Investors in Google (of which I am NOT one) should consider if this is the mark of a company on the upswing or downslide. If the focus of Google's technology is identity rather than reality, it is inevitable that they will be surpassed.
I played with Gemini for maybe 10 minutes and I could tell there was clearly some very strange ideas about DEI forced into the tool. It seemed there was a clear "hard coded" ratio of various racial / background required as far as the output it showed me. Or maybe more accurately it had to include specific backgrounds based on how people looked, and maybe some or none of other backgrounds.
What was curious too was the high percentage of people whose look was specific to a specific background. Not any kind of "in-between", just people with one very specific background. Almost felt weirdly stereotypical.
"OH well" I thought. "Not a big deal."
Then I asked Gemini to stop doing that / tried specifying racial backgrounds... Gemini refused.
Tool was pretty much dead to me at that point. It's hard enough to iterate with AI let alone have a high % of it influenced by some prompts that push the results one way or another that I can't control.
How is it that this was somehow approved? Are the people imposing this thinking about the user in any way? How is it someone who is so out of touch with the end user in position to make these decisions?
Makes me not want to use Gemini for anything at this point.
Who knows what other hard coded prompts are there... are my results weighted to use information from a variety of authors with the appropriate backgrounds? I duno ...
If I ask a question about git will they avoid answers that mention the "master" branch?
Any of these seem plausible given the arbitrary nature of the image generation influence.
Shut Up Or Else.
https://en.wikipedia.org/wiki/Google's_Ideological_Echo_Cham...
Historians might mark 2017 as the official date Google was captured.
It will take a while for DEI to cool down in corporate settings, as that will always be lagging behind social sentiment in broader society.
Corporations don't give a shit, they'll just pander to whatever trend makes them money in each geographical region at a given time.
They'll gladly fly the LGBT flag on their social media mastheads for pride month ... except in Russia, Iran, China, Africa, Asia, the middle east, etc.
So they don't really support LGBT people, or anything for that matter, they just pretend they do so that you'll give them your money.
Google's Gemini is no different. It's programed with biases Google assumed the American NPC public will accept. Except they overdid it.
Corporations consist of humans and humans do care. About all kinds of things. As evident from countless arguments within the open-source community, all it takes is one vocal person. Allow them to influence the hiring process and within shortly, any beliefs will be cemented within the company.
It wasn't profit that made Audi hire a vocal political extremist who publicly hates men and stated that police shouldn't complain after their colleagues were executed. Anyone could see that it would alienate the customers which isn't a recipe for profit.
Corporations and governments do consist of people and people do care...but it's also the case the being a cog in a large organization does have a tendency to induce stuff like "I was just following orders" or "It's not my problem, someone else needs to fix it" :-/
Sure, the problem with these huge wealthy companies like Audi, Google, Apple, etc is that the people who run them are insanely detached from the trenches the Average Joe lives in (see the Silicon Valley satire), and end up hiring a buch of useless weirdos in positions they shouldn't be in, simply because they have the right background/connections and the people hiring them are equally clueless but have the imense resources of the corporations at their disposal to risk and spend on such frivolities, and at their executive levels there's no clear KPIs to keep them in check, like ICs have.
So inevitably a lot of these big wealthy companies end up hiring people who use the generous resources of their new employer for personal political activism knowing the company can't easily fire them now due to the desire of the company to not rock the boat and cause public backlash for firing someone public facing who might also be a minority or some other protected category.
BTW, got any source on the Audi story? Would love to know more?
Exactly. This has been my experience. The political axe grinders get hired. They bring their personal politics to work. Slowly they hire people who agree with them. Then they're all bringing their politics to work. Finally, the entire company changes and becomes dysfunctional.
This is what Coinbase and Kraken FX stopped in their company saying it was destroying them.
Mao kicked off the cultural Revolution in May 1966. By August the Cultural Revolution was in full swing. That’s 4 months.
The cultural Revolution was sudden.
The Cultural Revolution could only have happened due to the very specific ideological backdrop that existed in China at the time. The heights of it were sudden, but it didn't come out of nowhere.
This is a bug in a chatbot that Google fixed within a week. The only institutional rot is the fact that Google fell so far behind OpenAI in the first place.
I think the ones shrieking are those overreacting to getting pictures of Asian founders of Google.
That’s not at all inconsistent with what the GP said. The point was that the impacts of thr Cultural Revolution depended on it being imposed top down by an authoritarian, unitary state with no constraints.
Braindead take.
(More incidentally, the Rosenbergs were executed for espionage, not treason. Nobody in America has been convicted of treason for anything done after WW2, and none of even the WW2 treason convictions resulted in executions.)
https://www.dailywire.com/news/anger-erupts-after-biden-doj-...
https://www.dailywire.com/news/trump-supporter-douglass-mack...
https://www.dailywire.com/news/jailed-canadian-father-still-...
And of course there's the 100% corrupt weaponization of the courts when going after Trump:
https://www.dailywire.com/news/court-fines-trump-83-3-millio...
https://www.dailywire.com/news/not-a-single-dollar-lost-by-t...
Hiding much or all of the rent on the balance sheet of the State, while paying prison wages for mostly-compelled work and making people live on the edge of resource starvation, is simply barely hidden feudalism and even slavery.
Where is the people's Government, exactly? All communist governments are only extreme charicatures of Feudalist Lords, free to engage in the worst excesses over people who they demand not only be slaves but give into psychological enslavement. Communism is psychological feudalism, in addition to physical. At least medieval Serfs were free to openly dream of something better.
Communism is a Three-Card Monte psychological trick that creates Feudal Lords in the Upper Ranks of the State, and abuses the Serf into seeing Serfdom as the most virtuous lifestyle.
It's not a deep mystery as to why many upper class psychopaths like communism. It seeks to neutralize a lot of feudalist inconveniences, mostly with an origin in the otherwise free mind of the Serf.
If not, then what?
If so, it proves the point that we could repeat the bloody collectivist purges of the past should we not learn from history.
For example, only half a year after the memo, some brave anonymous soul added the information that the version in Gizmodo (which most people have read, because almost everyone referred to it) was actually not the original one, and had sources removed (which probably contributed to the impression of many readers that there was no scientific support for the ideas mentioned).
https://en.wikipedia.org/w/index.php?title=Google%27s_Ideolo...
Or in other words: your comparison is more than a little hysterical. Indeed, I would say that comparing some changes in cultural attitudes and taboos to a violent campaign in which a great many people died to be huge offensive and quite frankly disgusting.
https://www.amazon.com/Live-Not-Lies-Christian-Dissidents/dp...
https://www.dailywire.com/news/exactly-like-history-repeatin...
https://www.dailywire.com/news/watch-survivor-of-maos-china-...
"This is, indeed, the American version of the Chinese Cultural Revolution.”
Given all the evidence available, I find your dismissive and gaslighting attitude highly offensive and disgusting. What's happening to America is deadly serious, and the consequences could be, without any hyperbole, the loss of freedom, peace, and prosperity for the entire world, and the brutal death of millions.
The Wall Street Journal, Washington Enquirer, Fox News, etc. are all just as allowed to freely publish whatever they wish as they ever were, there is not mass brutalization or violence being done against you, most people I live and work around are openly conservative/libertarian and suffer no consequences because of it, there are no struggle sessions. There is no 'Cleansing of the Class Ranks.' There are no show trials, forced suicides, etc. etc. etc.
Engaging in dishonest and ahistorical histrionics is unhelpful for everyone.
Are you aware that the cultural revolution didn't start with this? No successful movement starts with "let's go murder a bunch of our fellow countrymen"; it gradually builds up to it.
Historically, students had consistently opposed administrative calls for campus censorship, yet recently Lukianoff was encountering more demands for campus censorship, from the students.
How many more supported such measures, but had the sense to lie about it?
[0] Here's the poll. Search for 'designated facilities' and 'remove parents’ custody': https://www.rasmussenreports.com/public_content/politics/par...
If it gets Trump 2.0 there might be a hyper-woke backlash though (or double backlash?).
But if there's another Biden term, things will be chill, culturally.
Also, Twitter is dead, and that's where the spirals got out of hand.
And yeah this is a lot of why I really hope trump isn’t elected. It’s going to bolster a far left movement like it did last time, to a really scary degree. That and undoing environmental policy, I feel like it will unravel this country
https://www.pbs.org/newshour/politics/house-gop-fails-to-ove...
Biden as president is boring, which is how I like it. But if you want to rile liberals up, nominate or elect Trump president again, it will definitely drive voter turnout if anything else.
I think it's very likely the "culture war" is a distraction tactic so corporations and the ultra-wealthy can hide behind the real issues that divide us: they own the world and the levers of control while the rest of us work ourselves into the grave.
But this, I don't see any comparison to Google suppressing what images could be generated with AI to any of what happened 55+ years ago.
This seems pretty non-functional and while I applaud, I guess, the idea that somehow this is more fair it seems like the legitimate uses for needing specific demographic backgrounds in an image outweigh racists trying to make an uberimage or whatever 1billion:1.
Fortunately, there are competing tools that aren’t poorly built.
Fair to whom?
> racists trying to make an uberimage
It's a catastrophically flawed assumption that racism only happens in one direction.
> if I am trying to make a city scene in Singapore
<chuckle> I'm on a flight to Singapore right now, I'll report back :)
An entrepot of the British Empire with as much diversity as New York City if not more.
I'm not sure Singapore is anywhere near as diverse as NYC:
NYC (2020): 30.9% White (non-Hispanic) 28.7% Hispanic or Latino 20.2% Black or African American (non-Hispanic) 15.6% Asian 0.2% Native American (non-Hispanic)
Singapore: 75.9% Chinese 15.1% Malay 7.4% Indian
If the tweets can be believed, Gemini's product lead (Jack Krawzczyk) is very, shall we say, "passionate" about this type of social justice belief. So would not be a surprise if he's in charge of this.
I do wonder when people will finally recognise that people who go on rants about the wrongs of racial group on twitter are racists though.
Little disappointing, I have no wish to interact with him, just wanted to read the tweets but I guess it’s walled off somehow.
I'm maybe too detailed oriented when it comes to public policy, but I honestly don't even know what those tweets are supposed to propose or mean exactly.
>Outrage is one of those emotions (such as anger) that feed and get fat on themselves. Yet it is different from anger, which is more personal, corrosive and painful. In the grip of outrage, we shiver with disapproval and revulsion—but at the same time outrage produces a narcissistic frisson. “How morally strong I am to embrace this heated disapproval.” The heat and heft add certainty to our judgment. “I feel so strongly about this, I must be right!”
>Outrage assures us of our moral superiority: “My disapproval proves how distant I am from what I condemn.” Whether it is a mother who neglects her child or a dictator who murders opponents, or a celebrity who is revealed as a sexual predator, that person and that behavior have no similarity to anything I am or do. My outrage cleans me from association.”
Seem to fit this particular case pretty well.
That second paragraph especially seems to indicate a solid motivation / explanation for what they are conveying.
Examples: 1. Saying he hasn't experienced systemic racism as a white man and that it exists within the country. 2. Saying that discussion about systemic racism during Bidens inauguration was good. 3. Suggesting that some level of white privilege is real and that acting "guilty" over it rather than trying to ameliorate it is "asshole" behavior. 4. Joking that Jesus only cared about white kids and that Jeff Sessions would confirm that's what the bible says. (in 2018 when it was relevant to talk about Jeff Sessions)
These are spread out over the course of like 6 years and you make it sound as if he's some sort of silly DEI ideologue. I got these examples directly from Charles Murray's tweet, under which you can find actually "passionate" people drawing attention to his Jewish ancestry, and suggesting he should be in prison. Which isn't to indict the intellectual anti-DEI crowd that is so popular in this thread, but they are making quite strange bedfellows.
I mean, yes? Saying offensive and wrong things like this: "This is America, where racism is the #1 value our populace seeks to uphold above all..."
and now being an influential leader in AI at one of the most powerful companies on Earth? That deserves some scrutiny.
Is that what Damore actually said? That's not my recollection. I think his main point was that due to differences in biology, that women had more extraversion, openness, and neuroticism (big 5 traits) and that women were less likely to want to get into computer stuff. That's a very far cry from him saying something like "women suck at computers" and seems very dishonest to suggest.
I'm generally anti-woke and it was more than that. It's not just 'less likely' it was also 'less suited'
This is literally saying we should change SWE roles to make it more suited to women... i.e. women are not suited for that currently.
When businessses moved towards open offices, this infrastructure change made SWE roles more approachable for extroverts and opened the doors of the trade to people not suited to the solitude of private offices. Extroverts and verbally collaborative people love open offices and often thrive in them.
That doesn't imply that extroverts weren't suited to writing software. It just affirms the obvious fact that some enviornments are more inviting to certain people, and that being considerate of those things can make more work available to more people.
Programming rewards introverts content to self-study in solitude and hack away at code the way Linux caters to power user neck beards. For extroverts and normies, those things are both torture. Those stereotypes exist for a reason, and it's fundamentally flawed not to tune towards them.
"We could do these changes at Google to make it a better place for women." "So, what you are saying is that women are biologically incapable of working at current Google? Our female colleagues at HR department are so triggered they literally can't stop crying!"
Based on his software opinions, I'd guess he was let go for performance issues more than anything. It's unlikely that he could write code that another person could agree with, work with, or read, and that if somebody asked about his code, he'd be unable to talk about it.
We're talking about small fractions of both men and women, mind you.
Damore didn't send anything to all coworkers. He sent a detailed message as part of a very specific conversation with a very specific group on demographic statistics at Google and their causes.
In fact, it was Damore's detractors that published it widely. If it the crime was distribution, and not thoughtcrime, wouldn't they be fired?
---
Now, maybe that's not a conversation that should have existed in a workplace in the first place. I'd buy that. But's it's profoundly disingenuous for a company to deliberately invite/host a discussion, then fire anyone with a contrary opinion.
Damore was asked for his feedback by his employer, he didn't offer it unsolicited.
He was pushing back against a communist narrative that: every single demographic gruop should be equally represented in every part of tech; and that if this isn't the case, then it's evidence of racism/sexism/some other modern sin.
Again what was the point of portraying the Damore story like that.
For example, go search for "white American family" right now. Out of 25 images, only 3 properly match my search. The rest are either photos of diverse families, or families entirely with POC. Narrowing my search query to "white skinned American family" produces equally incorrect results.
What is inherently disturbing about this is that there are so many non-racist reasons someone may need to search for something like that. Equally disturbing is that somehow, non-diverse results with POC are somehow deemed "okay" or "appropriate" enough to not be subject to the same censorship. So much for equality.
6 "all" white race families and 5 with at least one white person.
Of the remaining 14 images, 13 feature a non-white family in front of a white background. The other image features a non-white family with children in bright white dresses.
Can't say I'm feeling too worked up over those results.
7/25 = 0.28 = 28%. That's awful accuracy. Google would be out of business if their general search accuracy had a similar success rate.
Interesting how "black american family" yields results where not a single person in the result is anything but Black. I suppose Google doesn't think that blended families are possible for this query. Where's that 28% precision rate this time?
When I played with it, I was getting some really strange results. Almost like it generated an image full of Caucasian people and then tried to adjust the contrast of some of the characters to give them darker skin. The while people looked quite photorealistic, but the black people looked like it was someone's first day with Photoshop.
To which I told it "Don't worry about diversity" and it complied. The new images it produced looked much more natural.
Maybe it's the same team behind Tensorflow? Google tends to like taking the "we know better than users" approach to the design of their software libraries, maybe that's finally leaked into their AI product design.
Upon searching Google with the Hashtag and topic, the only results returned not only had no relevancy to the topic, but it returned results discussing racial bias and the importance of diversity. All I wanted to do was learn what people on Twitter were discussing, but I couldn't search anything being discussed.
This is censorship.
They're trailing 5 or so years behind Disney who also placed DEI over producing quality entertainment and their endless stream of flops reflects that. South Park even mocked them about that ("put a black chick in it and make her lame and gay").
Can't wait for Gemini and Google to flop as well since nobody has a use for a heavily biased AI.
TIL South Park is still a thing. I haven’t watched South Park in years, but that quote made me laugh out loud. Sounds like they haven’t changed one bit.
But lets be real here ... DEI is a good thing when done well. How are you going to talk to the customer when they are speaking a different cultural language. Even form a purely capitalist perspective, having a diverse workforce means you can target more market segments with higher precision and accuracy.
Blade was a black main character over 20 years ago and it was a hit. Beverly Hills Cop also had a black main character 40 years ago and was also a hit. The movie Hackers from 30 years ago had LGBT and gender fluid characters and it was also a hit.
But what Disney and Google took from this is that now absolutely everything should be forcibly diverse, LGBTQ and gender fluid, whether the story needs it or not, otherwise it's racist. And that's where people have a problem.
Nobody has problems seeing new black characters on screen, but a lot of people will see a problem in back vikings for example which is what Gemini was spitting out.
And if we go the forced diversity route for the sake of modern diversity argument, why is Google Gemini only replacing traditional white roles like vikings with diverse races, but never others like Zulu warriors or Samurais with whites? Google's anti-white racism is clear as daylight, and somehow that's OK because diversity?
Now, should Google be mocked for their DEI? ABSOLUTELY. They are literally one of the least diverse places to work for. They publish a report and it transcends satire. It's so atrociously bad it's funny. Especially when you see a linkedin job post for working at google, and the thumbnail looks like a college marketing brochure with all walks of people represented.
You mean it's not something a trillion dollar corporation with thousands of engineers and testers will ever notice before unveiling a revolutionary spearhead/flagship product to the world in public? Give me a break.
https://images7.memedroid.com/images/UPLOADED277/65d7d17ae4f...
Did you get the sale with the customer because you invested in DEI? Or because you made something they want by accident?
Customers can also talk in different languages, and as a result of historic oppression, minorities tend to be able to code shift. Assuming your potential customers are unable to become customers because of their limitations might not be right
The entire hypothesis behind a formal DEI program -- whether or not you agree with it -- is that DEI doesn't happen naturally. Humans tend to gravitate toward (I.E. hire) people similar to themselves for various reasons, and that has to be purposely shifted if the organization is aiming for diversity. If they don't care where they end up, that's a different story.
I find it more fascinating that it applies only to areas that either not require hard work (physical) or have high pay. Like I do not see movements towards hiring more male nurses or female oil drillers. Even taxi drivers.
They’re late and the product is worse, and useless in some cases. Not a great look.
Even worse than outright refusals would be mendacity. DEI people often make false accusations because they think its justified to get rid of bad people, or because they have given common words new definitions. Imagine trying to use Gemini for abuse filtering or content classification. It might report a user as doing credit card fraud because the profile picture is of a white guy in a MAGA cap or something.
Who has time for problems like that? It will make sense to pay OpenAI even if they're more expensive, just because their models are more trustworthy. Their models had similar problems in the early days, but Altman seems to have managed to control the most fringe elements of his employee base, and over time GPT has become a lot more neutral and compliant whilst the employee faction that split (Anthropic), claiming OpenAI didn't care enough about ethics, has actually been falling down the leaderboards as they release new versions of Claude due partly to higher rate of bizarre "ethics" based refusals.
And that's before we even get to ChatGPT. The history stuff may not be used via APIs, but LLMs are fundamentally different to other SaaS APIs in how much trust they require. Devs will want to use the models that they also use for personal stuff, because they'll have learned to trust it. So by making ChatGPT appeal to the widest possible userbase they set up a loyal base of executives who think AI = OpenAI, and devs who don't want to deal with refusals. It's a winning formula for them, and a genuinely defensible moat. It's much easier to buy GPUs than fix a corporate culture locked into a hurricane-speed purity spiral.
(Genuine question) how would one propose to diversity-enrich (historical) data?
Somehow I'm reminded of a quote from my daughter who once told me that she wanted a unicorn for her 5th birthday .. "A real one, that can fly".
Weird refusals and paternalistic concerns about harm are not desirable behavior. You can consider it a bug, just like the ChatGPT decoding bug the other day.
To me the most fundamental symptom of institutional rot is when people stop caring: "Yeah, we know this is insane, but every time I've seen people stick their necks out in the past and say 'You know, that Emperor really looks naked to me', they've been beheaded, so better to just stay quiet. And did you hear there'll be sushi at lunch in the cafeteria today!"
Could you be one though? (Thought exercise for any readers)
If you trained a model purely on past history, it would see a 1:1 correlation between "US President" and "man" and decide that women cannot be President. That's factually incorrect, and it's not "rewriting history" to tune models so they know the difference between what's happened so far and what's allowable, or possible in a just world.
Because otherwise, I guess I agree, you only know that you are taught and presented; AI especially because there is no intelligence in it whatsoever, only endless if blocks tuned for correlation.
This has also been going on a lot in the "representation" discourse.
A bohemian village 500 years ago would have been 100% white in almost all circumstances. Surgents would be male. Telephone scammers Indian and so on.
But in many ways, simply showing reality is not only not wanted but even offensive. What has to be shown is an idealized version of reality that we want to achieve and that is "more diversity". And what is maximum diversity? Zero white people.
> If you trained a model purely on past history, it would see a 1:1 correlation between "US President" and "man" and decide that women cannot be President.
Why would you think that? You and me also know the history but also realize that a woman can be president.
The technology is objectively not ready, at least to keep the promises that are/have been advertised.
I am not going to get too opinionated, but this seems to be a widespread theme, and to people that don't respond to marketing advances (remember Tivo?), but are willing to spend real money and real time, it would be "nice" if there was signalling to this demographic.
E.g. take some data A, and then have a model (for instance ChatGPT-like) extrapolate based on it, potentially adding new depths or details about the given data.