"completely harmless if you're more than a few inches away"
This was in the "disassemble for scraps" tent. I do not think that's a particularly reasonable thing to assume.
95% of the time, someone unscrews the housing, sees the radiation symbol, behaves appropriately. 5% of the time it goes in a garage junk box, gets dropped, kid finds the shiny yellow disc and tries to open it.
As it was, this was hilarious, and perfectly on-brand for EMF. I'm glad there's not a small nuclear source sitting in some unknowing person's hackspace, I'm glad someone was watching out, and them wearing cute animal ears makes the story even better!
I actually think it somewhat proves the opposite. It's clear they've tacked on a "make images of a diverse set of people" in the setup prompt, rather than baking diversity in properly and with more nuance.
Tbh, I think I'd rather have diverse founding fathers by default (obvious problem, fixable by changing your prompt), than DALL-E's current default, which seems to just draw all white people. (I just requested a few images of doctors, nurses and patients, and an elven council - every last person of the 103 it drew was white).
I agree their phrasing could have been better. However in my country it typically takes 1-2 years to get a diagnosis and start treatment. It's even a few months to a year if you go private.
I can see this sort of thing being at least a stop-gap, and something that might help while someone's waiting to get access to the better treatment.
But wouldn't the experiences of an actual trans person be much more relevant to this discussion than our cis asses?
We're never going to have experiences of our own that can inform our judgement on this, so shouldn't we just listen to and trust trans people to work out how to best to make their way in the world?
That's a great resource (that links to even more great resources) - thanks for posting it.
I disagree that professional ethicists aren't needed in an organisation like Google though. Just as security should be lived and breathed by every developer, you still need some experts to consult when things get gnarly, light the way when you need many years of study to understand the concepts, and spot things that laypeople would miss.
The large language models Timnit was studying are a great example of this; people like her were (and are) needed to help the engineers understand the breadth and scope of the issues. Everyone in the org having their hearts in the right place doesn't replace expertise.
I used to have this view; a more precise miliatary has fewer civilian casualties, and shortens wars, limiting overall destruction.
However, my mind was changed by the idea that if killing people is easy, and there is little risk to the operators of the equipment, it will be done all the more. Countries will go to war much less overall, if there are higher stakes. I like there being high stakes.
He got fired for the other bits, the casually implying 40% of the workforce were, onr average, somehow inferior and less suited to their work than men, and for not taking it down but stirring the pot, when he realised that he'd wildly misjudged the reaction people would have to it.
"This leads to women generally having a harder time negotiating salary, asking for raises, speaking up, and leading."
Sure, there are statistical differences between genders. This is a total non-sequitur from that, with no citation in sight to try allay blame. This is just the 1950s calling with its stereotypes.
Women may be more agreeable, and more less assertive. However you had better provide fairly convincing proof that we are speaking up too little, rather than merely less than men do.
I think there are a lot of misconceptions about diversity programs, and their purpose. Let's use tech leadership as an example:
The problem they're solving shouldn't be "we need more women in tech, so let's lower the bar for them".
The problem they should be solving is "There are two candidates for this leadership position. One tends to use assertiveness and authority to get the job done. The other uses encouragement and persuasion. Who do you hire?". The default response is often to choose the person with the leadership style you've seen many times before, and thus the person who uses unfamiliar, equally effective techniques doesn't get a shot.
You sure as hell shouldn't hire me because I'm female. You should hire me because the set of techniques I employ to do my job are effective, even though they might be less familiar to you. Every tech lead has a different style, and that style has been tailored to that individual. Measure the outcome, not arbitrary characteristics that contribute to the outcome.
I agree - the discussion around this topic is very frustrating (more frustrating to me than the actual content of the manifesto by far), and everyone's trying to make it so black and white.
I think the current top comment has it right; it makes good points (there are statistical differences in traits across genders), it makes bad points (the wage gap...just doesn't work like that), and it contains some really bad non-sequitur conclusions (measure how a person leads on the outputs of their team, not how "assertive" or "technical" they are).
When it's a widely circulated internal document criticising the hiring practices behind 40% of the company's workforce, it's hardly surprising.
He put them in a difficult position. The brand and employee-esteem damage caused by tacitly condoning his manifesto by not firing him, vs just getting rid of one engineer. Easy maths. He's gone.
Yeah, I agree that on aggregate, men and women have subtly different tendencies.
The reason I bring it up though, is that I don't see or experience any particular bias around most of the the issues in the article. Most of the points raised really don't seem like gendered issues.
It's always alarming to read stories such as Susan Fowler's, that remind you sexism is still out there in force, in places. But it's great to read things like this, which implies at least a good chunk of fellow female devs are having the freedom to leave because their company doesn't care about software quality.
Can you pick out any problems in particular pointed out in this article, that are more frequently picked up on by women?
Most of the common ones seem to be a company or manager's words not matching their actions. The sample sizes I've experienced may be fairly small so far (very few female colleagues, as ever), but I've seen men and women leave in equal measure for that type of transgression.
Depends; the input data set could be massive, incorporating things like TV listings data (half-time kettle surges are a huge part of our power consumption profile), weather, publicly listed events, perhaps scraped from websites and so quite dirty.
I can see this transitioning quite easily from a 'we can sanitise this data ourselves' job to a 'screw it, let a massive neural net figure this out' one.
This was in the "disassemble for scraps" tent. I do not think that's a particularly reasonable thing to assume.
95% of the time, someone unscrews the housing, sees the radiation symbol, behaves appropriately. 5% of the time it goes in a garage junk box, gets dropped, kid finds the shiny yellow disc and tries to open it.
As it was, this was hilarious, and perfectly on-brand for EMF. I'm glad there's not a small nuclear source sitting in some unknowing person's hackspace, I'm glad someone was watching out, and them wearing cute animal ears makes the story even better!