Eric Schmidt's proposed 'hate spell-checker' has a few ethical problems | Hacker News Reader

Eric Schmidt's proposed 'hate spell-checker' has a few ethical problems

thestack.com

It is always weird reading anything Schmidt writes because it feels like he's using his own definitions of things that he never clarifies. He says he's against censorship, but then says that we need to act quick to remove certain videos. How does he define censorship?

He must define censorship as suppression of views he agrees with.

Why language filters don't work, is they can't properly analyze the intent of the language very well... I gave these two examples to "not" implement a language filter at a previous job. The "fowl language" comment was actually positive intent, while the proper language comment was somewhat vile.

"Man that is some bad-ass shit right there, fucking awesome!"

"The pile of human excrement shown isn't worth the cost of the tissue used to wipe my backside."

I think you're conflating two different things. The first quote is impolite but affirming, while the second one is polite but extremely negative, maybe even hateful.

It's true that a simple keyword filter won't work very well if your goal is to detect the hate rather than the cursing.

But there are also mature techniques for doing sentiment analysis. The canonical example is for automated analysis of movie reviews. Those techniques would be much more likely to pick up on the positivity in "bad-ass" and "awesome!" and the negativitiy in "pile of", "isn't worth", "wipe", and "backside".

https://en.wikipedia.org/wiki/Sentiment_analysis

Beyond ethical issues, the repercussions for services under its the Alphabet-Google umbrella would be bad. I can easily imagine YouTube videos suppressed simply because Google's speech-to-text algorithms, anyone who's try its auto Closed Caption probably has seen some insanely bizarre mis-translations. It's easy to start conjuring a lot hypothetical issues with this, from YouTube comments to Blogger posts.

Censorship! of course, why didn't we think of that before! lets limit the hate speech and create safe spaces for everyone so that pesky reality doesn't come and muck things up.

Goes to show how dumb some smart people can be. Technology isn't the be-all-end-all solution to all problems; and as Eric is well familiar with, it makes censorship and surveillance much, much easier.

Just because we can doesn't mean we should.

Reply on news.ycombinator.com