followed by periods of stagnation. What looks like a plateau from one angle may be a foundation being laid for the next leap. Conversely, what appears to be exponential progress might be the easy gains before the harder problems emerge.
The professor's comment arrives at a moment when AI regulation, safety concerns, and questions about labor displacement are dominating public discourse. These are urgent questions. But they may be questions asked about a technology that is still finding its shape. The real challenges—the ones that will define how AI integrates into human life—may not yet be visible. They may be waiting in the chapters we have not yet written.
Notable Quotes
We are still in prehistory— University of São Paulo AI professor
The Hearth Conversation Another angle on the story
When you say we're in prehistory with AI, do you mean we don't understand what we've built yet?
Partly that. But more that we lack the historical distance to see what's actually significant. We're too close to it.
So the breakthroughs we celebrate now—are they real, or are we just not seeing the bigger picture?
They're real. But they might be like the first stone tools. Genuinely useful, genuinely new. But not yet the thing that changes everything.
That's unsettling. It suggests the hard problems are still coming.
Yes. Right now we're solving problems we know how to solve. The unknown unknowns are still ahead.
What would it look like when we leave prehistory?
Probably when we can predict and control what these systems do with real precision. When we understand not just that they work, but why they work and what they'll do next.