A lot of management work is merely reporting up down sideways. The manager:engineer ratio is high because a human manager cannot keep up with so much reporting for so many people. They can't even keep up with all 1:1s.
With generative AI, a lot of reporting will become automated, reducing the need for managers to spend time on reporting. Efficient companies will figure out that manager:engineer ratio does not need to be so high.
That's the thing I'm seeing with AI from people with careers directly related to intelligence.
We keep raising the bar. First it was the the Turing test. AI has practically Rocketed past that and now a lot of people think it's a useless metric.
Now it beats mba students. I guess the new bar is CS students.
When your occupation is tied to your career people have a lot of pride invested in it. So they become in denial. They think in terms of their own skill and whether or not the AI can best them in the skills they spent years honing.
Two things:
1. An MBA student is a high bar. Any form of human intelligence is a high bar. Stop blinding yourself.
2. All trend lines point to AI surpassing CS students as well. AI has improved significantly in the last decade. The next decade will only be more improvement.
I have a CS degree and have worked for 20+ years and designed and architected pretty impactful systems.
Yet when I had a conversation with chatGPT to do some architectural brainstorming, it was on par with, or better than most architectural discussions I’ve had with others in my career. Including discussing tradeoffs, UI features, order of developing them, etc.
Sure you kind of have to know what to ask it. But it was the first moment for me where I realized even my job is at risk.
So I agree that MBA students aren’t a high bar. It even does a decent job with software solutioning and architecture.
>So I agree that MBA students aren’t a high bar. It even does a decent job with software solutioning and architecture.
Architecture discussions are basically similar to qualitative and often hand wavy conjecture... it's no surprise LLMs excel at this.
Ironically the industry has it backwards. They think "architecture" is some hyper advanced thing. It's actually far easier than coding. You have less primitives to work with and you can handwave shit and get things wrong all the time in discussions.
The MBA bar is the same bar as software architecture discussions. Same arena, same level of hardness.
The bar I'm more referring to is coding. LLMs have not quite matched us completely in that bar yet. But it's not far off.
The kind of engineer that requires the level of supervision chatgpt does is a non engineer. Someone that has not yet started coding but it just regurgitates text with little sense.
I have been told that "imagination" is a misnomer, much like "free will", it's an artifact of an observed phenomenon that has no actual basis in reality. We do not actually invent or imagine anything new, although we can conflate multiple memories into new forms, or even serendipitously discover things, we cannot actually invent any new ideas that we haven't been introduced to before. As an example, take the invention of the wheel. A tree falling over in a windstorm falls onto another tree that is already present, and as a result it begins to roll down a hill quite easily. This occurrence is observed by a human who later in time employs the principle to move a heavy object. The knowledge is preserved through language transfer, eventually developing into the "wheel" we know today.
20 comments
[ 4.3 ms ] story [ 90.1 ms ] threadWith generative AI, a lot of reporting will become automated, reducing the need for managers to spend time on reporting. Efficient companies will figure out that manager:engineer ratio does not need to be so high.
We keep raising the bar. First it was the the Turing test. AI has practically Rocketed past that and now a lot of people think it's a useless metric.
Now it beats mba students. I guess the new bar is CS students.
When your occupation is tied to your career people have a lot of pride invested in it. So they become in denial. They think in terms of their own skill and whether or not the AI can best them in the skills they spent years honing.
Two things:
1. An MBA student is a high bar. Any form of human intelligence is a high bar. Stop blinding yourself.
2. All trend lines point to AI surpassing CS students as well. AI has improved significantly in the last decade. The next decade will only be more improvement.
Yet when I had a conversation with chatGPT to do some architectural brainstorming, it was on par with, or better than most architectural discussions I’ve had with others in my career. Including discussing tradeoffs, UI features, order of developing them, etc.
Sure you kind of have to know what to ask it. But it was the first moment for me where I realized even my job is at risk.
So I agree that MBA students aren’t a high bar. It even does a decent job with software solutioning and architecture.
Architecture discussions are basically similar to qualitative and often hand wavy conjecture... it's no surprise LLMs excel at this.
Ironically the industry has it backwards. They think "architecture" is some hyper advanced thing. It's actually far easier than coding. You have less primitives to work with and you can handwave shit and get things wrong all the time in discussions.
The MBA bar is the same bar as software architecture discussions. Same arena, same level of hardness.
The bar I'm more referring to is coding. LLMs have not quite matched us completely in that bar yet. But it's not far off.
That's what's saving our jobs - for now.
I just do not respect MBA students.
Source: I used to teach MBA students.