The conversation around AI often drifts toward a single, unsettling conclusion: machines are coming for our jobs. But as with most technological shifts in history, this narrative is far more complex than it appears.
Mark Cuban, known for being one of the sharks at Shark Tank is also a businessman. He believes that AI won’t eliminate work. Instead, the technology will create more of it. And not just any work, but jobs deeply rooted in the uniquely human ability to sense, interpret, and navigate the living, breathing world.
According to Cuban, when humans wake up, they open their eyes and instantly, they're collecting information.
This isn’t just passive observation; it’s rapid, real-time interpretation that no artificial system can truly match. This innate data capture and processing ability is what separates biological intelligence from machines.
One particularly compelling example involves Cuban describing a dog. Despite having no access to software updates, LIDAR, or GPS, a dog can cross the street safely. If a variable changes—say, a car suddenly swerves or a cyclist appears—there’s no crash, no adversarial glitch.
The dog adapts instantly, processing new data without error.
That’s the magic of organic intelligence. And it's why AI, for all its power, still struggles to interact with the world the way even a dog can.
Yes, Large Language Models (LLMs) like GPT-4, Claude, or Gemini are marvels of machine learning.
They reason fast, distill logic efficiently, and can generate remarkable results from massive amounts of text-based training data. But their learning is limited. Most don't learn continuously from new, real-time experience—they learn through structured training cycles, often using a process called distillation.
And herein lies a challenge.
According to Cuban, text alone can’t fully capture the real world. LLMs don’t see. They don’t feel. They interpret symbols but lack first-person understanding. Even as these models expand into multimodal systems capable of analyzing images and audio, they’re still interpreting a filtered, flattened version of the world.
In contrast, people (and animals) live within that world. They absorb information from their environment holistically, using instinct, emotion, and memory to make decisions. For AI to catch up—even partially—will take time. Maybe decades.
For the creators of AIs, they know this, and they one big challenge to overcome this: data access.
And according to Cuban, intellectual property (IP), is king.
"Companies are realizing that their IP is their differentiation and salvation."
This should be an issue to tech companies willing to advance LLMs because as the value of proprietary knowledge becomes clearer, companies are beginning to restrict it.
This "data Balkanization" means general-purpose AI will increasingly be limited in what it can learn—unless it has access to scale and human feedback.
That’s where user engagement becomes vital.
Platforms with hundreds of millions of users, like OpenAI or Google, have a significant advantage. Not just because they host the best models, but because they learn from every interaction. The more people talk to them, correct them, and challenge them, the more these systems evolve.
This is why humans remain essential to the loop.
More jobs. Here is why. The minute you open your eyes, you are capturing information from everything you see. That makes humans and animals unique. My Australian Sheppard can cross the street safely. And if something changes , we don't have to worry about it creating an… https://t.co/gVHY5qCHQv
— Mark Cuban (@mcuban) April 2, 2025
People are still needed to convey the visual, tactile, and experiential world to machines. No matter how smart a model becomes, it will rely on people to teach it, steer it, and interpret the gray areas that data alone can’t define.
While automation will eliminate some jobs. But it will also create entirely new industries—ones centered around curating, contextualizing, training, maintaining, and regulating AI systems.
Think about prompt engineers, AI safety reviewers, model trainers, synthetic media designers, digital ethicists, and more.
Add to that the operational staff, creatives, and domain experts who will be needed to use AI tools responsibly and effectively.
And there’s another truth: companies will still need humans to run the show. To build products. To market them. To talk to customers. To innovate. To sell. And to lead.