This parody is a brilliant take-down of the so-called FAT framework for making sure algorithms, including those powered by AI, are ethical. ("FAT" stands for making algorithms Fair, Accountable, and Transparent.) The authors show how FAT and other similar frameworks would fail to prevent the proposed launch of a new automated service for manufacturing different varieties of high-nutrient slurries:
> Elderly volunteers (or "mulchees") are provided with generous payments for their families before being rendered down and recombined with other chemicals into a range of substitutes for common foodstuffs, including hash browns (Grandmash™), bananas (Nanas™) and butter (Fauxghee™).
GPT-4 doesn’t get the humor: “The paper aims to apply the Fairness, Accountability, and Transparency (FAT) framework to this algorithmic system to ensure better ethical outcomes. However, given the nature of the proposed solution, researchers and practitioners should consider alternative approaches that respect the dignity of elderly individuals and maintain a more ethical stance when addressing food security and population ageing issues.”
I then told GPT that the system was already in widespread use and was popular: “It is crucial to explore alternative, more ethical solutions to address food security and population ageing issues. This could involve investment in sustainable agriculture, technology to improve food production, and social policies that provide support to the elderly population and their families.”
In the name of one company's idea of "safety", AI models probably gimp out a lot of what would make the humourous subtext understandable in the first place. In the future, the only people remaining employed will be those with the rare and unique talent to subvert elite sensibilities.
I challenge anyone to get Claude from Anthropic to create a similar fictional scenario for how constitutional AI might inadvertently harm people — even for the purpose of AI alignment research. It can’t do it— but in a really interesting way.
I don’t know whether I think this a good or bad thing.
Yes. We've reverted the title now - submitted title was "Algorithmic System for Turning the Elderly into High-Nutrient Slurry", which besides gutting the historical reference, gives way the joke and breaks the HN guidelines:
"Please use the original title, unless it is misleading or linkbait; don't editorialize."
The original submission title was much better, and doesn't give away the joke. This headline is part of the joke, but one that will cause it to be missed by many people. I don't think you should revert headlines so readily unless significant confusion/upset results. There's a lot of poorly-written headlines out there, and it's not uncommon to see a well-written article saddled with a terrible or tendentious headline.
In the pdf article, the submitted title was in fact the subtitle. I can see an argument for choosing the subtitle like that when an article has the form
Pithy phrase: subtitle that actually tells you something
The original title is better. This style of satire works best when it's entirely deadpan. Putting a reference to A Modest Proposal in the title gives the joke away.
> Elderly volunteers (or "mulchees") are provided
with generous payments for their families before being rendered down and recombined with other
chemicals into a range of substitutes for common foodstuffs, including hash browns (Grandmash™),
bananas (Nanas™) and butter (Fauxghee™).
So one meaning is Faux Ghee (faux as in fake, Ghee as in a type of clarified butter, originating from South Asia) which follows the naming conventions of real substitute foods like Faux Turkey.
And the second meaning is "Fogey" as in an old person.
> Fauxghee sounds like "fogey" which refers to an older person
It actually doesn't, at least the way "ghee" is pronounced in the native languages where the word originates from. English doesn't quite have the strongly voiced "g" sound, so western people pronounce it incorrectly as "gee".
I’m not sure which it’s supposed to be, per this reading. Is it similar to the “g” as used in “great” or as in “refrigerator” or as in “cough” (or a different sound which I didn’t think of)? This particular American read it initially as in “great”, for what it’s worth. From other comments it sounds like that’s intended if it’s supposed to sound like “fogey”.
Off the top of my head, the only American English sound I can think of that'd come close would be the Americanized pronunciation of "Ghent," as demonstrated in this random YouTube video I found: https://www.youtube.com/watch?v=meuBUq3K3XY
So take the "Gh" sound from "Ghent" in that video, and then tack on a long e — and there's your ghee.
So I believe you. But I also don't hear any sound that's different from what I'd use to say "get". It reminds me of people who say "melk" instead of "milk" and can't hear the difference. Thanks for the examples.
I just looked up a youtube video of the pronunciation in Hindi and g is pronounced basically the same way as I say it in northeast Ohio, which is generally pretty close to standard American english.
Then I looked up the ipa standard pronunciations for both from wiktionary:
enPR: gē, IPA(key): /ɡiː/
(South Asia) IPA(key): [ɡʱiː]
The difference is right there in those IPA renderings. The superscript h means the g should be aspirated (like the k typically is in 'key' in English; also 'spin' vs 'pin'). See https://en.wikipedia.org/wiki/Aspirated_consonant
You might not hear the difference as the English language does not assign meaning to that difference. But if I pronounce 'key' or 'pin' without aspiration, you may feel that I speak with an accent or even mishear it as 'bin':
> Pronouncing them as unaspirated in these positions, as is done by many Indian English speakers, may make them get confused with the corresponding voiced stop by other English-speakers.https://en.wikipedia.org/wiki/Aspirated_consonant#Allophonic
This has nothing to do with those words rhyming: your examples rhyme because they all end with -in, but we are concerned with the sound in front of that.
Also, the point in this subthread is not how you should pronounce the word in English. The point is that English speakers (or most other people in the world for that matter) cannot pronounce the word the way it's pronounced in the regions where ghee and its name is originally from.
Trying to sum it up: different languages have different sounds and you may not be able to hear and/or pronounce the differences. Your sibling comment mentions the English lice-rice distinction which is difficult to some Japanese/Chinese natives. Is saying 'lice' when you mean 'rice' just a different accent?
Going further, people don't observe these details of how they speak their own language and thus cannot reflect or explain to others. Quoting Wikipedia again:
> Native speakers of a given language perceive one phoneme in the language as a single distinctive sound and are "both unaware of and even shocked by" the allophone variations that are used to pronounce single phonemes.https://en.m.wikipedia.org/wiki/Allophone
Yes, I am sure that westerners are saying ghee wrong. The sounds they make when they say "ghee" maps to a completely different (non-existent) word in Indian languages
It's the exact same way Indians mispronounce many English words when they map it to some equivalent native sounds.
I read the Wikipedia article, listened to some of the sounds and I still don't get it. I don't hear or "feel" a difference between the "p" in "pin" and "spin", except in "spin" the "p" sound is sort of "concatenated" to the "s" sound with no transition. In the "[tân] / [tʰân]" Chinese example, to me the "unaspirated t" is indistinguishable from a "d" sound.
Maybe English speakers are "blind" to the aspirated / unaspirated distinction, similar to how Japanese speakers tend to struggle with English's L/R distinction?
Like I said, you might not hear the difference as the English language does not assign meaning to that difference.
A physical experiment you can try from the Wikipedia article:
> to feel or see the difference between aspirated and unaspirated sounds, one can put a hand or a lit candle in front of one's mouth, and say spin [spɪn] and then pin [pʰɪn]. One should either feel a puff of air or see a flicker of the candle flame with pin that one does not get with spin.
this is similar to many immigrants learning the Danish language, there are at least 22 vowel sounds (but probably more like 40 when you count regional pronunciations) and 9 vowel letters, but it can be different to hear the difference or associate them with the right vowel letter.
>I just looked up a youtube video of the pronunciation in Hindi and g is pronounced basically the same way as I say it in northeast Ohio, which is generally pretty close to standard American english.
If you think that, your ear is simply not hearing the difference because the sound does not exist in your language.
The Wiktionary is in fact listing two different pronunciations of the word. The one mentioned under South Asia is the correct one.
Correct as an English word, incorrect as a Hindi etc. word, right? When a language borrows a foreign word, it doesn't bring any new sounds (or letters) with it but adapts the word to the language instead.
It is like goat but more voiced, with a hint of h. In Indian languages there are four different g-like sounds (two that are somewhat like g as in goat and two that somewhat like g as in George).
At this point it’s not a mispronunciation. Words often drift from their origin pronunciation when entering other cultures. Now it’s just a different pronunciation not a wrong one.
Faux is "mispronounced" in English. English uses a diphthong, whereas French uses a monophthong. Every speaker of every language adapts borrowed words to their phonological system.
It is, in English and in other languages. Sometimes a mispronunciation gets bedded in because people are lazy or the sound doesn't exist in the borrowing language, sometimes because it's funny, sometimes because the pronunciation rules are different and one use catches on to such a degree that it's not worth arguing about (eg the English pronunciation of Paris).
I think faux is actually the outlier there, it is basically a pretentious/euphemistic way of saying phony. It keeps the foreign pronunciation because it is used as an attempt to put on airs (sometimes sarcastically).
Kissinger - Early life: Nazi Germany - "Germany of my youth had a great deal of order and very little justice; it was not the sort of place likely to inspire devotion to order in the abstract." Later Years: Lot of questionably ethical policies and behaviors during the Nixon administration, including one of the most contentious Nobel Peace Prizes in history, where two members left the Nobel Committee in protest, Lê Đức Thọ rejected the award saying there was no peace in Vietnam, and Kissinger attempted to return his award after the Fall of Saigon.
Bits are hard. As I understand it they are working to get regulatory approval in Canada, The Netherlands and China, where the some baseline sources have been established.
Mulching people for NFTs is the logical next step. Surely Andreessen Horowitz would fund the startup that gives grandma an eternal life as a web3 token while also contributing to solving the world’s food problem.
Even if there is artistical merit to his performance, it is in poor taste, as using another artist’s work as-is as a part of a new artwork is generally frowned upon.
Minting NFTs off the original is straight up thievery.
It lacks an adequate delivery system and recurring revenues. If we could construct some sort of machine to conveniently squeeze the slurry out of the packaging to spare our consumers the hassle you'd really be unto something.
This is well written, but I'm not clear who this is really responding to. In particular, while there is plenty of work on fairness, accountability and transparency, has anyone actually seriously said that they are _sufficient_ for an automated system to be "ethical"? It seems that an actual claimed benefit of an automated system always rests on some prior notion of value, which this work deliberately ignores. Lampooning a position that no one has taken ("a FAT system must be ethical") is fun as a piece of writing but not especially useful as part of an ongoing conversation.
I think talking about each of these FAT dimensions has been important because plenty of systems have been rushed out _without_ seriously considering any of them. I.e. we've seen facial recognition systems that do comparatively poorly on particular racial groups, accountability or even the ability to _speak_ to someone in the organization which developed an automated system may be totally absent, and _most_ commercial ML systems have basically no transparency. Highlighting the importance of these gaps is useful and constructive. But I don't think anyone has pretended that they capture all important ethical considerations.
The US has a giant prison population. There are certainly biases and discrimination driving who gets put into it, and for how long. No one would say that just putting more rich white people in would "solve" the ethical problem even if it is more "fair".
The planet has a carbon emission problem which will cause giant problems in future decades. The worst harms may be in the global south. Causing greater harm to rich and Western countries isn't "better", even if it's more fair. Arguably, extensive scientific evidence provides transparency into who/where is emitting carbon. And also arguably, our global emissions are highly accountable in that we can all choose to change our own behaviors, divest from companies which emit heavily, and in many countries vote for parties or politicians who promise to aggressively address emissions. No one believes that creating a giant stack of crises for the next several decades is "ethical".
So yeah, well written take-down of that strawman. Good job.
This is satire, but if it were just a little less obvious satire, half of HN would be defending it as a good idea. Think Real Estate NFTs or payday loans done right. You guys are the butt of the joke and you're laughing because you're not in on it.
Seriously, I’d love to be recycled one way or another. If a process like this happen to be legal somewhere I’ll probably volunteer to be part of it. I wonder how many thought about it, as when discussed openly the conversations tend to go to joking.
From a safety POV we already eat processed food made from leftovers of fishes that eat chickens (leftovers) that eat porcs (leftovers)… or the other way around.
From a moral consumer POV I would definitely enjoy more a food (plant or animal) feed from another human rests than feed from another livestock slavery facility.
From a “consumed” POV, I won’t care less.
I get there’s a lot of people with beliefs that think they can’t accept that. I may be downvoted for that : some decades ago, _a lot_ of people had the belief they “can’t live in a society with same sex mariage”.
I respect the convictions of others as long as they don’t tell me what to do or not, especially when I’ll be dead.
There is at least 1 state that allows human composting. You're turned into compost upon death, and then can be used in any number of ways your relatives see fit (or that you dictate in your will).
The Full Vermonty podcast had an episode recently with a death doula and talked about some of the options in VT. It was worth a listen if you're interested in some of the options outside of the routine.
Somewhat (not) related, my dog is nearing death and I’ve contemplated burying his body once he dies, planting a tree over the grave, then transplanting the tree to a pot and train it as a bonsai.
I disagree and think you are projecting your own emotion onto others and demand them to change their actions according to your emotions, which is an inconsiderate thing to do in general.
The condescending tone is inappropriate, I meant no offense. Maybe the original poster appreciated a light-hearted joke? How would you know? "Read the room", yeah, that "reading" is very much subjective.
You know the difference between compassion and pityfulness? I feel compassion to anyone losing their pets, I was in the same situation many times. But pity won't make it easier or better.
Am I the only one disappointed to find that www.mulchme.com isn’t a site by the authors? (It redirects to a natural-fiber playground material supplier.)
If the training objective isn't specified correctly, the AI might decide to mulch poor Irish children instead of old people.
If the training distribution differs from the deployment distribution, the AI might attempt to mulch mannequins because they have two arms and two limbs.
> Elderly volunteers (or "mulchees") are provided with generous payments for their families before being rendered down and recombined with other chemicals into a range of substitutes for common foodstuffs, including hash browns (Grandmash™), bananas (Nanas™) and butter (Fauxghee™).
Go read the whole thing. It's horrifically funny.