Video není dostupné.
Omlouváme se.
Biases are being baked into artificial intelligence
Vložit
- čas přidán 16. 08. 2018
- When it comes to decision making, it might seem that computers are less biased than humans. But algorithms can be just as biased as the people who create them.
signup.axios.com
So ... biased people are going to add algorithmic bias in order to compensate for raw data. How about just get more data so the accuracy improves without humans baking it?
No, biased people are misinterpreting results they don't like as biased.
@@dud3man6969 The video explicitly references verifiably incorrect data being generated by these reports. Unless I misunderstand you, you are suggesting figures such as with the white and black criminal fallback rates are indeed accurate or what are you trying to convey?
@carldressler3262 does that automatically prove some kind of malicious intent, or could it just be that the technology still has a ways to go? It is not perfect yet and it sometimes has a very hard time understanding some concepts that would be obvious to a toddler. I think it is just the state of the technology.
@@dud3man6969 Nope, absolutetly no malicious intent. Technology has no will of its own, it just reproduces what it sees. A common misconception is that algorithms or AI (a very fancy kind of algorithm) do some sort of logical reasoning, but they do not. Just take a look at ChatGPT doing maths. It is only predicitions on what it's seen before and because of that is very bad at maths or other logics-based issues. The AI does not care about race or anything, it just reproduces what it sees.
I don't believe it's reasonable to say that the algorithm is 'behaving in a prejudiced way.' It's behaving the way it was trained. Its bias is not sociological, it's data-driven. Even in those court examples, the AI probably acted very similarly to the human judges and such who provided the data. It's ironic that it's then easy to attack an AI that demonstrates this behavior, but it's apparently NOT easy to solve the problem when it's humans make the same type of decisions.
The algorithms will always be biased so we must always take that into account. We can't just brush it aside with the easy fix of "diversity" in algorithm construction.
Why do i hear Naruto
So you are saying ALL algorithms are race biased 99% of the time. I would really like to see facts about this. It seems more that your own video is biased and SJW propaganda...
He is saying the ones who are creating them might have or in some cases have a cognitive biased towards certain people. Lets be honest, most Tech firms are Liberal SJW.
Stuart: thanks for your reply.
I admit I was a bit fast on my assumption for this video.
I guess I'm getting tired of the media bombarding us with race and gender issues all the time, when in reality, it has never been that equal for race and gender in human history.
Glen Reer it definitely is, they want to enforce their secular religion on everyone.
@Chanel Sterling Like I said, I would really like to see facts about this.
Translation: They're having a hard time writing their biased woke ideology into an algorithm.
Striving for woke AI will hold the technology back at least one decade.
>Muh Libs
Kaveh Waddell looks strange, very odd; mixed race, probably Persian and Scottish: a stingy waspy mullah? No wonder he muses on algorithms and race sensitivity, another millenial wizard, "ever learning, and never able to come to the knowledge of the truth" (HOLY BIBLE; 2 Timothy 3:7).
Become not living