4 ms·

I absolutely loathe this puerile tendency to attribute anything even remotely relatable to modern data systems as Shikata ga nai by definition. Grow up. this

I absolutely loathe this puerile tendency to attribute anything even remotely relatable to modern data systems as Shikata ga nai by definition. Grow up. this meatgrinder didnt just condense from aether. Someone wrote this, someone can change it, and anyone in the git log can be held accountable for what they did at any time. Enough of this late-stage capitalist attempt to offshore even the very concept of accountability.

lovich 4y ago
Individuals can’t enact this sort of heroic change on their own by design with systems this large. There is no single point where the systems behavior can be pivoted. It requires society as a whole to demand the change.
- wombatpm 4y ago
  Until we can have open season on MBA’s and lawyers to thin the herd we can’t change anything
eli_gottlieb 4y ago
Speaking of late-stage capitalism, the actual problem here is that we've built the infrastructure for a rather authoritarian form of computerized socialism -- but under capitalist ownership. So we all walk around pretending a car that can repossess itself if you miss a loan payment is a piece of private property, when the physical infrastructure says that it's social property.
voytec 4y ago
The Trolley Problem is seen differently by someone who has to push the lever inside the trolley, by someone who has to do it remotely from far away and by a coder programming trolley's software.
koboll 4y ago
>Someone wrote this, someone can change it, and anyone in the git log can be held accountable for what they did Uh, not really, no. The whole point of neural nets is they independently discover novel and unexpected ways of predicting the probable answer.
- nerdponx 4y ago
  This is a seriously naive misunderstanding of how neural networks (and machine learning "algorithms" in general) work. Someone chose the features, the model architecture, the training data, and SGD parameters; it might not even be a NN (and probably isn't) — it could be a traditional linear regression model, or gradient-boosted decision trees. Someone analyzed the model performance on a test set and maybe even ran simulations to see how effective the model was at learning a known functional form. Someone made reports on that performance and included estimates like "probability of falsely denying claims on average" broken down by 10-year age bracket. An intern might have helped. A team lead coordinated meetings with whatever VP was running the project and signed off on the model. Senior management budgeted months worth of person-hours on model development and integration/deployment into production systems. And the model could have been rejected or modified at any point. The model training code is probably sitting on someone's laptop right now, and absolutely can be revised at any point in time, either slightly or significantly. This is no magic in something like this. There are no positronic brains to wonder about. It's just a fucking best-fit line, and someone had to both draw the damn line and sign off on it.
  - twawaaay 4y ago
    Yeah... "unsupervised" learning my ass. After ten thousand tries and tinkering with parameters until it finally clicked. Now if you could only make an algorithm that would learn how to tune algorithms...
    - nerdponx 4y ago
      > Now if you could only make an algorithm that would learn how to tune algorithms... That exists! https://www.automl.org/ https://www.automl.org/

lovich 4y ago

Individuals can’t enact this sort of heroic change on their own by design with systems this large. There is no single point where the systems behavior can be pivoted. It requires society as a whole to demand the change.

wombatpm 4y ago
Until we can have open season on MBA’s and lawyers to thin the herd we can’t change anything

eli_gottlieb 4y ago

Speaking of late-stage capitalism, the actual problem here is that we've built the infrastructure for a rather authoritarian form of computerized socialism -- but under capitalist ownership. So we all walk around pretending a car that can repossess itself if you miss a loan payment is a piece of private property, when the physical infrastructure says that it's social property.

voytec 4y ago

The Trolley Problem is seen differently by someone who has to push the lever inside the trolley, by someone who has to do it remotely from far away and by a coder programming trolley's software.

koboll 4y ago

>Someone wrote this, someone can change it, and anyone in the git log can be held accountable for what they did Uh, not really, no. The whole point of neural nets is they independently discover novel and unexpected ways of predicting the probable answer.

nerdponx 4y ago
This is a seriously naive misunderstanding of how neural networks (and machine learning "algorithms" in general) work. Someone chose the features, the model architecture, the training data, and SGD parameters; it might not even be a NN (and probably isn't) — it could be a traditional linear regression model, or gradient-boosted decision trees. Someone analyzed the model performance on a test set and maybe even ran simulations to see how effective the model was at learning a known functional form. Someone made reports on that performance and included estimates like "probability of falsely denying claims on average" broken down by 10-year age bracket. An intern might have helped. A team lead coordinated meetings with whatever VP was running the project and signed off on the model. Senior management budgeted months worth of person-hours on model development and integration/deployment into production systems. And the model could have been rejected or modified at any point. The model training code is probably sitting on someone's laptop right now, and absolutely can be revised at any point in time, either slightly or significantly. This is no magic in something like this. There are no positronic brains to wonder about. It's just a fucking best-fit line, and someone had to both draw the damn line and sign off on it.
- twawaaay 4y ago
  Yeah... "unsupervised" learning my ass. After ten thousand tries and tinkering with parameters until it finally clicked. Now if you could only make an algorithm that would learn how to tune algorithms...
  - nerdponx 4y ago
    > Now if you could only make an algorithm that would learn how to tune algorithms... That exists! https://www.automl.org/ https://www.automl.org/