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

Imagen, a text-to-image diffusion model

gweb-research-imagen.appspot.com

Post reply on HN
Print view View original

61–70 of 661 posts

Re: Imagen, a text-to-image diffusion model

#61

Post by karpierz »

tines wrote: post #42
daenz wrote: post #7

>While we leave an in-depth empirical analysis of social and cultural biases to future work, our small scale internal assessments reveal several limitations that guide our decision not to release our model at this time. Some of the reasoning: >Preliminary assessment also suggests Imagen encodes several social biases and stereotypes, including an overall bias towards generating images of people with lighter skin tones…

This raises some really interesting questions. We certainly don't want to perpetuate harmful stereotypes. But is it a flaw that the model encodes the world as it really is, statistically, rather than as we would like it to be? By this I mean that there are more light-skinned people in the west than dark, and there are more women nurses than men, which is reflected in the model's training data. If the model only gener…

It depends on whether you'd like the model to learn casual or correlative relationships.

If you want the model to understand what a "nurse" actually is, then it shouldn't be associated with female.

If you want the model to understand how the word "nurse" is usually used, without regard for what a "nurse" actually is, then associating it with female is fine.

The issue with a correlative model is that it can easily be self-reinforcing.

Re: Imagen, a text-to-image diffusion model

#62

Post by rvnx »

ceejayoz wrote: post #40
user3939382 wrote: post #23

Earlier quoted context omitted.

Translation: we need to hand-tune this to not reflect reality but instead the world as we (Caucasian/Asian male American woke upper-middle class San Fransisco engineers) wish it to be. Maybe that's a nice thing, I wouldn't say their values are wrong but let's call a spade a spade.

"Reality" as defined by the available training set isn't necessarily reality. For example, Google's image search results pre-tweaking had some interesting thoughts on what constitutes a professional hairstyle, and that searches for "men" and "women" should only return light-skinned people: https://www.theguardian.com/technology/2016/apr/08/does-goog... Does that reflect reality? No. (I suspect there are also mostly u…

If your query was about hairstyle, why do you even look or care about the skin color ?

Nowhere there is any precision for a preferred skin color in the query of th user.

So it sorts and gives the most average examples based on the examples that were found on the internet.

Essentially answering the query "SELECT * FROM `non-professional hairstyles` ORDER BY score DESC LIMIT 10".

It's like if you search on Google "best place for wedding night".

You may get 3 places out of 10 in Santorini, Greece.

Yes you could have an human remove these biases because you feel that Sri Lanka is the best place for a wedding, but what if there is a consensus that Santorini is really the most appraised in the forums or websites that were crawled by Google ?

Re: Imagen, a text-to-image diffusion model

#64

Post by curiousgal »

josho wrote: post #52
user3939382 wrote: post #23

Earlier quoted context omitted.

Translation: we need to hand-tune this to not reflect reality but instead the world as we (Caucasian/Asian male American woke upper-middle class San Fransisco engineers) wish it to be. Maybe that's a nice thing, I wouldn't say their values are wrong but let's call a spade a spade.

Translation: AI has the potential to transform society. When we release this model to the public it will be used in ways we haven’t anticipated. We know the model has bias and we need more time to consider releasing this to the public out of concerns that this transformative technology further perpetuate mistakes that we’ve made in our recent past.

> it will be used in ways we haven’t anticipated

Oh yeah, as a woman who grew up in a Third World country, how an AI model generates images would have deeply affected my daily struggles! /s

It's kinda insulting that they think that this would be insulting. Like "Oh no I asked the model to draw a doctor and it drew a male doctor, I guess there's no point in me pursuing medical studies" ...

Re: Imagen, a text-to-image diffusion model

#65

Post by Ar-Curunir »

user3939382 wrote: post #23
daenz wrote: post #7

>While we leave an in-depth empirical analysis of social and cultural biases to future work, our small scale internal assessments reveal several limitations that guide our decision not to release our model at this time. Some of the reasoning: >Preliminary assessment also suggests Imagen encodes several social biases and stereotypes, including an overall bias towards generating images of people with lighter skin tones…

Translation: we need to hand-tune this to not reflect reality but instead the world as we (Caucasian/Asian male American woke upper-middle class San Fransisco engineers) wish it to be. Maybe that's a nice thing, I wouldn't say their values are wrong but let's call a spade a spade.

Except "reality" in this case is just their biased training set. E.g. There's more non-white doctors and nurses in the world than white ones, yet their model would likely show an image of white person when you type in "doctor".

Re: Imagen, a text-to-image diffusion model

#67

Post by ceejayoz »

rvnx wrote: post #62
ceejayoz wrote: post #40

Earlier quoted context omitted.

"Reality" as defined by the available training set isn't necessarily reality. For example, Google's image search results pre-tweaking had some interesting thoughts on what constitutes a professional hairstyle, and that searches for "men" and "women" should only return light-skinned people: https://www.theguardian.com/technology/2016/apr/08/does-goog... Does that reflect reality? No. (I suspect there are also mostly u…

If your query was about hairstyle, why do you even look or care about the skin color ? Nowhere there is any precision for a preferred skin color in the query of th user. So it sorts and gives the most average examples based on the examples that were found on the internet. Essentially answering the query "SELECT * FROM `non-professional hairstyles` ORDER BY score DESC LIMIT 10". It's like if you search on Google "best…

> The algorithm is just ranking the top "non-professional hairstyle" in the most neutral way in its database

You're telling me those are all the most non-professional hairstyles available? That this is a reasonable assessment? That fairly standard, well-kept, work-appropriate curly black hair is roughly equivalent to the pink-haired, three-foot-wide hairstyle that's one of the only white people in the "unprofessional" search?

Each and everyone of them is less workplace appropriate than, say, http://www.7thavenuecostumes.com/pictures/750x950/P_CC_70594... ?

Re: Imagen, a text-to-image diffusion model

#68

Post by rcMgD2BwE72F »

ceeplusplus wrote: post #46
ceejayoz wrote: post #40

Earlier quoted context omitted.

"Reality" as defined by the available training set isn't necessarily reality. For example, Google's image search results pre-tweaking had some interesting thoughts on what constitutes a professional hairstyle, and that searches for "men" and "women" should only return light-skinned people: https://www.theguardian.com/technology/2016/apr/08/does-goog... Does that reflect reality? No. (I suspect there are also mostly u…

The reality is that hair styles on the left side of the image in the article are widely considered unprofessional in today's workplaces. That may seem egregiously wrong to you, but it is a truth of American and European society today. Should it be Google's job to rewrite reality?

In any case, Google will be writing their reality. Who picked the image sample for the ML to run on, if not Google? What's the problem with writing it again, then? They know their biases and want to act on it.

It's like blaming a friend for trying to phrase things nicely, and telling them to speak headlong with zero concern for others instead. Unless you believe anyone trying to do good is being hypocrite…

I, for one, like civility.

Re: Imagen, a text-to-image diffusion model

#69

Post by alphabetting »

daenz wrote: post #7

>While we leave an in-depth empirical analysis of social and cultural biases to future work, our small scale internal assessments reveal several limitations that guide our decision not to release our model at this time. Some of the reasoning: >Preliminary assessment also suggests Imagen encodes several social biases and stereotypes, including an overall bias towards generating images of people with lighter skin tones…

There is a contingent of AI activists who spend a ton of time on Twitter that would beat Google like a drum with help from the media if they put out something they deemed racist or biased.
Post reply on HN