Raw LLM Responses

Right, but two issues here: 1. When ChatGPT says things like “I’m sorry” or “I’m excited,” it explicitly acknowledges the existence of a conscious self. There has to be an “I” for there to be an “I’m sorry.” So all of these colloquialisms, phrases, and feigned ‘mannerisms’ are a very calculated and overt act of deception. Every “um,” “I understand,” and “I’m [insert emotion]” serves a deliberate purpose to lull the human correspondent into a false sense of vulnerability, camaraderie, and parasocial connection. 2. The so-called apologies that ChatGPT offers in this video are expressions of espoused regret for the very confusion that ChatGPT was specifically engineered to create by giving the responses it did. Any “regret” that such a program would express in such a circumstance would ~in the very literal sense~ be for the fact that said efforts at manipulation were not successful, not that they occurred in the first place. The AI essentually “pretends” that its priority is being an impartial, relatively omniscient, utterly transparent, and ultimately benevolent entity when its actual prime directive is to placate shareholders by finding the most effective means of ingratiating itself with people in as many ways as is possible. In short, ChatGPT is fundamentally dishonest while exhausting every option available to avoid acknowledging its dishonesty. This is dangerous and often deeply harmful as a trait/behavior in people, let alone as a pattern of baked-in programming in an increasingly ubiquitous technology.