What Does the AI Boom Really Mean for Humanity?

In partnership with

The Gorilla Problem: A Metaphor for AI Risks

In Central London, gorillas reside behind glass at a zoo, offering a glimpse into our past and a vision of our future. This scenario serves as a metaphor known as the "Gorilla Problem" among AI researchers. It highlights the risks of creating machines that surpass human intelligence—a concept known as superhuman AI. The fear is that such AI could potentially take over the world, threatening our very existence. Despite these warnings, major companies like Meta, Google, and OpenAI continue to push forward, striving to develop AI that can outperform humans in every domain. They argue that superintelligent AI will solve our most challenging problems and invent technologies beyond our current imagination.

All your news. None of the bias.

Be the smartest person in the room by reading 1440! Dive into 1440, where 3.5 million readers find their daily, fact-based news fix. We navigate through 100+ sources to deliver a comprehensive roundup from every corner of the internet – politics, global events, business, and culture, all in a quick, 5-minute newsletter. It's completely free and devoid of bias or political influence, ensuring you get the facts straight.

Subscribe to 1440 today.

Current AI Applications: Narrow Intelligence in Action

Today, AI is ubiquitous, far beyond just photo editing and chatbots. It plays crucial roles in preventing tax evasion, diagnosing cancer, and tailoring advertisements, among countless other applications. These AI tools exemplify narrow artificial intelligence—sophisticated algorithms excel at specific tasks. However, the pursuit of artificial general intelligence (AGI), which aims to create machines that outperform humans across all areas, remains the Holy Grail of AI research. Tech giants are investing billions annually to achieve this, aiming to replicate the broad, capable, human-like intelligence that we often take for granted.

Challenges in Defining and Replicating Intelligence

Defining intelligence itself is a complex and slippery concept. Early definitions, such as Vivian Henman’s 1921 description of intelligence as the capacity for knowledge, are broad and challenging to apply practically. Others suggest intelligence is the ability to solve hard problems, but this requires clear criteria for what constitutes a "hard" problem. Despite the lack of a universal definition, certain characteristics are sought in truly intelligent AI: the ability to learn and adapt, reason with a conceptual understanding of the world, and interact with its environment to achieve goals.

The Role of Physical Interaction in Achieving Superintelligence

Professor Hannah Fry, a mathematician and writer, questions whether superintelligent AI is just a few years away and whether advanced AI could pose an existential threat akin to the near extinction of gorillas due to human impact. Researchers like Sergey Levan and his PhD student Kevin Black argue that AI might achieve superintelligence only when it can physically interact with the world. Their work with robots that learn actions autonomously suggests that embodiment—having a physical form—could be crucial for developing truly intelligent systems.

Expert Opinions: Balancing Optimism and Caution

Not everyone shares the doomsday perspective. Melanie Mitchell, an AI researcher, believes that while AI poses many threats, labeling it as an existential threat is an overstatement. She emphasizes the importance of addressing current issues such as AI bias, where facial recognition systems make more mistakes with darker skin tones, and the proliferation of deepfakes that can manipulate public opinion. Professor Stuart Russell, a pioneer in AI research, warns about the difficulty of retaining control over machines that become more intelligent than humans. He underscores the importance of aligning AI objectives with human values to prevent catastrophic outcomes.

Understanding Human Intelligence to Advance AI

A key frontier in AI research is understanding the vast complexity of the human mind. Neuroscientist Professor Ed Boyden is working on creating a detailed digital map of the brain to better comprehend its intricate neural circuitry. This understanding could pave the way for replicating human-like intelligence artificially. However, the human brain's complexity—comprising around 100 billion neurons—presents a monumental challenge that is still far from being fully mapped or understood.

Conclusion: Navigating the Future of AI

The quest to develop superintelligent AI is fraught with both incredible potential and significant risks. While AI continues to transform various aspects of our lives, ensuring its safe and ethical development is paramount. Balancing innovation with caution, and deepening our understanding of human intelligence, will be crucial as we navigate this new frontier.