Artificial intelligence (AI) has undoubtedly taken the world by storm, with rapid advancements in language models, computer vision, and a wide range of practical applications.

However, in this episode of the #shifthappens podcast with Peter Voss, the CEO and Founder of Aigo.AI, we see how the current state of AI is still a far cry from the ultimate goal of true, human-level artificial general intelligence (AGI).

Read on as we explore the fascinating story of Voss, who shares his decades-long journey into AI, starting with his own software company and eventually setting his sights on the challenge of building AGI systems.

The Limitations of Narrow AI

Voss began by drawing a clear distinction between narrow AI and AGI. “Narrow AI,” he explains, refers to the current state of AI technology, where systems are trained to excel at specific tasks, like playing chess or generating human-like text. While impressive in their own right, these systems lack the fundamental qualities of human-level intelligence.

“If you look at all the efforts over the decades on AI, it’s always been narrow AI trying to solve one particular problem at a time,” Voss says. “The focus is no longer on having the intelligence in the system itself that can learn.”

Instead, the intelligence lies in the data scientists and programmers who design the algorithms and fine-tune the systems to achieve their desired outcomes. This, Voss argues, is a far cry from the original vision of AI, which was to create “thinking machines” that can learn, reason, and adapt as humans do.

The Path to Artificial General Intelligence

Recognizing the limitations of narrow AI, Voss and a small team of like-minded individuals set out in the early 2000s to pursue the holy grail of AI: artificial general intelligence. This led them to coin the term “AGI” in 2002 to differentiate their approach from the increasingly narrow focus of the field.

“In 2001, when I started on my project, I realized that we need to get back to the original vision of AI—to build thinking machines and get away from narrow AI,” Voss says.

The key to achieving AGI, according to Voss, lies in understanding the fundamental aspects of human intelligence, such as concept formation, metacognition, and the ability to learn incrementally and adapt in real time. With their reliance on massive datasets and brute-force computing power, current AI systems fall short in these areas.

“We need to focus on the adaptive nature of intelligence and what it requires,” Voss explains. “We need incremental real-time learning and metacognition to help the systems understand what they need to learn compared to simply understanding how to complete one task.”

The Promise of True AGI

Voss’ vision for the future of AGI is nothing short of transformative. He envisions a world where AGI systems can revolutionize scientific research, dramatically reduce the cost of goods and services, and serve as personalized assistants that help us make better decisions and live more fulfilling lives.

“Imagine having one AGI that has trained itself to be a PhD level cancer researcher. So, now you can make a million copies of that AI. You have a million PhD-level cancer researchers chipping away at the problem, each pursuing different ideas across various paths, but collaborating much better than humans will because they don’t have an ego getting in the way.”

Similarly, Voss believes that the abundance created by AGI-powered automation could radically reduce the cost of goods and services, and the personalized nature of AGI assistants could help us all make better decisions and achieve our personal and professional goals.

“Imagine each person in the world having a personal assistant. It serves your agenda, not some mega corporations’ agenda. Imagine it being hyper-personalized to you: it gets to know your history, your goals, your dreams, who your friends are, what you’re working on, and so on.”

Challenges and Cautionary Notes

Of course, Voss acknowledges that the path to AGI is fraught with technological and practical challenges. He warns against the tendency of companies to chase the hype around current AI models, such as large language models, without fully understanding their limitations.

“Companies need to ask themselves what AI technology is appropriate for different tasks. GenAI cannot be used in critical situations without enormous guardrails.”

Voss also cautions against the dangers of regulatory capture, where large tech companies could use their influence to shape the rules governing AI development in a way that stifles innovation and protects their own interests.

Nonetheless, Voss remains optimistic about the future of AGI and the profound impact it could have on the human condition. He eloquently states, “To me, that is the future I see with AGI: the phenomenal improvement of the human condition.”

Continuing the Pursuit of True, Human-Level AGI

Peter Voss’ journey into the world of artificial intelligence has been a long and arduous one. Still, his unwavering commitment to pursuing true, human-level AGI is truly inspiring. Through his insights and vision, he has challenged the status quo of narrow AI and painted a compelling picture of a future where AGI systems can revolutionize scientific discovery, dramatically improve our quality of life, and unlock new human potential.

As the world continues to grapple with the rapid advancements in AI technology, we must heed the lessons and warnings that Voss has shared. By maintaining a clear-eyed perspective on the limitations of current AI models and the immense challenges of achieving AGI, we can work towards a future where the true power of artificial intelligence is harnessed to benefit all of humanity.