> Before the pernicious or the insidious, we of course begin with the deeply, hilariously stupid: from screenshots I’ve obtained, an insistence engineers no longer use phrases like “build ninja” (cultural appropriation), “nuke the old cache” (military metaphor), “sanity check” (disparages mental illness), or “dummy variable” (disparages disabilities). One engineer was “strongly encouraged” to use one of 15 different crazed pronoun combinations on his corporate bio (including “zie/hir,” “ey/em,” “xe/xem,” and “ve/vir”), which he did against his wishes for fear of retribution. Per a January 9 email, the Greyglers, an affinity group for people over 40, is changing its name because not all people over 40 have gray hair, thus constituting lack of “inclusivity” (Google has hired an external consultant to rename the group).<p>This is both hilarious and sad
If you want to skip all the culture wars stuff, the interesting bit here is the alleged description of how Gemini's image generation prompting is preprocessed:<p>"<i>Roughly, the “safety” architecture designed around image generation (slightly different than text) looks like this: a user makes a request for an image in the chat interface, which Gemini — once it realizes it’s being asked for a picture — sends on to a smaller LLM that exists specifically for rewriting prompts in keeping with the company’s thorough “diversity” mandates. This smaller LLM is trained with LoRA on synthetic data generated by another (third) LLM that uses Google’s full, pages-long diversity “preamble.” The second LLM then rephrases the question (say, “show me an auto mechanic” becomes “show me an Asian auto mechanic in overalls laughing, an African American female auto mechanic holding a wrench, a Native American auto mechanic with a hard hat” etc.), and sends it on to the diffusion model. The diffusion model checks to make sure the prompts don’t violate standard safety policy (things like self-harm, anything with children, images of real people), generates the images, checks the images again for violations of safety policy, and returns them to the user.<p>“Three entire models all kind of designed for adding diversity,” I asked one person close to the safety architecture. “It seems like that — diversity — is a huge, maybe even central part of the product. Like, in a way it is the product?”<p>“Yes,” he said, “we spend probably half of our engineering hours on this.”</i>"<p>If true (admittedly a big if), it torpedoes the suggestion made by many that Gemini's misbehavior was simply accidental teething problems that were missed in testing.
That does not bode well. When HR is the problem then you’ve got a serious problem. Very hard to get rid of problematic HR people because they obviously know all the tricks. And any sort of attempt at changing culture can also get derailed by them.<p>That leaves you with just override from above ie top management. And nobody sane is taking an even vaguely anti-DEI stance. So they seem a little stuck.<p>Won’t be the last major fk up we seen from them on this topic as a result.<p>> diversity architecture<p>Jikes
I’ve been saying this since DALL-E was released three years ago: image generation is going to be a can of worms, forever.<p>Google was clumsy here (in the same way that OpenAI was with DALL-E) but no amount of patches are going to fix the culture wars. Taking a text prompt and turning it into a picture is always going to be contentious. There will always be some amount of real bias and some amount of perceived bias on top of that.<p>And it’s incredibly cheap and easy for users to find real or perceived bias if they’re looking for it.
> Per a January 9 email, the Greyglers, an affinity group for people over 40, is changing its name because not all people over 40 have gray hair, thus constituting lack of “inclusivity” (Google has hired an external consultant to rename the group)<p>Oh boy.
>Per a January 9 email, the Greyglers, an affinity group for people over 40, is changing its name because not all people over 40 have gray hair, thus constituting lack of “inclusivity” (Google has hired an external consultant to rename the group).<p>I doubt firing the CEO will do much at this point, the rot is institutional.
Conclusion: "Google is sitting on an enormous amount of cash, but if the company does lose AI, and AI in turn eats search, it will lose its core function, and become obsolete. Talent will leave, and Google will be reduced to a giant, slowly shrinking pile of cash."
I found I could defeat the AI filtering if I added some positive adjectives.<p>Bad result:<p>Show me an Amish person eating chicken.<p>Better result:<p>Show me an awesome and amazing Amish person eating chicken.
Is this what people mean when they say "woke mind virus". I've never really understood the term.<p>My perception was that certain groups took the idea of diversity too far and made it into strict quotas. Certain speech is restricted. Promos went from being on merit to based on the skin color you were born into.
I opened this expecting to see legitimate criticisms about the culture problems at google, but I can't get past the excessive whining about wokeness. I think that pseudo-progressive policies at hyper-capitalist corporations can contribute to the types of problems we have seen in Gemini. However, conflating those issues with some employee complaining about being "strongly encouraged" to choose their pronouns, and being disgusted by the "15 different crazed pronoun combinations" made me tap out. I don't care to read some culture warrior's weird manifesto.
Was Gemini a "disaster"? That seems like too strong of a word. For the first time after trying Gemini 1.5 Pro, I feel like they are turning the ship around. Sure they haven't caught up to GPT-4 but it does not feel like a joke anymore. Does anyone else feel the same after trying 1.5 Pro?<p>Surely they could do the same for their image generation efforts?<p>Frankly are the "anti-woke" people really that fundamental to where the country is going? Do their complaints even matter all that much? On the US political side, they have been dead wrong in every political prediction since the 2016 election<p>1. 2018: record election year for the opposition<p>2. 2020: after a mess in handling covid the incumbent party is kicked out when most recent presdents get two terms<p>3. 2022: a supposed red wave never materializes.<p>4. 2024: ?<p>(This is not to mention all the underperformances for every "anti-woke/far right" candidate in special elections).<p>It sounds like they are just a loud minority.