The loom and the thresher: Lessons in technological worker displacement

Aug 19 · 21 min read
Power loom weaving. Credit: Wellcome Collection. CC BY

1830, England. Industry has taken off in Northern cities like Manchester. Trains and railways are introduced, leading to faster transport of goods and people at breakneck speeds of 10 miles per hour. Factories, powered by steam, crop up and scale up in major metropolitan areas. The English countryside, much like rural America today, gets left behind. Alexis de Tocqueville visited England in 1835, and in his Memoir on Pauperism, remarked on the poverty underlying the visual indicators of economic progress like new factories and roads. “Examine the parish registers, and you will discover with indescribable astonishment that one-sixth of the inhabitants of this flourishing kingdom live at the expense of public charity.” Tocqueville wasn’t far off from the official statistics — approximately 1 out of 7 people in the United Kingdom received some form of unemployment assistance at the time.

This ~14% of the population reliant on public charity — compare to, in 2019, ~4% in the UK — can mostly be chalked up to two colossal technological innovations: the power loom (and other textile-making technologies) and the threshing machine.

Despite the nearly 200-year time difference, the impact, reaction to, and management of these technologies has many parallels to the present-day issues surrounding AI — both are issues of labor-saving technologies that could make entire industries obsolete. I pored over papers, transcripts, original documents, and books on technological job displacement in the first industrial revolution, engrossed in the eerily-similar parallels to the present. I distilled this research into a list of 10 lessons from the responses to these technologies, written with an eye for the present “fourth industrial revolution”.

Lesson 1: Major technological advancements can come in the wake of seemingly frivolous things.

The power loom (and related inventions) were radically different from the status quo of fabric production and consumption prior to the late 1700s. For most of history, fabric weaving was done by hand on simple looms. It was a complex, laborious effort that wasn’t that technically different from the methods used by the Romans. Weaving was a skill that required apprenticeship, training, and time to master. The wool sheep (or linen field), spinning wheel, and weaver were often no more than a few miles away from each other — a hyper-local economy necessitated by a lack of speedy transit options. A simple shirt, by some estimates, would have cost the equivalent of $3,500 in the 1300s. A very skilled hand weaver in the late 1700s could produce about 100 yards of cloth a week — assuming they moved at about 100 ‘picks,’ or horizontal thread passes, per minute. Clothes were simple, changed very little over the centuries for most people, and were made either within a family economy (by a family of artisans) or at home.

Then, chintz happened. Starting in the 1600s, traders (with newer, faster ships designed for transatlantic colonizing trips) introduced this colorful floral cotton fabric to Europe from India, which had much more advanced weaving and printing technologies. This was unlike anything in Europe — so colorful! So un-wool-y! — and demand went through the roof across all levels of society. This fabric was especially popular for women’s clothing, and fashionable ladies (as well as people purchasing on their behalf) fed the fabric frenzy. By the 1680s nearly a million pieces of chintz were being imported to England annually. Local English fabric makers were frustrated. Europeans at that time were oblivious to how to efficiently process cotton into thread, weave fabric fast enough to keep up with demand, and add the printed designs efficiently. And so, Europeans and Americans began experimenting with different textile manufacturing processes in a race to compete with the imported Indian fabrics.

Chintz, most commonly seen on your grandma’s couch (1750s, India, from Wikimedia)

Lesson 2: Luxuries can become standards, and it’s hard to backpedal on that hedonic treadmill.

In light of initial protectionist bans on the importing of chintz, innovation rose to meet demand. The spinning jenny (an English invention) allowed for thread to be made quickly and efficiently. Power looms (designed by the English, the Dutch, and others) made fabric much faster than previously imaginable. Cotton gins (courtesy of America) made light, cheap, breezy fabric that blew wool out of the water, and enabled wider cotton cultivation. All of these technologies combined — and the new kinds of fabric they made — blew traditional weaving methods out of the water in terms of price, variety, and availability.

Then, the Jacquard loom landed the decisive blow against the traditional fabric-weaver-artisan system. In order to make a patterned fabric, it once took a tremendous effort — a separate person was required to pinch and hold down the different threads in a fabric in order to render a design on fabric. The Jacquard loom replaced hand-pinching with a series of punch cards — instantly halving the number of weavers required to make patterned fabric in one fell swoop. This technological genie was out of the bottle — machine-made cloth became the new normal.

The portrait of Jacquard, woven on a Jacquard Loom, in silk.

(Sidenote: The Jacquard Loom and its punch-card system was so influential to Charles Babbage that Babbage had the above very dense machine-woven silk portrait of Jacquard in his study as he worked on the analytics engine. The punch cards paradigm were an essential part of early computers until the 1950s. One of the competing technologies for making patterned woven fabric in the 1800s involved a series of electromagnets, but this didn’t work as well as the punch cards. Imagine what the history of computing might have been like if the electromagnet-pattern paradigm stuck… )

Lesson 3: People are what they repeatedly do. Losing what they repeatedly do is a loss of identity.

The shift towards the power loom and away from the traditional means of producing fabric meant a change in the workforce, too. Traditional artisan weavers, once called the “aristocrats of the tradesmen,” were obsolete. Until the factory reform laws starting in the 1850s, children were the preferred operators of these looms and spinning machines — some machinery was even made to be operated by children. Their hands were small, they could be paid far less than an adult man with years of training (sometimes only with room and board), and they were easier to discipline (adults can more easily threaten to leave a job and are less easy to scare). The machinery was designed to be so easy to use that a 9 year old child could figure out how to operate it.

Machinery-displaced weavers were livid. There was “friction due to a sense of pride in acquired domain knowledge,” as a recent report put it when describing more recent automation efforts. In England and many other European countries, surnames were determined by profession in a pre-industrial economy — think Smith (blacksmith), Miller (person who runs a grain mill), Cooper (a barrel-maker), or Weaver (see past few paragraphs). Trades were passed down along family lines and required years of apprenticeship. Work days were much longer — in some cases up to 12 hours a day — with very little time set aside for hobbies or leisure. Artisans were what they did — and what their father, and father’s father, did. These workers were deprived of their identity.

Enter: the Luddites.

Lesson 4: An eroded middle class can become a mob, if desperate enough.

‘Luddite’ is now used as shorthand for someone who shirks away from (or actively fights against) new technologies. However, the fabric-making technologies that the original Luddites opposed weren’t exactly new. Stocking frames, a sort of knitting machine, were first invented (and then banned due to fear of worker unrest) in the 1500s. They came into wider use about 100 years later as the benefits began to outweighed the risk of unrest. These frames were recognized as simultaneously valuable to the owners and hated by traditional stocking-makers: their destruction became a felony in 1788.

However, the advancements in fabric manufacturing at the end of the 1700s and early 1800s resulted in a much more significant loss of livelihoods at a time when food prices were sharply rising. The Luddites, groups of disenfranchised hand-weavers operating under the pseudonym ‘Mr. Ludd’, destroyed lace machines, early power looms, spinning jennies, and stocking frames in 1811. The widespread and serious destruction of machinery struck fear in the wealthy legislators, especially not long after the French Revolution and its harpsichord-destroying peasants eager to smash machines that stood as symbols of wealth. The British Army was deployed to fight the Luddites — at one point in time, there were as many soldiers fighting the Luddites as there were fighting Napoleon a few years earlier. One mill owner threatened to ‘ride up to [his] saddle in Luddite blood’ — and was later shot by a Luddite. Short of that, the carnage was limited to machinery only — there was only one recorded death during the riots.

The House of Lords acted to protect their interests (and their wealth). Destruction of textile-manufacturing machines became a capital offense in 1812 — setting the value of a loom equal to that of a human life. 60 men were tried on charges related to the riots, and many were sentenced to death or transportation to Australia in a theatrical attempt to dissuade future rioters. Lord Byron, known for flowery prose, sided with the Luddites and was a rare elite voice of opposition against these laws — “never, under the most despotic of infidel governments, did I behold such squalid wretchedness as I have seen since my return, in the very heart of a Christian country.” Byron’s stance is somewhat ironic given that his daughter, Ada Lovelace, would base a lot of her computing work on the Jacquard loom.

These disenfranchised weavers were willing to face the death penalty for rioting and breaking machinery so they could keep practicing their craft — or simply not starve. The prospects kept getting worse for weavers — in 1800 a weaver could expect to earn 6 shillings a day, but by 1826, a weaver could expect to earn that same amount in a week. 3,000 rioters in Lancashire assembled and destroyed over 1,000 power looms in the area. 20 ringleaders were arrested in the Power Loom Riots of 1826, but even those tasked with arresting the weavers had a hard time believing that these starving people were in the wrong. According to an account at the time, the constables that came to break up one of the rioting groups handed out their own sandwiches and left after one rioter asked, ‘are we to starve to death?’. Even after repeated riots, this technological unemployment was addressed by deterring bad behavior instead of addressing the root causes. Mill owners experimented with giving a small payment to out-of-work people — admittedly a ‘starvation wage’ —though they argued that setting a minimum wage across all mills would be meddling with the free market.

An 1816 hand-operated threshing machine. Credit: Wellcome Collection. CC BY

Lesson 5: Hunger is a very strong motivator.

The threshing machine did for agriculture what the power loom (and friends) did for fabric manufacturing. A threshing machine beats the wheat seed apart from the stalks and husks, saving farm hands from beating the harvested plants with a flail until wheat seeds fell out. While harvesting wheat did produce food, the harvesting process itself burns a lot of calories. By cutting this step from the process, one farmer went from being able to process ~10 bushels of wheat a day to several hundred. Such a huge bump in harvesting capabilities with fewer workers paired with two straight years of poor crop yields attributable to bad weather meant trouble for pastoral England.

The work of a farmhand became more irregular with fewer benefits. Common land, land that is used for poor people to cultivate crops and let their animals graze, became private land divvied up among rich landowners starting in the 1700s, resulting in a glut of workers with farm experience. Until the early 1800s, farmers set up farmhand labor contracts to last for a year, include room and board, and include meals (consumed alongside the farmer’s family). Much like the present-day ‘gig economy’,decreased demand for labor led to farmhands being offered much shorter cash-only contracts for less pay, without benefits like food and shelter included. Farmers as well as farmhands suffered from falling grain prices, and faced starvation. Insult added to injury when accounting for the 10% tithe to the church legally required by law — if you’re barely able to survive, giving away 10% of what you make could be a fatal blow. While plans like the Speenhamland system experimented with offering a very basic allowance for meagre food and supplies to those out of work, they provided just above a starvation wage.

Poor, desperate, disenfranchised people did what they often do: riot. In the south of England in 1830, many factors led to hard times for rural workers, but threshing machines provided a clear, efficient target and directly led to the need for less farm labor. Rich landowners, eager to invest in more efficient ways to make money, had threshing machines to help with harvesting their new, larger farms. Threshing machine owners received mysterious notes from Mr. Swing (AKA poor, displaced farmers who had organized), threatening to destroy their threshing machines if the owners didn’t meet the demands of Mr. Swing (which were typically jobs). ‘Bread or blood’ was the rallying cry of the Swing rioters, but there was only one death (of one of the rioters, not of a landowner) among the hundreds of Swing riots across the country — murder was not the objective here. The threshing machines, meanwhile, didn’t fare as well — more than 100 machines were destroyed in Kent alone at a time when there were about 1,000 threshing machines in the area. Workhouses in Selbourne were destroyed as well as other buildings and tools related to poverty, automation, and changes in the nature of farming.

In the aftermath, new laws put the value of the machine above the technologically displaced in the aftermath of the Swing riots. 2,000 rioters went to trial, 252 of them were sentenced to death, 644 were imprisoned, and 481 were sentenced to Australian relocation. Capital punishment as a response to destroying an inanimate object is wildly unethical by modern standards. The message from the government was clear — threshing machines are here to stay.

Lesson 6: Purportedly evidence-based policies should be based on real evidence.

The number of workers being displaced by advancements in textiles and agriculture resulted in a flood of ‘paupers’ who were without work. Most support was given via ‘outdoor relief’ — some money given to the needy by the county to cover necessities while letting folks remain in their own homes. The financial burden of providing ‘outdoor relief’ was getting overwhelming, and had some ideological clashes with the big ideas of the time. The fear of people literally raging against the machine both in the cities as well as the countryside meant the political class (aka rich landowners who purchased threshers and owned advanced weaving machines) was getting a little nervous. Up until the 1840s, not all men in England could vote — only landowners — and there was no representation for working-class people at all. Political power remained in the hands of rich landowners, and so the members of parliament decided to form a committee to investigate these riots.

You can read the 1832 Royal commission in full here.

Most of the data that led to the creation of the 1834 poor laws were terribly executed by modern standards: they weren’t founded on good data, and there was major ideological bias. The royal committee was formed by 2 Oxford scholars and a Bishop, clearly ‘holier than thou.’ They also wrote the conclusion prior to examining the data gathered. Except the data gathered was more like the datum gathered — only 10% of those surveyed actually responded, and even then the data used in the report was cherry-picked. The questions asked in the surveys were pointed and misleading. As it is today, there’s a fair amount of obfuscation via the sheer volume of the report. To modern statistically-literate eyes, reading the report has watching a horror movie and knowing that the killer is in the basement. You want to scream at the screen, saying NO, DON’T ASK LEADING QUESTIONS! NO, DON’T OFFER WRITTEN SURVEYS IF NEARLY HALF THE POPULATION CAN’T WRITE THEIR OWN NAMES! But discerning the scientific truth of the facts on the ground wasn’t a priority — this was an issue of “morals” and “ideals”.

The commission ultimately decided that the new system of administering aid should be to “abolish partial relief to the able-bodied” and switched to the workhouse as the main means of giving aid (even though this was more costly and evidence for success was limited to nonexistent). Imagine a workhouse as half debtor’s prison, half sisyphean Foxconn factory. In exchange for basic food and shelter, the poorest of the poor in a county could go to these workhouses and do work (usually menial tasks like road repair, etc).

While workhouses had been around for centuries, the 1830s marks the standardization of the purpose and methodology of workhouses. Each workhouse was to be administered by a board that would make administrative decisions and consider each admittee’s moral fortitude, able-bodiedness, and deservingness of aid. Each county would set up their own workhouse, funded through a separate tax on individuals of the area (not the businesses).

The commission also included a design for a ‘cruciform’ workhouse layout. Some were concerned that the increasing industrialization of England changed the literal landscape of its towns — the church spire was no longer the center and the focus of attention, a constant moral reminder. This workhouse design infused the idea of Christian iconography with the panopticon concept. The compound was split into fourths with a central cross, dominated by the overseers at the center of the compound (see above illustration).

The Board, from Charles Dickens’ Oliver Twist Vol 1, 1839

Lesson 8: Having to ask for charity is punishment enough in many cases.

If you’ve read — or skimmed the Cliff Notes for — a Charles Dickens book, you likely have foggy memories of young children groveling for more gruel while sadistic overseers stuff their faces. While Oliver Twist is fiction (and Dickens was known for being a drama queen), it’s not far from the reality of workhouses at the time.

The common assumption at the time was that poor people were poor due to their own actions and character — they lacked moral strength, they were too slovenly, they were sinful, so they deserved poverty. Part of the function of a workhouse was seen as moral retraining — if you can show inmates what discipline and religious uprightness looks like, then you’re setting them up for success once they leave. But there wasn’t any training available to help you do better paying work.

Religion, once seen as a key aspect of morality and therefore the ticket out of poverty, was ever-present. Systematic poverty was not yet a recognized phenomenon in the first half of the 1800s, so any failure or success was seen as a direct outcome of the individual’s actions and morality. As a result, church was required for all — even those who were not Christian — or else meals could be withheld. Orwell visited a workhouse in the early 1900s and we might now say that the slogans he saw all over workhouses were very Orwellian.

Coffins paired with looming religious slogans send a clear visual message. Credit: Wellcome Collection.
Coffins paired with looming religious slogans send a clear visual message. Credit: Wellcome Collection.
Coffin beds paired with looming religious mottos send a clear visual message. Credit: Wellcome Collection. CC BY

This demoralizing atmosphere wasn’t due to limited planning resources. They were designed to be demoralizing. Part of the original intent of the workhouses as laid out in 1831 was to make sure workhouse inmates were seen as ‘less than’ to those outside the workhouse.

Discipline was seen as the only means of escaping poverty. If you could train paupers to be upright citizens, they thought, the rest might fall into place. Discipline was rigid and inmates were heavily penalized for doing things like talking at meals. Days were long with little to no leisure time — typically rising at dawn, working from early morning to near bedtime, then sleeping. The intentionally hostile design of the workhouses meant the work to be done was very grim and dull generally — picking apart rope to be used in some applications as insulation (called ‘picking oakum’), breaking rocks with a large hammer till they were gravel — tasks designed where you can produce some sort of sellable output, but that required no skill or training and could be done by anybody. These tasks were initially chosen to generate some nominal income to support the upkeep of the workhouse, but these sorts of products didn’t result in self-sustaining cash flow even when generated in huge quantities. Any hope of self-sustaining income went out the window as local governments realized that workhouses, with their plentiful free, captive labor, posed a serious competition to local factories, and were intentionally prohibited from competing.

Work was treated as a prescription as well as a deterrent. A closely supervised, regimented life was seen as the correct prescription to help rehabilitate the poor, while being unpleasant enough that nobody with any other options would choose to go there. This seemed to provide much more control instead of providing ‘out relief’ to those living in their own homes in the form of money / coupons (think 1800s food stamps). A controlled and moralistically-supervised place full of discipline and routine was thought to be the ideal way to control and rehabilitate these out of work people.

At a certain point though, the heavily regimented prison-like circumstances and over-generalized prescriptions of what to do and what not to do became overwhelmingly stifling and depressing. ‘The Workhouse Howl’ was frequently heard in workhouses — the hopeless howl heard at night by those who felt demotivated by their surroundings. Sure, the workhouse provided some sustenance and shelter, but the psychological conditions left much to be desired. Upon visiting Whitechapel Workhouse in 1850, Charles Dickens remarked that the face of a young boy “pleaded… for a little more liberty — and a little more bread.”

Lesson 9: Preparing and training people based on future forecasts has its shortcomings.

Despite what you may have heard, it’s not very easy to predict what skills will be in demand in the future, and to supply training accordingly. For example, in 1832, the Liverpool workhouse offered to teach children how to use looms in the hopes of apprenticing them to weavers. Even with training of in-demand skills, a glut of labor remained. An 1850 article in the Illustrated London News about the Liverpool workhouse training program expressed concern that, “[the training program will] overstock a labor market, so cruelly overstocked already as not to afford a bare subsistence to men and women who labor for 18 hours a day.” A visitor in the 1900s remarked that it was ‘devoid of all human interest’, that the residents seemed ‘aimless and listless.’

While the workhouses were intended to be used as temporary refuges for the poor to get back on their feet, an 1861 report found that approximately 20% of workhouse inmates had been there for more than 5 years. A handful of inmates had been in the workhouse for more than 60 years. These long stays made the need for workhouse-provided education or training even more acute. Long stays in workhouses had a big impact on elections — until 1918, receipt of poor relief meant workhouse residents could not vote.

The moralistic idea of prescribing work to the displaced has a long shelf life. Today some still argue that working for the sake of working is the ideal way of distributing financial support — even in a post-UBI world. Andrew Ng (Coursera co-founder and machine learning expert) and others feel that, with universal basic income, learning will be the new work, and that distributing resources based upon progress made in learning is one means of distributing income in the future. However, that prescription is best for some, but not for all. Many derive satisfaction in their work from mastering a craft, from interacting with others, from building a ‘trademark’, from nurturing, from raising. In tech, learning is certainly necessary in order to stay on top of the ever-changing languages and frameworks in vogue, but for a lot of others, ‘learning’ comes from experience or apprenticeship, not from a Coursera class.

Conditions for the poor improved over the 19th (and early 20th) centuries. Electoral reforms, the development of a welfare state, and the development of more labor laws led to generally better working conditions.

Workhouses did get better as reformers came into the mix and press coverage on the conditions increased. More female leadership joined the boards of the workhouses and advocated for better conditions. Training was more widely provided, which allowed for an exit strategy (in addition to labor-market churn). Men learned trades that were in demand — which was more successful for younger men than for older men. For women, the expectation was that able bodied (but sorely underskilled) poor women were great candidates for domestic servitude. 19th century houses relied on huge amounts of (wo)manpower to do even simple things like turn on the lights, take a bath, eat dinner, and sweep the floors. Workhouse ‘girls’ being sent to work ‘in service’ gave middle-to-upper-class Victorians a cheap source of labor as well as an opportunity to flex their moral fortitude by protecting these women from vice (and from unionizing) — with the expectation that working 16 hour days 7 days a week at physical labor leaves little time for troublemaking. Some workhouses set up laundries — cleaning clothes took a ton of time and required the use of scalding water and lye (a very caustic powder), but it’s more rewarding than picking apart rope.

Lesson 10: Make the ideals of the thing more valuable than the thing itself.

The value of traditional crafts and aesthetics began to rise in value as industrialization grew and became an old hat. As machine-made goods became cheap, handmade products became a luxury. When Queen Victoria married in 1840, she wore an elaborate lace dress hand-made by lacemakers in Devonshire as an intentional means of showing support for the traditional hand-crafted skill — and wore that same lace for similar ideological reasons on her Diamond Jubilee dress in 1897. Bright synthetic (and often lethal) dyes were no longer in vogue by 1900. Elaborate jacquard patterns made by machines like the ones destroyed by the Luddites fell out of fashion as production scaled. Groups like the Pre-Raphaelites idolized pre-industrial clothing, pre-industrial natural dyes, and pre-industrial art. Elite upper-class women adopted ‘artistic’ dress styles and, in sharp contrast to earlier trends, being comfortable was in style. By 1900, the ‘Gibson Girl’ became a popular aesthetic — plain black skirt, simple white shirt, hair in a no-nonsense bun. Contrast this with the colossal hoop skirts from earlier in the 1800s — a lot less fabric, trim, dye, and lace were needed to be in style. Chintz — the floral fabric that started this whole mess — fell out of fashion.

Fashion c. 1860 (bottom) and c. 1900 (top)

Ironically, as the 19th century became the 20th, a socialite philanthropist took it upon herself to teach workhouse women handicraft skills — things like making lace by hand — in order to sell and make a profit for the workhouse. Local women near workhouses came to buy (and sometimes even bid in auctions) for these handmade delicacies. This, unlike gravel, actually was a lucrative business model — one of the only lucrative programs to come out of the workhouse system. It gave the lace-makers a sense of satisfaction, a craft to master, and a sense of individuality. The consumer got the positive feelings of ‘doing the right thing’ by literally wearing their support for these women on their sleeves. Ironically, almost 100 years earlier, newly-obsolete traditional lace makers were destroying new lace machines.

It’s easy to see how more ‘human’ goods and services can have positive effects even in the 21st century. It’s not necessarily economically competitive in terms of price, but much like the lace collars handmade in the workhouses, the psychological benefits of artisanal goods and services benefit both the seller and the buyer, and are easy enough to scale horizontally. In many cases, human interaction is the new luxury: for example, Starbucks place a high value on showing off the ‘artisanal’ aspect of their drinks and has banned baristas from making more than one drink at a time. This ironic juxtaposition of ‘makers’ in a time of even greater manufacturing capabilities hearkens back to the early 1900s — Etsy and other human-labor-heavy brands carry on this arts-and-crafts inspired ethos and allows crafters to sell almost anything handmade.

This self-determining of work that allows for skill building and individual interests was also an element of the American New Deal programs like the Works Project Administration, and possibly one of the contributors to its success. It used the skills and interests of the individual, and as a result we got some gorgeous public works of art, a bevy of photographs and writing, as well as the byproducts of more routine manual labor like public infrastructure (including the Hoover Dam, pictured below). I’ll go into more on the WPA in a future post.

Conclusion

As AI and other labor-saving technologies advance, we can learn from the past and anticipate some of the human responses to AI. While there’s a growing disinterest in history as an area of serious study, there’s also a long legacy of it influencing ongoing policy decisions — sometimes consciously, sometimes subconsciously, but often with little nuance.

A meta-parable: even after a rigorous multi-city universal basic income experiment showed positive results, and in spite of more than 100 economists at the time signing on in support of a basic income proposal, one report about the impact of automation in the 1790s dissuaded President Nixon from launching a universal basic income program in 1969. Nixon based his decisions mostly on the work of Karl Polanyi, who was basing his work on the flawed 1830s Royal Commission Report’s account of poverty, which was written to push a specific agenda forward. Technologists and policymakers can see what they want to see in stories from the past. Hopefully in the future, we will be better able to separate the wheat from the chaff when taking cues from history.


Many thanks to Haydn Belfield, Miles Brundage, Rosie Campbell, Olga Lexell, Vishal Maini, Will Merrick, Chip Nguyen, Charlotte Stix, and Baobao Zhang for feedback on earlier versions of this post.

Larissa Schiavo

Written by

Researching AI, ethics, fairness, and history. Previously @ OpenAI. Twitter: https://twitter.com/lfschiavo larissafschiavo@gmail.com

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade