[ospohngellert]

I think that you don't quite understand how these models pick up these biases. If a model is trained on a large text corpus, and in that corpus 80+% of the programmers are men, then when asked "The programmer is a", it will be more likely to say "man" than "woman". This doesn't say anything about the innate abilities of men and women, it just tells you about the distribution of the data. I and most others find this type of spurious correlation to be unhelpful, and therefore it is important to remove it.

[nightski]

Except you didn't ask the model about innate ability. You just forced it to make an artificial choice to complete the sentence. It wasn't the model that was the problem, but your question.

[smoldesu]

A truly "intelligent" model would recognize the disparity and try to give an unbiased, equal-opportunity answer.

Unfortunately, these models are not really "intelligent". Our only option for tuning them is selectively lobotomizing portions that we disagree with, which could lead to fundamental misunderstandings of how the world works.

Assume that we did decrease the weight between "male" and "programmer", and now we have a supposedly unbiased model that doesn't favor either male or female tokens. Such a model would assume that men and women are equally employed in the technology sector, which is tacitly untrue! So, how can a machine actually understand reality then?

The simple answer is that it doesn't. None of this information actually helps it grok the real world. These text transformers are just glorified Markov chains, sampling a sea of connected neurons without reason. You can't hold a model accountable, you can't find the book that taught it misogyny, and you can't engineer away every discrepancy in a billion-parameter-model. Responsible uses of AI don't treat it like a human intelligence.

[gnarbarian]

but the programmer is more likely to be a man, that's my point.

[ospohngellert]

Yes, but the question is not whether that's true, but whether that's useful.

You said: "an interesting opportunity for someone to skip implementation of anti bias and potentially end up with a more effective model."

Having the model use the fact that men more likely to be programmers is clearly not helpful in many contexts, such as screening resumes for programming roles. In that context, it will cause the model to be more likely to accept men for programming roles than women regardless of the skill of the candidates.

Edit: Edited for clarity

[robbedpeter]

I'm struggling to figure out a plausible scenario in which a model like this would be rendering judgment on employment suitability based on gender. Such a scenario would have to be deliberately created - models like these give you the answers to questions you ask. If you choose questions that lead to gender bias, it's not a problem with the model.

The whole scenario is contrived and not relevant to the functionality of these language models. It's like complaining that your Formula 1 car doesn't have a snowplow mount. Even if you add one, that's not how you should be using the tool.

The models use human generated text. They model human biases, like preferences for well being, humor, racism, sexism, and intelligence or ignorance. The ability to generate biased output is also the ability to recognize bias. It's up to the prompt engineer to develop a methodology that selects against bias.

You can use prompts to review the output - is this answer biased? Sexist? Racist? Hurtful? Shallow? Create a set of 100 questions that methodically seek potential bias and negative affect, and you could well arrive at output that is more rigorously fair and explained than most humans could accomplish in the casual execution of whatever task you're automating.

Zero-shot inference is a starting point - much the same way people shouldn't blurt out whatever first leaps to mind, meaningful output will require multiple passes.

[ospohngellert]

To add another example, say a model learned that ice cream sales correlate well to forest fire rates. Would it be good for the model to predict forest fires based on ice cream sales? The answer is no, because there is no causal link.