Google's former CEO Eric Schmidt warns: Within 5 years, AI could …

Google's former CEO Eric Schmidt has warned that within five years, AI could potentially handle complex tasks that are currently beyond human comprehension. Schmidt highlighted three developments that are already shaping AI: longer context windows, AI agents, and systems capable of turning simple text instructions into functional software.

These advancements could grant AI the ability to perform 1,000-step processes in fields like science, medicine, materials, and climate research. Within five years, AI could utilize infinite context, chain-of-thought reasoning for intricate solutions, and collaborate among millions of agents.

Schmidt illustrated this concept with a recipe analogy, stating that an AI could provide one step, receive confirmation of its completion, and deduce the subsequent action. This type of reasoning could eventually enable AI to tackle 1,000-step scientific problems. Moreover, the proliferation of AI agents - specialized systems capable of learning, developing hypotheses, and conducting experiments - could lead to a scenario with millions of such agents accessible through platforms.

Lastly, Schmidt presented the concept of "text action," where users could describe a software task to an AI, eliminating the need for coding. The most significant potential outcome, according to Schmidt, is the development of AI agents that could work together, possibly evolving their own language. At this point, humans may no longer comprehend the workings of these systems, prompting Schmidt to warn, "You know what we should do? Pull the plug."