It’s because their positions are often like that “rest of the owl” drawing meme, only it makes sense to them because other people do the filling in of the details and solving the problems. So when an AI can produce the early part of that drawing and confidently promises that it can fill in the rest of the owl, they see it as the same as what their teams were doing prior and unironically believe that them saying “ok, go do that” is the important part, so an LLM should be as competent as a team of engineers.
It takes an engineer who knows the material well enough to see that LLM accuracy is incredibly low, even when it seems to be making sense.
Conversations about AI aside for a moment, God bless random trade dudes making YouTube videos. Thanks to them, I’ve learned about 80% of most jobs can be picked up with minimal training.
It takes an engineer who knows the material well enough to see that LLM accuracy is incredibly low, even when it seems to be making sense.
This was my take until even this year, but honestly it has improved since a year or even six months ago.
It still lies to you and needs to be given pointers constantly, and many other caveats, but the reality is that all of the investment and coming up with the failure loop perfected by Claude Code changed things IMO.
It’s really depressing to think about how all of these rich fucks set a trillion dollars on fire to eliminate one of the only good paying careers available. It’s almost like it’s time to riot or something. 🤷
I still don’t think that means the c suite will be able to fire all of the programmers. It’ll still be the nerds’ job to get the robot to produce the software. It’s likely just going to make life more miserable for the remaining programmers because more and more will be expected of less of them.
It’s because their positions are often like that “rest of the owl” drawing meme, only it makes sense to them because other people do the filling in of the details and solving the problems. So when an AI can produce the early part of that drawing and confidently promises that it can fill in the rest of the owl, they see it as the same as what their teams were doing prior and unironically believe that them saying “ok, go do that” is the important part, so an LLM should be as competent as a team of engineers.
It takes an engineer who knows the material well enough to see that LLM accuracy is incredibly low, even when it seems to be making sense.
I bet AI is especially enticing to those of the “It can’t be that hard” mentality.
Conversations about AI aside for a moment, God bless random trade dudes making YouTube videos. Thanks to them, I’ve learned about 80% of most jobs can be picked up with minimal training.
This was my take until even this year, but honestly it has improved since a year or even six months ago.
It still lies to you and needs to be given pointers constantly, and many other caveats, but the reality is that all of the investment and coming up with the failure loop perfected by Claude Code changed things IMO.
It’s really depressing to think about how all of these rich fucks set a trillion dollars on fire to eliminate one of the only good paying careers available. It’s almost like it’s time to riot or something. 🤷
I still don’t think that means the c suite will be able to fire all of the programmers. It’ll still be the nerds’ job to get the robot to produce the software. It’s likely just going to make life more miserable for the remaining programmers because more and more will be expected of less of them.