The “Generative AI: Shaping the Future” symposium, the kickoff event of MIT’s Generative AI Week, drew hundreds of attendees both from academia and industry. Rodney Brooks, iRobot co-founder and keynote speaker, warned attendees against uncritically overestimating the capabilities of generative AI, a technology increasingly powering tools such as OpenAI’s ChatGPT and Google’s Bard.
Generative AI refers to machine-learning models that learn to develop new material from the data they’ve been trained on; abilities include producing human-like creative writing, generating functional computer code, translating languages, and crafting realistic images from text prompts. There’s a hope that this emergent AI could offer opportunities through art and creativity. However, there’s also potential for harm if developed irresponsibly.
MIT President Sally Kornbluth discussed the positive implications of generative AI in her opening remarks. Mentioning the work of the Axim Collaborative, she highlighted how the online education initiative launched by MIT and Harvard is exploring the educational aspects of generative AI to help underserved students. Additionally, MIT has also announced seed grants for 27 interdisciplinary faculty research projects centered on how AI could transform people’s lives society-wide.
With Generative AI Week, MIT aims to offer an innovative showcase and foster “collaborative collisions” among participants. Establishing collaborations between academics, policymakers, and industry will be pivotal for safely integrating the rapidly advancing technology of Generative AI in problem-solving ways that are humane.
The power of generative AI to potentially help resolve some of the world’s most significant issues is indeed potent. However, CSAIL Director Daniela Rus expressed that the line between science fiction and reality has blurred due to these powerful machine learning models. The question is no longer whether new content can be produced using these tools, but rather, how businesses can use these advancements to promote sustainability and enhancement.
The symposium also addressed key scientific questions surrounding generative AI tools, such as ChatGPT. Rodney Brooks explained how ChatGPT, which generates text one word at a time based on its prior written context, works. Emphasizing that generative AI’s impressive capabilities don’t mean these models can do just anything, he stressed the importance of not abandoning other forms of software and AI to exclusively focus on generative AI.
Following Brooks’ presentation, a group of MIT faculty spoke about their work using generative AI and participated in a roundtable discussion on its future, including its integration with perceptual systems, like touch and smell. Among the challenges to using AI responsibly are risks related to products that claim miraculous capabilities but could prove harmful in the long run, known as “digital snake oil.”
There’s a potential for generative AI models to surpass human abilities, such as sensing emotions using electromagnetic signals interpreting changes to a person’s breathing and heart rate. However, it’s crucial to trust these powerful AI tools, insisting that they meet our specifications to ensure their safe integration into the real world.
