I really think that this blog is part of the IPO story from Anthropic or OpenAI.
Because the reality is: A generalist can simply not guide the LLMs through the system nor can he/she verify the output. LLMs are force-multipliers and if there is nothing to multiply, it won't work.
As long we don't have AGI, humans will guide AI and if we get AGI, that's not something to even think of anymore. Software engineering was always about automation, if you want to be an artist, you better look for something else.
Because the reality is: A generalist can simply not guide the LLMs through the system nor can he/she verify the output. LLMs are force-multipliers and if there is nothing to multiply, it won't work.
As long we don't have AGI, humans will guide AI and if we get AGI, that's not something to even think of anymore. Software engineering was always about automation, if you want to be an artist, you better look for something else.