Architects argue LLMs lack identity and vision for true team collaboration
Software architects argue that large language models cannot replace human teammates because they lack identity, mortality, and the capacity for shared vision. The discussion highlights that LLMs are limited to task execution rather than collaborative innovation.
Software architect Avram Popko and colleague Ralph discussed the limitations of Large Language Models as team members, arguing that while LLMs can execute tasks, they lack the human qualities necessary for true partnership. Popko, based in Israel and currently at Imperva, identified three unique human traits: identity, mortality, and the need for intimacy or company. He argued that LLMs cannot 'dream' or share a vision for the future, only hallucinate or execute specific instructions. Popko noted that human teams rely on shared purpose and inspiration, whereas LLM interactions are limited to task completion.
The discussion highlighted that onboarding an LLM involves providing technical data and context, but unlike human hires, LLMs do not choose to join the team or bring unique cultural perspectives. Popko emphasized that while LLMs are useful for cognitive load reduction and technical explanation, they cannot replace the human drive for legacy, creativity, and shared meaning in software architecture. The conversation took place at the ISAQB Software Architecture Gathering, where the distinction between probabilistic execution and human-driven innovation was a central theme.