AI Won't Need to Take Commands from Humans Soon: Ex-Google Boss Eric Schmidt
Former Google CEO Eric Schmidt has issued a warning that artificial intelligence is nearing a stage where it may evolve without human intervention. Speaking at a recent event organized by the Special Competitive Studies Project, a think tank he founded, Schmidt expressed concerns about the implications of this technological shift for human roles in the workforce.
He noted that we are on the verge of a new era where AI systems can learn and enhance themselves independently, potentially displacing highly skilled professionals. "Computers are now capable of self-improvement... They’re learning how to plan without needing our guidance," Schmidt explained. He introduced the concept of "recursive self-improvement," where AI systems create new hypotheses, test them in robotic labs, and integrate the results back into their learning processes all without human input.
Schmidt described this as the future model of AI and biotechnology integration, where AI generates various candidates for testing, and the robotic lab evaluates them before restarting the cycle.
Despite the frequent discussions surrounding AI, Schmidt believes it is "underhyped, not overhyped." He pointed out that while tools like ChatGPT and others are becoming integral to daily life, many people still do not fully comprehend the extent of their impact. He highlighted that individuals are using AI for sensitive tasks, such as relationship and psychological advice, despite these systems not being specifically trained for such purposes.
Schmidt also predicted significant changes in the job market, suggesting that within a year, most programmers could be replaced by AI-driven programming tools. He anticipates that AI will soon outperform top human talent in areas like mathematics, leading to the emergence of "superintelligence" machines that surpass human capabilities.
However, he emphasized that AI should remain a tool under human control, stating, "Scientists are in charge, and AI is assisting them, which is the right order." He stressed the importance of high-quality human-generated data for AI systems and called for the creation of open, reproducible, peer-reviewed datasets to foster responsible innovation.
Schmidt raised concerns about the national security risks associated with unregulated open-source AI models, warning that they could potentially create harmful pathogens. He urged for increased government involvement in AI research funding and oversight, advocating for a system where public support drives university-led innovation. He argued that private philanthropy cannot replace the substantial funding provided by the government.
In response to audience questions, Schmidt reiterated the necessity of human oversight in AI systems, cautioning that just because AI excels at a task does not mean the processes it operates within are sound.
