Even the 'godmother of AI' does not have clarity on what AGI is.
Do you have questions about artificial general intelligence, or AGI? It refers to that goal that OpenAI aims to achieve, with the intention of making it something that "benefits everyone."
Artificial General Intelligence (AGI) is a concept that has generated confusion and debate in the fields of technology and artificial intelligence. OpenAI recently raised $6.6 billion with the goal of advancing its development, yet its definition remains unclear to many, including experts in the field.
In a recent discussion at the Credo AI Responsible AI Leadership Summit, Fei-Fei Li, a recognized researcher in artificial intelligence, shared her own uncertainty about the meaning of AGI. Despite her academic background in AI and her involvement in the development of foundational technologies like ImageNet, Li admitted that she does not have a clear idea of what AGI really entails.
Li commented that, although she comes from an academic approach to artificial intelligence, she does not feel familiar with many of the terms surrounding AGI and AI singularity, expressing that she believes there are more relevant issues to address.
The definition of AGI has been complicated, even for OpenAI, which has proposed multiple levels of progress toward this goal. OpenAI’s CEO, Sam Altman, described AGI in an interview as an entity that can perform tasks equivalent to those of an average human peer. However, the company’s internal classifications range from chatbots to systems capable of running an entire organization. The complexity of these concepts leaves many, including Li, in a state of confusion.
Since her beginnings in the field, Li has been fascinated by intelligence and its potential. In 2012, her work with ImageNet and technologies like AlexNet and GPUs was pivotal in driving the revolution of modern artificial intelligence. This advancement is attributed to the convergence of massive data, neural networks, and advanced computing.
Regarding the recent California artificial intelligence law, SB 1047, Li was cautious in addressing the topic due to its veto by Governor Newsom. However, she expressed optimism about being invited to participate in a task force that will develop policies for the safe deployment of artificial intelligence in the state. Li emphasized that it is crucial to focus regulations not only on technology but also on how it affects people and communities.
Aside from her role in AI regulation, Li heads her new startup, World Labs, in San Francisco. She highlighted the need for greater diversity in the artificial intelligence ecosystem, convinced that human diversity will contribute to a more inclusive and effective technological development.
In the coming years, Li intends to work on "spatial intelligence," proposing that understanding the world in three dimensions will pose a greater challenge than interpreting human language, which has evolved over millennia. For Li, this involves not just making machines see, but also understanding and acting in a three-dimensional context.
