You have an expert right in your terminal these days. Have you noticed that the junior engineers in your team are reaching out to you with questions less and less? That's because they get most of the answers from the LLMs and not experts/gurus/seniors these days. Also, if I am a generalist I know enough to know what I want to achieve, the end goal. Then I can ask the agent to go deep and build it.
I have a completely different opinion: experts will benefit more in the future if LLM use remains the norm.
LLM is not an expert. You need to be an expert to judge if the output generated by an LLM is good. It may *look* good, but it's not always the case - you could face all kinds of subtle (or not so subtle) issues in the future (performance, security, etc.). I don't believe it's sustainable to rely on a few generalists to handle everything. But the future will tell.
I'm planning to write a post specifically about this.
PS: Don't even start on the junior engineers - that's a planetary-scale problem that we'll need to be solving in the next few years.
You have an expert right in your terminal these days. Have you noticed that the junior engineers in your team are reaching out to you with questions less and less? That's because they get most of the answers from the LLMs and not experts/gurus/seniors these days. Also, if I am a generalist I know enough to know what I want to achieve, the end goal. Then I can ask the agent to go deep and build it.
I have a completely different opinion: experts will benefit more in the future if LLM use remains the norm.
LLM is not an expert. You need to be an expert to judge if the output generated by an LLM is good. It may *look* good, but it's not always the case - you could face all kinds of subtle (or not so subtle) issues in the future (performance, security, etc.). I don't believe it's sustainable to rely on a few generalists to handle everything. But the future will tell.
I'm planning to write a post specifically about this.
PS: Don't even start on the junior engineers - that's a planetary-scale problem that we'll need to be solving in the next few years.