David Robinson, former safety engineer at OpenAI, warned on Saturday that AI companies are not being precise enough. These statements come in light of growing concerns that AI could become powerful enough to pose a threat to humanity.
The Need for Safety Measures Similar to Nuclear Energy
Robinson stated in an article for The Atlantic that AI requires safety measures similar to those in place for nuclear energy and aviation. He added, "The time for trial and error is over." He emphasized the necessity of prioritizing safety and research before large tech companies develop more advanced AI models to prevent human errors that could lead to disaster.
Read more: Greg Stanton Does Not View the Use of Flock Cameras as a Tool for Totalitarian Control
Criticism of OpenAI's Approach
Robinson, who oversaw safety reports for 12 AI model startups, acknowledged that recent errors are common as tech giants prioritize "speed and flexibility." He said, "The environment in which these things can happen is not a suitable place to nurture artificial minds that may be smarter than us and may not act the way we want them to."
Robinson noted that OpenAI relies on a "gradual deployment" strategy, releasing new AI models first and only tightening security measures after problems arise. He added, "The company fails to achieve the level of precision required while jumping from one launch to another."
In response to recent criticisms, OpenAI stated that it is ensuring its models do not progress beyond a level that can be safely managed and will take action if it needs to slow down training or hold back models.
In recent months, repeated warnings have been issued by technology experts and policymakers about the rapid advancement of AI models that could take control and potentially destroy humanity. However, U.S. President Donald Trump is strongly opposed to strict regulations, believing that such actions would slow down innovation.
Read more: Hakeem Jeffries Criticized Trump's Opposition to AI Restrictions · The Livelihood Rights Award Was Given to Women's Rights Activists and AI Researchers
