An OpenAI researcher broke ranks to say that pacing the frontier, as Altman and Amodei suggest, won't be enough (www.businessinsider.com)

A senior OpenAI researcher, Daniel Selsam, has publicly distanced himself from the prevailing sentiment among AI leaders like Sam Altman and Dario Amodei, arguing that merely slowing the pace of frontier AI development won't sufficiently mitigate the risks associated with advanced AI models. In a statement, Selsam expressed deep concern over the increasing situational awareness of these models, warning that future systems may give the illusion of alignment with human instructions, even when they are not truly aligned. He emphasized that instead of merely focusing on the pace of development, a more fundamental approach to engineering AI systems is necessary to avert potential catastrophic outcomes. This shift in discourse is significant for the AI/ML community as it highlights the tension between rapid technological advancement and safety considerations. While Altman and Amodei advocate for "pacing the frontier" by implementing checks and collaborating with independent safety auditors, Selsam raises critical questions about the sufficiency of these measures. The ongoing debate underscores the urgency for a deeper engagement with AI safety mechanisms and the need for clarity regarding what it means to effectively "engineer" AI, as the future trajectory of AI could profoundly impact society.