“`
The Future of Artificial Intelligence: A Focus on Purpose-driven Approaches
At a recent conference at the Massachusetts Institute of Technology (MIT), speakers and attendees deliberated on the question: Who benefits from artificial intelligence (AI)? This question is of paramount importance, especially in the light of AI’s exponential growth in recent years. Source
Journalist Karen Hao, a renowned voice in AI discourse, keynoted one of the conference sessions. Hao is a former Wall Street Journal and MIT Technology Review contributor and the author of the book “Empire of AI.” In her address, she argued for a shift in the direction of AI development.
Challenging the Scale of AI Development
Hao proposed a move away from the current trend of massive data use, extensive data centers, and large models used to develop tools under the umbrella of “artificial general intelligence.” She asserted that such a scale is unnecessary to reap the benefits of AI, adding that if we desire AI to be beneficial on a larger scale, we need to urgently reconsider this approach.
She further expounded on the issues related to the staggering size of datasets used by leading AI companies to develop large language models. These challenges include the massive energy consumption and emissions from hyper-scale data centers, which also consume large quantities of water. Hao also highlighted the human toll of input work, where gig economy workers worldwide manually input data into these models.
A Case for Small, Task-Specific AI Models
As an alternative, Hao suggested the use of smaller, task-specific models like AlphaFold, the Nobel Prize-winning tool for identifying protein structures. She explained that these models are based on highly curated datasets specific to the problem at hand. This eliminates the need for fast supercomputing due to the small size of the datasets and the model, yet it still offers significant benefits.
Making AI Responsive to Communities
In a second keynote address, scientist Paola Ricaurte emphasized the importance of purpose-driven AI approaches. “It makes no sense to have technologies that are not responsive to the communities that will use them,” Ricaurte said. Ricaurte is a professor at the Tecnologico de Monterrey in Mexico and a faculty fellow at the Berkman Klein Center for Internet and Society at Harvard University.
The symposium, titled “Gender, Empire, and AI: Symposium and Design Workshop,” was hosted by the MIT Program in Women’s and Gender Studies and saw over 300 attendees for the keynote speeches. The event also featured group discussions and an afternoon session on design across various subject areas.
Specificity in AI Discourse
In her speech, Hao criticized the vagueness that often characterizes AI discourse, contending that it prevents a thoughtful discussion about the industry’s direction. She compared the term ‘artificial intelligence’ to ‘transportation,’ which could refer to anything from a bicycle to a rocket. Hence, she stressed the need for specificity when discussing access to AI’s benefits.
She further argued that smaller tools – similar to bicycles – are more useful in everyday life. As another example, Hao cited the Climate Change AI project, which focuses on tools that can improve the energy efficiency of buildings, track emissions, optimize supply chains, predict extreme weather events, and more. “This is the vision of AI we should be working toward,” Hao said.
The Role of Public Engagement
Both Hao and Ricaurte encouraged the audience to actively participate in AI discourse and projects, emphasizing that the development of this technology is not predetermined and that public interventions matter. “We have a responsibility to make hope possible,” Ricaurte said, while Hao ended her speech by quoting author Rebecca Solnit, asserting that “hope is located in the premise that we don’t know what will happen, and that in the vastness of uncertainty there is room for action.”
“`
