Genuinely shocked to see something like this on the front page of HN. Framing your misinformation in the form of a debunking doesn't mean it's true.
130 karma · joined September 3, 2019
While it would be a symmetric situation a vacuum, we do not live in a vacuum. And acknowledging dataset bias by itself doesn't address the bias meaningfully. In practice, we often treat the bias as an exogenous factor when it is not, moving it outside the scope of our responsibility. But it is very much the product of our work, a reflection of our choices, values, and beliefs about what to prioritize. We can't abdicate our responsibility for it; the stakes are too high.
That's the deeper question ML research/industry has to grapple with. When these models are deployed increasingly quickly and at scale in ways that can potentially cause massive harm, why is it okay to keep doing the same exact harmful things?
LeCun mentioned that there would have been opposite problem if it were trained on a dataset from Senegal. But why wasn't it trained on a dataset from Senegal? Why do we always see these errors where white-centric datasets produce white-centric results?
It's obvious that while it would be a symmetric situation a vacuum, we do not live in a vacuum. We live in a world with deep sociocultural biases in favor and against various racial groups. And it is unjust to let AI perpetuate and entrench these biases by acting at with this bias at scale.
Acknowledging dataset bias by itself doesn't address the bias meaningfully. In practice, we often treat the bias as an exogenous factor when it is not, moving it outside the scope of our responsibility. But it is very much the product of our work, a reflection of our choices, values, and beliefs about what to prioritize. We can't abdicate our responsibility for it (even if we choose not to prioritize it.)
If reducing crime is our only objective, we can do so trivially by indefinitely detaining anyone who ever does something suspicious. I think most people would find this system to be deeply unjust, even if it eliminated crime.
As a society, we have some tolerance for crime, preferring to err on the side of personal liberty against totalitarianism.
Asking for fairness in this recidivism model isn't a case of political correctness run amok, it's a reasonable request from stakeholders in the system it affects: Why should an individual be punished for the actions of others who happen to have similar identity markers?
You can say that a biased system will be "more efficient", but efficient to what end? A system that is unfair may reduce crime but it won't achieve "justice".
Since the senate is fixed at 2 senators per state, it massively advantages small states in terms of political power per person. The number of representatives per state is currently set at 435, and is allocated proportionately to population. This number is arbitrary, and can be changed by an act of congress. Increasing it will dilute the power of the electors from the Senate.
There's also compelling reason to increase the number of for its own sake: the number of people per representative is higher now than it has ever been. And it's much higher than other Western democracies.
When the Constitution was written, the number of representatives was set so there'd be one for every 30,000 people. The idea was that the number would be increased over time, and there was almost almost an amendment ratified to ensure this would happen, where the ratio would never get above 50,000 per.
But the number of representatives hasn't been updated since 1929. Back then, there was one representative per 218,000 people. Now there's one representative per 744,000. If we had the 1929 ratio in place, we'd have 1505 representatives. And if we set representatives at 50,000 per person (as preferred by founders such as Washington), we'd have 6564.
Right, they passed a law, AB5, to better do that job.