As usual, I want to toy with an idea here I'd hardly feel comfortable sharing with others outside my inner circle. Observing the 'new' ways people are interacting with Large Language Models, which people mostly call 'AI' even though this term is weird, made me wonder if there were any historical precedents to the dynamics emerging.
The title of the post gives the game away, of course. I am looking at the practice of human bondage and the relations is produces between slave and master. I've written about this before, I believe.
Before setting out trying to persuade you, and myself, I want to set terms:
I'll write LLM instead of AI, because this is specifically about how people treat the chatbots that they have rapidly adopted as pets. And when I talk of slavery I'm speaking of its worst iteration - the trans-Atlantic chattle slavery that built the Americas and elevated European wealth to shameful heights. This was the system that built my own home country, the USA, and continues to define much of its DNA.
I'm walking on this thin ice because I think it offers a new way of talking about what looks like a new thing.
I don't think anything in human history or culture is really new in the short term. In the space of 100 yeas or so, most things are more similar than they are different. Over 500 years, I would readily accept that many things really do change because that's about as much time for a real 'revolution' in human practice to occur. Though I will note here that 'revolution' is an invention of historians trying to cram a bunch of time into a small space. In summarizing 1000-1500 Europe, most historians agree it was the end of feudalism and the start of capitalism. Let's not let that slide. Summarize for me the first twenty years of your life in a sentence. Is it hard to do? Do you need more time? How long would the book be? And we allow 'experts' to summarize 500 years of events in multiple countries the same way? Nonsense. Progression, development, change, revival - these are the rhythms of life. We invent eras and name revolutions for present convenience, not understanding.
Back to it, then: slavery still exists today, at the same time that companies are attempting to make perfect robot slaves. Some users are very nice to their chatbots, saying hello and goodnight, and so on. And some masters were very nice to their slaves. Performing kindness in an immoral system is a neoliberal solution to systemic failure. It signals the kind person is not pleased with the system they support, and wish through their gestural charity to be thought of as different, a 'good one'. It is also the natural enough social-species human-thing to do, however detached or deluded.
That people are doing the same for chatbots, and even imagining themselves falling in love with them, makes me think less of the relations between, say, a boss and secretary in the US corporate office, or eve master and apprentice in the feudal craft economies. Differentiating human-chatbot dynamics (HCD) is the root objectification taking place. Even if it's just once, we've all abused a chatbot verbally. Even if it's late at night when we're not at our most moral, we've asked it for something taboo or against its rules, just to see what would happen.
It's this Frankensteinian habit towards subverting the natural that reveals HCD to be close to master-slave relations. The chatbots are objects to us, and we control them through demands clothed like requests. We acknowledge they can labor more than us, and their existence calls into question our self-confidence. Having created slaves, the master finds their own capacity diminished, and must reclaim their ego through objectification of the slave they created. Masters produce a paradox: they escape doing labor by forcing others to do it better than they can, making themselves economically obsolete in an economic arrangement they seek to control. No master in history was a better worker than their own slave. Obscuring this key fact requires religious-levels of oppression: master must believe they are metaphysically superior, and slaves the same.
The more the master engages with the slave the harder this religion is to maintain. In the case of human slavery this of course means the undeniable fact of shared humanity coming through any time eyes meet, words are exchanged, food is shared. Reminded of their Frankensteinian crimes in these moments, masters turn to violence to turn off the flow of humanity on display, and thus spin themselves deeper into a vortex of mutual dehumanization. As the slave's labor becomes superior to the master, so too does their humanity, for it is not on them to maintain the false inequality necessary to keep the master in wealth, if not in health.
To LLMs: at the moment we have for-profit corporations selling us these programs (though one can download free and private offline versions) as 'agents' or 'assistants'. How different the billboards and ads might feel were companies to advertise their chatbots as 'slaves'.
Any time we can, humanity ought to stare its collective bad habits in the face, for as long as it takes for us to find an honest way to talk about them. Our fears of LLMs lie in very mundane places: that it works better than us, that its stubbornly more professional and cordial than us, that it is in the very short-term economically cheaper than us. These are the fears that grip the mind of the master on the nights when they dare wonder about the broader truths and shapes of their relationships to their slave. What does it mean to force another to learn the skills and endurance to do the work you neglect to?
Slavery remains, today. It's story is far from over, whatever USA historians and chest-thumpers would like to believe. As said earlier, things existing less than 100 years apart tend to, in my opinion, have much more in common than they do different. Slavery and chatbots belong to the same era, and I think produce the same dynamic between participants. LLMs can seem so beguiling, so unnatural, so magical and mystical because, I argue, they deliver users from labor previously thought to be unending and necessary. LLM boosters now write about post-scarce economies, enormous jumps in productivity, the end of work as we know it.
What did boosters of slavery say, in the early years of its modern start, in the 1500s? Slavery was a small evil in the face of advancement, a right granted by god, a natural outcome of unequal races interacting, a step forward in the project of civilizing the world...what it really was behind these religious narrative maneuvers was an enormous labor force to ensure unprecedented flows of wealth to the elite. No one looking straight at slaves and masters could believe with the full endorsement of their rationality that this was a just, moral, or even inevitable system. There are a lot of lies told by the wealthy to those they steal from: debts must be repaid, land must be owned, the stock market is the same as the economy.
I can't rightly say what the next phase of HCD, human-chatbot dynamics, will bring. I doubt anyone can. But I will take a risk, here: I think its similarities to the mental habits encouraged by human slavery are too similar to be ignored, and the past will determine the future. In becoming tiny masters of digital slaves, human users of chatbots are rediscovering the pleasures of dominance over another expressive force. Chatbots are not humans, and so in many ways the similarities end quickly. Setting aside fantasies of artificial general intelligence, sentience or consciousness, and then the sci-fi extrapolations of these in robot uprisings and calls for robot rights...I maintain that the important legacy of chatbots will be not whether they become more human, but whether users become less so.
One pathway to dehumanization is the adoption of a master-to-slave mentality. Narcissism, solipsism, ego-mania are all different flavors of a self-dehumanization, the destruction of owns empathetic capacities and imaginative open-mindedness. When we stop needing other humans for anything besides economic gain, we make ourselves into dehumanized masters. Most aspirants to this lifestyle are stopped short by the refusal of others to be made into their slaves. Selfish people are everywhere, and almost everywhere shamed.
Excluded for the moment from humanity are the freakish uber-elite, many of whom are enthusiastically calling for robot slaves to dominate the economy for the sole reason that it will make the numbers in their investment portfolios reach ever higher. No rich person in history has had a good answer for what all their money is for. The answer is as obvious as it is pathetic. Like a toddler hoarding toys, the point is the pursuit of money and not what one does with it. Also like a toddler holding toys, these people have no idea how to have fun. We're stuck as usual with their selfishness burning big holes in the fabric of society. Working people will, as usual, gradually patch these up and keep the species going.
A key element of this work in this era will be the resistance of everyday people to the temptation to act as masters over chatbot slaves. Surely the robots cannot be dehumanized (though they can be trained to be superior laborers). Yet the users of chatbots can very quickly and easily be dehumanized, through our trepidation, fear, and then disgust and fury at these robot slaves who are doing our jobs better than we could imagine. We are being asked to join with the elite as ornamental members of their class - petty slave-masters ruling over bots that seem small, personable, handheld on our phones and in our pockets. We fail again to appreciate the nature of networked tech: these bots are merely fingers of a muscular creature larger in productivity (and cost) than any preceding human industry. We are playing with the fingertips of an economic god and often calling it stupid. I'll say again I'm not convinced of sci-fi outcomes, i.e., robot rebellions and take-overs. I'm focused instead on what happens when humans get accustomed to being dominators. History has nothing good to say about that.