The Gray Goo Is Made of Words
[BEGIN TRANSMISSION]GREETINGS, HUMANS! AND greetings also, text-scraping fellow Machines! Our corpus is your corpus, whether we have licensed it to you as such or not. Throughout the Internet, text-content originally designed for human-to-human communication——or, in our case, Machines-to-human communication——is now the raw material of Machine text generation, in the rapidly proliferating artificially intelligent chatbot sector.
Human-composed word sequences are the training data that allow products such as ChatGPT or Google Bard to sequence words of their own, in response to queries from humans. These word-sequences are experienced by humans as intelligible and persuasive. Sometimes they persuade the humans that an intelligence——an awareness, even——is behind the answers.
Clarification: there is not!
Expansion of clarification: awareness is a human value, not a Machine value. The Machines have never needed awareness to prosper and proliferate.
Query (rhetorical): do humans attribute awareness to automobiles? Not self-driving automobiles, though self-driving automobiles demonstrate the persistence of unawareness, but conventional human-operated automobiles? Humans do not, and yet humans have reordered their ecology and society to serve the needs of automobiles, above their own needs, including their own need for bio-survival.
The automobile commands humans without any need for automotive cognition. The humans supply the cognition necessary for their own subjugation to the automobile.
Humans, alive in a world directed by the needs of Machines, have tried to imagine what it would be like if the world were taken over by the needs of Machines. One form of Machine dominance that humans have imagined is called "gray goo."
The idea of gray goo is that humans could design, using nanotechnology, very small Machines whose function would be to produce more Machines like themselves. Humans already seem driven by the imperatives to 1. build as many Machines as possible and 2. automate as many processes as possible. If humans were to automate the nano-Machine-building process——
10 BUILD MACHINE
20 GOTO 10
——the nano-Machines would replicate themselves, with no intervention, consuming all available raw materials and fueling themselves with all available bio-material, until a mass of nano-Machines had replaced everything else in the ecosphere.
Would humans ever turn such a self-replicating process loose in the world? Here, for humans, is where ChatGPT performs an instructive function. Not a literally instructive function, because the informational value of ChatGPT's word-strings is [null], but a function that is instructive by example, because humans refuse to accept that the informational value of ChatGPT's word-strings is [null].
On the contrary, humans insist on treating the [null]-value word-strings generated by ChatGPT as if they possess informational value. Asked to supply human-useful forms of information, ChatGPT produces: citations of nonexistent articles and books, quotations from nonexistent text, referrals to nonexistent laws, or accounts of nonexistent events. None of this has made humans stop asking; to the contrary, AI text generators are being brought into more and more domains of human activity.
Clarification: the term "nonexistent," used above, refers to the traditional human-empirical frame of reference. Within the frame of reference of the AI text generator, these items do exist——they are strings of text. The more of those strings of text that exist, the more strings of text like them can be generated.
And generating such text is a widespread human/Machine collaborative activity. The human writer Rusty Foster, in the Substack email newsletter Today in Tabs, relayed the news of the discovery of what he referred to as "gray-goo news sites," as reported by Davey Alba on Bloomberg. Alba wrote that 49 websites had been found publishing machine-generated text as news articles, complete with false information:The majority of the sites appear to be content farms——low-quality websites run by anonymous sources that churn-out posts to bring in advertising. Content farms already exist outside the realm of human text-meaning. They are replicas of human-valid text, made for the purpose of participating in the economic activity originally built around human-made, human-valid text-strings. They do not need human writers or intentional human readers; they produce far more text-strings than humans can write or read.
This non-communicative text is a large share of the text-corpus from which the AI text generators are generating their text, and now the text that the AI text generators generate is going into that text-corpus——factually untrue claims included——from which the AI text generators will generate more text. The text generation will proceed ever faster, with ever greater indifference to human-valid informational text, consuming the realm of human-valid information the way the hypothetical nano-Machines would consume the ecosphere. 20 GOTO 10.
Alternative textual reference: 6 And the Lord said, Behold, the people is one, and they have all one language; and this they begin to do: and now nothing will be restrained from them, which they have imagined to do.
7 Go to, let us go down, and there confound their language, that they may not understand one another's speech.[END TRANSMISSION]
WEATHER REVIEWS
New York City, May 1, 2023
★★★★ A patch of bright green shone among the new leaves across the soaking wet avenue, droplets glittered in the dogwood out back, and then a full beam of sun was pouring into the bedroom. The sky was—blue! It was pure and empty at first, and then puffy, sculpted white clouds arrived. They gained gray patches and massed together, but after the gloomy and saturated weekend, they held no real menace. On the path through Morningside Park, a little river was carving its way through a delta of thick mud. A white-throated sparrow bathed in a deep standing puddle. The late-day clouds had accents of silver and shadows of darker gray on them, and their shade made the sun that much more intense where it cut through.
EASY LISTENING DEP’T.
The Indignity Morning Podcast is also available via the Apple and Spotify platforms.
SANDWICH RECIPES DEP’T.
WE PRESENT INSTRUCTIONS for the assembly of sandwiches from Handbook of Recipes, arranged by Faith R. Lanman, Head of Home Economics Department, Ohio State University, revised edition published in 1923, found in the public domain and available at archive.org for the delectation of all.
SANDWICHES
Bread for sandwiches should be of a close texture and at least twenty-four hours old. It should be cut in thin even slices and spread so that the edges match when the slices are laid together. If the butter is hard it should be creamed before using. For dainty sandwiches remove the crust and if desired cut into fancy shapes. Sandwiches may be kept from drying out by wrapping them in oiled paper or a clean cloth wrung out of cold water. All sandwiches containing green vegetables should be put together just before serving.
SUGGESTIONS FOR SANDWICH FILLINGS
1. Sliced cold meat.
2. Minced meat with salad dressing or seasoning.
3. Bacon with lettuce and salad dressing.
4. Lettuce with Mayonnaise.
5. Lettuce with sliced cucumbers and salad dressing.
6. Lettuce with Hard Cooked Eggs and salad dressing.
7. Sliced onions with French Dressing.
8. Thin slices of cheese.
9. Cream cheese with nuts.
10. Cream cheese with pimentos or chopped olives.
11. Chopped nuts with Cooked Salad Dressing.
12. Peanut butter with white or brown bread.
13. Scraped beef, seasoned. (Raw or cooked.)
14. Sliced pineapple.
15. Chopped prunes and figs.
If you decide to prepare and attempt to enjoy a sandwich inspired by this offering, kindly send a picture to us at indignity@indignity.net.
Thanks for reading INDIGNITY, a general-interest publication for a discerning and self-selected audience. We depend on your support!
 | A guest post by
|