Contra Yishan: Google's Gemini issue is about racial obsession, not a Yudkowsky AI problem.
@yishan wrote a thoughtful thread:
Googleâs Gemini issue is not really about woke/DEI, and everyone who is obsessing over it has failed to notice the much, MUCH bigger problem that it represents.
[...]
If you have a woke/anti-woke axe to grind, kindly set it aside now for a few minutes so that you can hear the rest of what Iâm about to say, because itâs going to hit you from out of left field.
[...]
The important thing is how one of the largest and most capable AI organizations in the world tried to instruct its LLM to do something, and got a totally bonkers result they couldnât anticipate. What this means is that @ESYudkowsky has a very very strong point. It represents a very strong existence proof for the âinstrumental convergenceâ argument and the âpaperclip maximizerâ argument in practice.
See full thread at link.
Gemini's code is private and Google's PR flacks tell lies in public, so it's hard to prove anything. Still I think Yishan is wrong and the Gemini issue is about the boring old thing, not the new interesting thing, regardless of how tiresome and cliched it is, and I will try to explain why.
I think Google deliberately set out to blackwash their image generator, and did anticipate the image-generation result, but didn't anticipate the degree of hostile reaction from people who objected to the blackwashing.
Steven Moffat was a summary example of a blackwashing mindset when he remarked:
"We've kind of got to tell a lie. We'll go back into history and there will be black people where, historically, there wouldn't have been, and we won't dwell on that.
"We'll say, 'To hell with it, this is the imaginary, better version of the world. By believing in it, we'll summon it forth'."
Moffat was the subject of some controversy when he produced a Doctor Who episode (Thin Ice) featuring a visit to 1814 Britain that looked far less white than the historical record indicates that 1814 Britain was, and he had the Doctor claim in-character that history has been whitewashed.
This is an example that serious, professional, powerful people believe that blackwashing is a moral thing to do. When someone like Moffat says that a blackwashed history is better, and Google Gemini draws a blackwashed history, I think the obvious inference is that Google Gemini is staffed by Moffat-like people who anticipated this result, wanted this result, and deliberately worked to create this result.
The result is only "bonkers" to outsiders who did not want this result.
Yishan says:
It demonstrates quite conclusively that with all our current alignment work, that even at the level of our current LLMs, we are absolutely terrible at predicting how itâs going to execute an intended set of instructions.
No. It is not at all conclusive. "Gemini is staffed by Moffats who like blackwashing" is a simple alternate hypothesis that predicts the observed results. Random AI dysfunction or disalignment does not predict the specific forms that happened at Gemini.
One tester found that when he asked Gemini for "African Kings" it consistently returned all dark-skinned-black royalty despite the existence of lightskinned Mediterranean Africans such as Copts, but when he asked Gemini for "European Kings" it mixed up with some black people, yellow and redskins in regalia.
Gemini is not randomly off-target, nor accurate in one case and wrong in the other, it is specifically thumb-on-scale weighted away from whites and towards blacks.
If there's an alignment problem here, it's the alignment of the Gemini staff. "Woke" and "DEI" and "CRT" are some of the names for this problem, but the names attract flames and disputes over definition. Rather than argue names, I hear that Jack K. at Gemini is the sort of person who asserts "America, where racism is the #1 value our populace seeks to uphold above all".
He is delusional, and I think a good step to fixing Gemini would be to fire him and everyone who agrees with him. America is one of the least racist countries in the world, with so much screaming about racism partly because of widespread agreement that racism is a bad thing, which is what makes the accusation threatening. As Moldbug put it:
The logic of the witch hunter is simple. It has hardly changed since Matthew Hopkinsâ day. The first requirement is to invert the reality of power. Power at its most basic level is the power to harm or destroy other human beings. The obvious reality is that witch hunters gang up and destroy witches. Whereas witches are never, ever seen to gang up and destroy witch hunters.
In a country where anyone who speaks out against the witches is soon found dangling by his heels from an oak at midnight with his head shrunk to the size of a baseball, we wonât see a lot of witch-hunting and we know thereâs a serious witch problem. In a country where witch-hunting is a stable and lucrative career, and also an amateur pastime enjoyed by millions of hobbyists on the weekend, we know there are no real witches worth a damn.
But part of Jack's delusion, in turn, is a deliberate linguistic subversion by the left. Here I apologize for retreading culture war territory, but as far as I can determine it is true and relevant, and it being cliche does not make it less true.
US conservatives, generally, think "racism" is when you discriminate on race, and this is bad, and this should stop. This is the well established meaning of the word, and the meaning that progressives implicitly appeal to for moral weight.
US progressives have some of the same, but have also widespread slogans like "all white people are racist" (with academic motte-and-bailey switch to some excuse like "all complicit in and benefiting from a system of racism" when challenged)
and "only white people are racist" (again with motte-and-bailey to "racism is when institutional-structural privilege and power favors you" with a side of America-centrism, et cetera)
which combine to "racist" means "white" among progressives.
So for many US progressives, ending racism takes the form of eliminating whiteness and disfavoring whites and erasing white history and generally behaving the way Jack and friends made Gemini behave. (Supposedly. They've shut it down now and I'm late to the party, I can't verify these secondhand screenshots.)
Bringing in Yudkowsky's AI theories adds no predictive or explanatory power that I can see. Occam's Razor says to rule out AI alignment as a problem here. Gemini's behavior is sufficiently explained by common old-fashioned race-hate and bias, which there is evidence for on the Gemini team.
Poor Yudkowsky. I imagine he's having a really bad time now. Imagine working on "AI Safety" in the sense of not killing people, and then the Google "AI Safety" department turns out to be a race-hate department that pisses away your cause's goodwill.
---
I do not have a Twitter account. I do not intend to get a Twitter account, it seems like a trap best stayed out of. I am yelling into the void on my comment section. Any readers are free to send Yishan a link, a full copy of this, or remix and edit it to tweet at him in your own words.
Anya is live and ready to show you everything. Watch her strip, dance, and perform exclusive shows just for you. Interact in real-time and make your fantasies come true.
â Live Streamingâ Interactive Chatâ Private Showsâ HD Qualityâ Free Actions
Free to watch ⢠No registration required ⢠HD streaming
Two hundred thousand pounds a year to be Company Priest
Nice gig, if you can get it.
I go on sometimes about the Peace and Truce of Westphalia, the big European religious peace arrangement from 1648 where the Catholics would have Catholic countries, and the Protestants would have Protestant countries, and they had a mutual agreement to stop trying to compel each other to convert, which laid the foundation for later "secularism", whatever that is, because I have growing doubts about that word.
Individual countries (and some sub-country units) still had state religions and might even be theocratic, which is an important piece of understanding what kind of religious truce it was. Roman Catholicism was the rule in Austria, while Sweden had Lutheran Protestantism as its state religion, but the Peace of Westphalia established an arrangement where Austria would not try to invade Sweden to change Sweden's state religion, or vice versa.
The First Amendment operated similarly in America a century later. The individual states could have state churches, and several of them did, but the Federal government was supposed to stay out. This was a truce and a compromise to reduce fighting by having there be multiple small 'prizes' rather than an existential fight for the single big 'prize' of capturing the central government.
Later, there was a rise in "secularism", and a lot of very clever abstract thinkers looked at an ad hoc compromise signed in blood, and decided they wanted to make a generalized universal principle out of it. 'European countries and American states should be secular too! For that matter, individual companies also should be secular - enforced by state power if necessary! Abolish the state churches! No more religious discrimination in the workplace!'
(I feel like this is a repeating problem: abstract thinkers trying to make contingent agreements into universal principles and overreaching, but that's another essay.)
The major churches mostly agreed and complied with the truce, with some foot-dragging, and gradually settled down in the hope of a peaceful new era and doing missionary work by reason instead of the sword, which was a lot less risky for everyone involved. Even if it was tempting for a church to [re]capture the state and simply order everyone to praise Jesus while helping itself to tax money and organizational power.
And then this happens. A "Chief Equity, Diversity and Inclusion Officer" is a Company Priest for a sort of nontheistic religion that is capturing the state and breaking the truce. It names itself 'antidiscrimination' in order to cover the very discriminatory practice of requiring it be taught to nonbelievers in schools, funded by the state, enforced at companies, and demanding important people sign affirmations of faith commitments to diversity. It is not a formalized religion - formalization would break its cover - but it is a particular religion with beliefs, creeds, and moral demands (particularly racial quotas) that are not merely the neutral "secularism" of keeping Protestants and Catholics from fighting. Its restrictions on "hate speech" and "hate facts" amount to nontheistic blasphemy laws.
This is not a novel thought, but sometimes I see a striking illustration of it like the exchange above, and I'm torn:
Part of me says that maybe someone could patch up the edges of "secularism", perhaps expand the definition of "religion", name the Name-Eating Thing as a religion to be kept away from government, or ban ideology from government (lol) and do classical liberalism better by preventing it from decaying into...this. The free market is the best cure for bigotry; Diversity Officers are useless wastes at best and bigots themselves at worst.
Part of me says that classical liberalism was always going to lead here via modern liberalism and leftism, and I should shed my lingering qualms about supporting a coercive state church, because evidently there is no alternative. There will be Company Priests, and given that fact, it is best the Company Priests be Christian not heathen. What looked like growing secularism was in retrospect a transitional period, not a condition that can be sustained in its own right.
I have previous long essays and more in the drafts, but I want to have a summary I can refer to here even if it's crude:
"secularism" is used about several different vague things: Westphalian truce, American 1A Federal truce, tolerant state ideology, state neutrality - and the last one looks increasingly fake to me.
"religion" was a sensible category at Westphalia, but it isn't anymore. The Westphalian-descended systems as in America (1A+14A) contributed to this by creating a security hole: to control the state, your ideology must be nontheistic. This splits religion-as-theism from religion-as-moral-order and induces Culte de la Raison shenanigans. People saying "x is a religion" are a symptom of this.
a rule that tells us to âkeep Mithra out of the schoolsâ is overspecified, unless you think Mithra in specific is the great danger to impressionable young minds. If we keep Mithra out of the schools but we say nothing about Baal, Baal will outcompete Mithra
but "keep gods out of the schools" or even "keep religion out of the schools" is also overspecified, but generalizing gets you "keep ideology out of the schools" and that's incoherent nonsense. schools will teach something other than maths, and states will impose something by law, that's what states are. there is no getting rid of states and there is no getting rid of state ideology, there is only playing wordgames as to why Heavily Armed Enforcement Organization Of Your Choice is technically not a state.
Insofar as I have an opinion on the latest flareup of the Middle East conflict, it's to point to Moldbug's question: Why does not Israel simply win? Israel is overwhelmingly more powerful than Palestine.
To summarize ten thousand words of moldbuggery, it's because America is backing Palestine.
This may sound implausible. You may think it's obvious that America is all-in for Israel. But that should imply an even greater force disparity in Israel's favor if that were true, where Israel trivially rolls over Palestine.
Do you ever despair of the state of the West? Do you think the decline is reversible?
I refuse to despair, and I think the decline is completely irreversible!
Quoth Moldbug:
Since there is no credible alternative to USG, its opponents have no Schelling point. Moscow could surrender to Washington. Washington has no one to surrender to. The East had a West; the West has no West. Thus, its only option is to live forever.
Nothing lasts literally forever, but I expect USG is going to hold on unusually long as it declines, and then collapse unusually hard.
Are you familiar with the phrase "Arsenal of Democracy"?
The term was originally a description of Detroit, the highly productive industrial center in WW2. Detroit produced a double-digit percent of America's wargear. (Exact numbers vary with how you measure.)
Detroit has declined a lot since then.
But Detroit can't finish up and collapse, because it's propped up by USG.
That, I think, is a useful miniature reference for how the whole thing is going to go: USG is committed to the Detroitification of the world, despite the benefits of Not Doing That. Every failing component is going to be propped up, painted over, propagandized about, ignored when inconvenient, and paid off with other people's money, until USG dies. Then a lot of things are going to collapse at once.
Not this year, and not next year, though. There's a lot of ruin in a nation, and, I repeat, no alternative to flee to. The white flight mostly stayed in the US, too.
Anya is live and ready to show you everything. Watch her strip, dance, and perform exclusive shows just for you. Interact in real-time and make your fantasies come true.
â Live Streamingâ Interactive Chatâ Private Showsâ HD Qualityâ Free Actions
Free to watch ⢠No registration required ⢠HD streaming
WATCH: Judge Amy Coney Barrett apologizes for her earlier comments referring to sexual âpreferenceâ instead of orientation: âI certainly didnât mean and would never mean to use a term that would cause any offense in the LGBTQ community.â https://nbcnews.to/3lLBEbMÂ
1. Identify a respected institution.
2. kill it.
3. gut it.
4. wear its carcass as a skin suit, while demanding respect.
-@iowahawkblog
(The full tweet in question adds #lefties, but Iâve increasingly come to feel that the left wasnât the monster itself, just the most prominent suit worn by the Skinsuit Monster at the time.)
Todayâs case of skinsuiting comes to us courtesy of the MIT Technology Review, which posted this dumb tweet:
Originally I was just going to share it with my friends on ephemeral media and have a small chuckle about how this was omnicide-bait and MIT Technology Review really should have thought out their phrasing better. No man, no problem, as Rybakov put it. (often attributed to Stalin)
Then I read the full article, âWhat Buddhism can do for AI ethicsâ, and had a feeling of [screaming internally] before falling into cynicism. Itâs not just the twit who tweeted omnicide-bait. The full article repeatedly tees up the case for omnicide and hardly seems to notice.
Buddhism proposes a way of thinking about ethics based on the assumption that all sentient beings want to avoid pain. Thus, the Buddha teaches that an action is good if it leads to freedom from suffering.
The Buddha also taught a way of thinking about how best for everyone to die and stay dead (not reincarnate), because all life on Earth would inevitably involve suffering. A substantial chunk of Buddhist meditation is oriented to becoming a zombie of sorts: walking dead, not feeling suffering because youâre not feeling life. If you gave an archetypal Buddhist a lasting killswitch for Earth, heâd press it. I contend this would be an important subject to cover in an article on AI ethics -- if you gave a fuck about Buddhist teachings rather than being a monster wearing Buddhism as a skinsuit, doubly so when youâre halfway to making the case for the killswitch yourself.
A Buddhist-inspired AI ethics would also understand that living by these principles requires self-cultivation. This means that those who are involved with AI should continuously train themselves to get closer to the goal of totally eliminating suffering.
A Buddhist-inspired AI ethics would also understand that the best way of totally eliminating suffering is, again, omnicide. This is a controversial implication that Soraj Hongladarom should wrestle with!
But itâs not merely the unfortunate implications of the writer which leads me to cry âSkinsuitâ. Itâs the way the author so blithely skims over these implications as though he never thought of them. Whether this was by knowing more about Buddhism and what it advocates, or knowing more about Yudkowskyism and what it warns about, or knowing more about pop culture and the constant Robot Rebellions in fiction looking to end human suffering by ending humanity, or knowing something about moral philosophy... Iâm sure there are other ways of noticing the problem here, Iâm an internet commenter and I thought of four on the spot.
I donât demand that he should have reached any particular conclusion. But if youâre a professor of philosophy, as Hongladarom supposedly is, and youâre going to write a longform article about the potential influence of Buddhism on AI ethics, I feel itâs grossly negligent not to even mention the possibility that this might imply or lead the AI towards a conclusion of omnicide.
But the Skinsuit Monster doesnât know or care. The Skinsuit Monster is content to just gut Buddhism and gut the MIT Technology Review for new skin to wear as it continues monstering. (Demonstrating, we might say. The etymology of monster is an attention-grabbing spectacle, just like the modern mass-protest.) And the bloody claw of the Skinsuit Monster sticks out here:
Using technology to discriminate against people, or to surveil and repress them, would clearly be unethical.
âDiscriminateâ is a bait and switch word. Means-testing is discrimination. Progressive taxation is discrimination. Age limits are discrimination. Take your pick; some discrimination is clearly ethical - and the switch is something like âdiscriminate unethicallyâ, which it is uselessly tautological to call unethical.
And here:
For any of this to be possible, companies and government agencies that develop or use AI must be accountable to the public. Accountability is also a Buddhist teaching, and in the context of AI ethics it requires effective legal and political mechanisms as well as judicial independence.
When the Skinsuit Monster says that such-and-such must be accountable, what it means is something like obedient. And when the Skinsuit Monster claims that some millennia-old philosophy happens to require public accountability and antidiscrimination and judicial independence too, you can see the whole suit tearing. This has fuck-all to do with Buddhism. This is someone using Buddhism as a meatpuppet to get a second vote for what he wants.
If you donât like the skinsuit analogy, try the framing from @raggedjackscarletâ:
thereâs more intellectual integrity in âI met Cthulhu and he told me the world should be ruled by zombie flamingos from Neptuneâ than in âI met Cthulhu and he told me to vote for Hillaryâ
Buddhism isnât exactly Cthulhu, but Buddhism is far weirder than MIT Technology Review gives it credit for. Which is extra ironic when Soraj Hongladarom, a professor at Chulalongkorn University in Bangkok, is playing the part of the put-upon foreigner trying to ensure his native traditions get a say in the face of the overwhelming West-
Many groups have discussed and proposed ethical guidelines for how AI should be developed or deployed: IEEE, a global professional organization for engineers, has issued a 280-page document on the subject (to which I contributed), and the European Union has published its own framework. The AI Ethics Guidelines Global Inventory has compiled more than 160 such guidelines from around the world.Unfortunately, most of these guidelines are developed by groups or organizations concentrated in North America and Europe: a survey published by social scientist Anna Jobin and her colleagues found 21 in the US, 19 in the EU, 13 in the UK, four in Japan, and one each from the United Arab Emirates, India, Singapore, and South Korea.Guidelines reflect the values of the people who issue them. That most AI ethics guidelines are being written in Western countries means that the field is dominated by Western values such as respect for autonomy and the rights of individuals, especially since the few guidelines issued in other countries mostly reflect those in the West.
-only to turn right around and endorse another flavor of Westernism, now in Thai skinsuit. Whereâs your flamingo, Hongladarom? I donât see you advocating a single uniquely Buddhist goal. Barely anything novel or interesting, let alone controversial. Your article might as well have been titled âHow Buddhism Can Be A Loyal Handmaid To Western Progressivism.â
AI ethics guidelines should draw on the rich diversity of thought from the worldâs many cultures to reflect a wider variety of traditions and ideas about how to approach ethical problems.
Gandhi is said to have answered âWhat do you think of Western civilization?â with the barb âI think that would be a good ideaâ. True or not, that exchange nicely sums up how I feel about Hongladarom here. Some diversity of thought would be a good idea, and Iâm not seeing any.
Finally, lemme back up to a phrase I skimmed past earlier:
Accountability is also a Buddhist teaching
This is false.
There is indeed a Buddhist teaching which is often translated to English as âresponsibilityâ or âaccountabilityâ in the headline summary. But the content of that teaching is along the lines of: âI own my actions, I bring consequences upon myself, I do things and I am responsible for what results...â
This is a far cry from Hongladaromâs call for accountability to the âpublicâ, with the public of course being interpreted through appropriate political mechanisms.
Further reading: Moldbugâs How Dawkins got pwned series, if you have a few days.
A bunch of the twitter responses are along the lines of âthatâs your job, you <expletive>, this is why people make fun of journalistsâ.
But Vox has run many pieces arguing that the President is incompetent and wrong and lying, as well as the case for abolishing the Supreme Court, the case for packing the Supreme Court, why the Senate is deeply problematic and undemocratic, why the Electoral College is the absolute worst, et cetera, et cetera. Vox has explainers and articles dedicated to âAbolish ICEâ, âAbolish DEAâ and for that matter âAbolish the Senateâ.
From a normie perspective, Vox has absolutely been pursuing the hypothesis.
If we charitably suppose that the staff writer at Vox is generally familiar with what Vox has published, the inference that follows is: Kelsey Piper doesnât think any of the abovementioned are âgovernmentâ. Hmmm. đ¤
-
(My speculative pattern-matching, hinted at by the rest of Kelseyâs thread: Kelsey is influenced by a certain school of thought that implicitly views the true and rightful government as being composed of the experts and scientists she looks up to, and the merely elected government is an unwelcome imposition on those beloved degree-holders.)
