Schottky Diode,Schottky Barrier Diode,Schottky Diode Forward Voltage,Sic Schottky Diode Shenzhen Kaixuanye Technology Co., Ltd. , https://www.icoilne.com
AI philosopher: artificial intelligence should first study cells instead of brains
[Netease Smart News, August 29] From the very beginning, we’ve always been told that intelligence and the brain are inseparable. Intelligence is often used as a synonym for cognition, and discussions about talent and wisdom frequently use the brain as a metaphor. Naturally, when technology evolved to the point where humans began attempting to replicate human intelligence in machines, our goal became to simulate the brain in artificial intelligence.
But what if this approach is flawed? What if all the talk about creating a "neural network" and a robotic brain is a misguided effort? Perhaps we’ve overlooked something simpler yet more profound: smaller cells.
This counterintuitive perspective is the work of Ben Medlock, who isn’t your typical academic. As the founder of SwiftKey, a company that uses machine learning algorithms to design smartphone keyboard apps, his day-to-day work revolves around enhancing everyday tools with artificial intelligence.
But Medlock is more of an AI philosopher. His vision isn't about shaving seconds off texting times. Instead, he aims to shift the paradigm of AI research and redefine intelligence itself.
"I lead a double life," Medlock said. "On one hand, I collaborate with SwiftKey to make AI practical. That's my day job. But I also spend a lot of time pondering the philosophical implications of AI and the very human values tied to intelligence."
This line of thought led him back to the foundation of human life—cells.
"In fact, I think we should start with eukaryotic cells," Medlock argued. "Rather than treating AI as an artificial brain, it’s better to envision the human body as an 'incredible machine.'"
Typically, AI scientists favor the brain as a model for intelligence. Hence, some machine learning methods are described as "neural networks." These systems mimic neuron connections and neural structures, but they don’t fully replicate the complexity of the human brain.
Medlock, however, challenges this metaphor. "There's a gap between current AI capabilities and human intelligence," he stated.
Currently, AI researchers tackle intelligence by breaking it into smaller tasks, training machines to perform specific actions step-by-step. The more these machines learn to recognize patterns and execute tasks, the more "intelligent" they seem.
Yet Medlock insists this isn’t how humans operate. "If you study human intelligence, it’s a mistake to start with the brain," he explained. "Cells are like microscopic information processors. They’re flexible, communicating in networks, working together."
Medlock deepened this idea by referencing DNA replication. Geneticists once believed evolution relied on random mutations, but new studies show cells are remarkably precise during DNA copying—only one error per four billion bases. This precision introduces a corrective mechanism, allowing cells to adapt positively to their environment.
"Intelligence isn’t about playing chess," Medlock emphasized. "It’s about processing environmental data and acting accordingly. Cells are the origin of all organic intelligence—they’re data processors."
Organic intelligence offers a concrete model for understanding consciousness. "When conflicting data enters, it’s compared against predictions built from past experiences," Medlock noted.
He suggested that if our aim is to create machines as adaptable and intelligent as humans, we need to build AI systems with similar concrete models. This would give machines the same power and flexibility as humans.
Of course, this raises the larger question: Is this what we truly want from AI? Medlock admitted that if we prefer AI to focus narrowly on specific tasks, we can continue down the current path. But he believes this approach may limit creativity and innovation.
The brain model is useful for developing AI that excels in particular areas, but it restricts growth beyond those boundaries. "We're hitting limits with current methods like deep learning and neural networks," Medlock said. "I don't think we need to simulate organic intelligence's evolutionary process, but it’s an intriguing question."
Medlock doesn’t have all the answers. He proposes studying AI from a cellular rather than a brain-based perspective. While his idea remains abstract, he envisions future machines as independent entities with their own experiences, memories, and decision-making processes—much like humans.
Practically, Medlock suggests these machines should be:
1. Independent information processors capable of analyzing physical data without constant reliance on external servers.
2. Mobile, with the ability to interact with the world and adapt physically.
3. Self-aware, understanding themselves and their place in the universe.
While this vision sounds futuristic, it's not entirely far-fetched. Consider autonomous cars: today’s models rely on sensors for basic navigation, but imagine a vehicle covered in nanomaterials that can detect touch, temperature, and movement, then respond accordingly. Such machines could do far more than transport people.
Ethical concerns remain a significant hurdle. Redefining AI metaphors could shift development toward creating emotional, sentient beings. Medlock acknowledges potential risks but emphasizes the importance of discussing regulations, laws, and rights as AI progresses.
In conclusion, while Medlock’s ideas aren’t yet realized, they spark important conversations. His vision challenges the status quo and inspires exploration into uncharted territories of AI. Time will tell how these concepts evolve, but Medlock has plenty of opportunities to refine his ideas and influence the future trajectory of AI.