Artificial General Intelligence (AGI) has become a hot topic in the world of technology and AI. Its potential for advancement and impact on society has captivated the imagination of many, including tech giants like OpenAI. With a recent funding round of $6.6 billion, OpenAI is making serious efforts to bring AGI to life, with the ultimate goal of benefiting humanity. However, even experts in the field, such as Fei-Fei Li, admit to being unsure about what AGI truly is.
Fei-Fei Li, often referred to as the “godmother of AI,” is a renowned researcher who contributed significantly to the development of modern AI by creating ImageNet, a groundbreaking AI training and benchmarking dataset. She has an extensive background in the field, having served as the Chief Scientist of AI/ML at Google Cloud and currently leading the Stanford Human-Centered AI Institute (HAI). Yet, when asked about AGI, Li expressed her confusion and lack of understanding.
Li mentioned that she comes from an academic AI background, where rigorous and evidence-based methods are emphasized. She admitted that she doesn’t spend much time thinking about the definitions and concepts surrounding AGI because there are more important things to focus on. Despite her deep involvement in AI research and development, Li’s perspective highlights the ambiguity and complexity surrounding AGI.
To make sense of AGI, let’s explore the definitions provided by OpenAI and its CEO Sam Altman. OpenAI defines AGI as highly autonomous systems that outperform humans in most economically valuable work. Altman further explains that AGI is equivalent to a median human that you could hire as a coworker. While these definitions provide some insight, they still lack clarity on the exact capabilities and characteristics of AGI.
OpenAI has developed a progression model consisting of five levels to assess its progress toward AGI. The first level is chatbots, followed by reasoners, agents, innovators, and finally organizational AI. These levels are meant to represent increasingly advanced AI systems, but they still don’t provide a concrete definition of AGI. The complexity of AGI surpasses what a median human coworker can achieve, as it includes capabilities such as innovation and the ability to perform the work of an entire organization.
Fei-Fei Li’s journey into AI began long before it became a profitable field. She developed a fascination with intelligence at a young age, leading her to study AI when it was still in its infancy. Li highlighted the pivotal moment in 2012 when her ImageNet, combined with AlexNet and GPUs, ushered in the birth of modern AI. The convergence of big data, neural networks, and GPU computing laid the foundation for the AI boom we are experiencing today.
During the discussion, Li was asked about California’s controversial AI bill, SB 1047, which aimed to impose regulations on AI models. She expressed caution and avoided reigniting the debate, as Governor Newsom had recently vetoed the bill. Li’s approach focuses on looking forward and finding the right balance between preserving human safety and not burdening technologists. She believes that penalizing the engineers behind AI models for any misuse would not make the technology safer. Instead, she advocates for continued innovation, coupled with a well-defined regulatory framework that ensures the responsible use of AI.
While advising California on AI regulation, Li is also leading her startup, World Labs. As a woman at the forefront of AI research, she acknowledges the lack of diversity in the AI ecosystem. Li believes that diverse human intelligence will lead to diverse artificial intelligence and ultimately result in better technology. She is particularly excited about advancing “spatial intelligence,” which involves creating computer systems that have a deep understanding of the three-dimensional world. This goes beyond simply recognizing objects but focuses on enabling computers to navigate, interact, and perform tasks in the physical world.
In conclusion, AGI remains an elusive concept that even experts in the field struggle to define with certainty. Fei-Fei Li, despite her significant contributions to AI, admits to not fully grasping AGI’s meaning. The complexity and potential of AGI drive the ongoing research and development efforts by OpenAI and other organizations. As we continue on this journey toward AGI, it is crucial to advocate for responsible AI practices, promote diversity in the field, and strike a balance between innovation and regulation to ensure the safe and beneficial adoption of this transformative technology.
Source link