DeepMind Co-founder Foresees the Rise of Interactive AI Over Generative Models Like ChatGPT

CHATGPT TO BE REPLACED BY "INTERACTIVE AI"

Mustafa Suleyman, the co-founder of DeepMind—a company that Google acquired for $500 million in 2014—recently shared his insights on the future of artificial intelligence. According to Suleyman, the existing wave of generative AI models, such as ChatGPT, is expected to give way to a new era of "interactive AI."

Suleyman envisions this transformation as a significant milestone in the technological landscape, moving humanity beyond mere information retrieval to a state where AI can execute tasks and services autonomously. In an interview with MIT Technology Review, he articulated the progression of AI technology in three waves. The first wave focused on classification, where deep learning enabled computers to categorize various types of data, from images to language. The current, second wave is characterized by generative models that take this categorized data and produce new information.

The anticipated third wave, dubbed "interactive AI," aims to take the capabilities of AI a step further. Suleyman believes conversational interfaces will become more integral in this phase, replacing traditional forms of user interaction, such as clicking buttons or typing commands. The advent of interactive AI will allow users to delegate tasks to their AI systems, which will then communicate with other people or AI systems to carry out those tasks.

According to Suleyman, the transition to interactive AI marks a profound shift in technological capability and offers a level of agency to tools we create. "It’s truly a step change in the history of our species that we’re creating tools that have this kind of agency," he commented.

Suleyman also emphasized the importance of ethical considerations in this emerging landscape. When queried about the risks associated with granting AI a degree of autonomy, he advocated for clear boundaries to ensure the technology remains aligned with human interests. During his tenure at DeepMind, the team worked on safety measures, including the conceptualization of an emergency "off switch" for AI systems that may operate outside of intended parameters.

The perspective offered by Suleyman underscores the continuing evolution of AI and its increasing influence on various aspects of life, marking what he describes as a "very, very profound moment in the history of technology."