Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Facebook apology as AI labels black men 'primates'

bbc.com

Post reply on HN
Print view View original

61–70 of 291 posts

Re: Facebook apology as AI labels black men 'primates'

#61

Post by shadowgovt »

nolaspring wrote: post #54
IncRnd wrote: post #51

Earlier quoted context omitted.

The reason these companies don't fix these systems is because they don't know how. It is easier to remove certain outputs or retire the whole system. There is no line of code they can tweak.

Richen the dataset it’s trained on enough so that the model is correct before you release it to prod.

We don't actually know how to do that, or how rich is "rich enough." It's an open avenue of research to be able to extrapolate how well-tuned a neutral net is on data not in its training set.

Not to imply the problem is unsolvable, just that if an institution has zero tolerance for this mistake, the fix your describing is no guarantee it won't occur.

Re: Facebook apology as AI labels black men 'primates'

#62

Post by JasonFruit »

MattGaiser wrote: post #40
scotty79 wrote: post #5

Is that a result of a skewed training set or are people really hard to tell apart from gorillas if there are no obvious tells like large difference in brightness of different areas of the face?

I would be interested to know if young children would do much better in many cases. When I was quite young, I referred to some firefighters as robots.

It reminds me of some of the explorers' tales of people who were half-human, half some other animal, or of people covered in hair, the first of which may have originated from seeing people riding animals, and the others to various (actual) primates. If humans can make such mistakes, certainly Facebook's AI can be excused for its confusion.

Re: Facebook apology as AI labels black men 'primates'

#63

Post by MBCook »

IncRnd wrote: post #59
nolaspring wrote: post #54

Earlier quoted context omitted.

Richen the dataset it’s trained on enough so that the model is correct before you release it to prod.

That's sort of obvious. How do you know that wasn't attempted?

If it was we wouldn’t expect this problem to occur, correct?

Re: Facebook apology as AI labels black men 'primates'

#64

Post by tartoran »

q-rews wrote: post #43
tartoran wrote: post #31

Earlier quoted context omitted.

How would you feel if you’d be labeled supremacist (only if your skin complexion was white)? After all the human species is the supreme species... Oh well I guess that wouldn’t bother you and you’d take it as a compliment. Now imagine the reverse of a compliment, to be given one of the labels that has been used again and again to belittle you, to make you feel inferior (only to make others feel above you), a label wh…

I think the core difference is that one is a honest mistake and the latter is just deprecating. It’s fine if AI calls me “white treehouse” but it’s less fine if it labels me “stupid”, even if my face really looks stupid. The reality is that people have become extremely sensitive about some topics even if they’re not the recipients of the perceived insult.

I don’t think any algo should be in the business of labeling any people period. What you call sensitive is offensive to others as much as you being labeled stupid by the look of your face. It is wrong in your case and it is also wrong in the cases you perceive as sensitive. I actually find it quite insensitive of you to dictate what others should feel based on your experience. And selfish too

Re: Facebook apology as AI labels black men 'primates'

#65

Post by belltaco »

IncRnd wrote: post #59
nolaspring wrote: post #54

Earlier quoted context omitted.

Richen the dataset it’s trained on enough so that the model is correct before you release it to prod.

That's sort of obvious. How do you know that wasn't attempted?

That makes it sound even worse that they knowingly released it without fixing it.

Re: Facebook apology as AI labels black men 'primates'

#66

Post by MathMonkeyMan »

tomrod wrote: post #28
civilized wrote: post #14

Earlier quoted context omitted.

Deep learning, for all its recent glories, still suffers from relatively crude, slow-converging training algorithms compared to other areas of ML and statistics. Maybe to your typical SGD-type algorithm, working off a dataset filled with mostly light skin toned people, skin tone just looks like a real solid first-order way to distinguish humans and primates, and picking up the black people / primate distinction seems…

Consider too what they are likely using for inputs: photos with associated comments. I don't know Facebook's TOS sufficiently to know whether they are using private groups as source material, but if you're utilizing bigoted content to train pattern recognition, you will replicate bigoted content.

I wonder why this was downvoted. It's an interesting hypothesis.

Re: Facebook apology as AI labels black men 'primates'

#68

Post by MBCook »

yanlezeiler wrote: post #38

I worked for another computer vision company, Clarifai that had the same issue. One of the employees noticed it and we retrained the model before it became public.

This is what amazes me. Given this exact thing has happened in the past and resulted in public humiliation of the companies involved, how did they not notice this? Why didn’t they check for it?

Re: Facebook apology as AI labels black men 'primates'

#69

Post by systemvoltage »

dadoge wrote: post #52
ALittleLight wrote: post #49

The video features white and black men. It seems like concluding the algorithm is calling black men primates is the same kind of error people are accusing the algorithm/Facebook of. i.e. The reason you think it's racist is because you assume it's talking about black people specifically suggesting you think the word is more apt to describe black people. Primates and humans are similar labels. This was almost certainly…

Intentional or not, the outcome is all that matters. In every aspect of your life

Speak for yourself. Intent is one of the key factors in crime investigations. Even in 'every aspect of life', intent plays a critical role in greasing the society's abrasion. It helps us understand each other better. Did you accidentally bumped me or did you try to push me out of the bus?
Post reply on HN