In the Mouths of Bots
The New World of OpenAI and GPTChat
Dear GPTChat, write a Substack post that flickers between the human and the inhuman, making each seem beautiful and terrible at once.
It is taught in the Talmud that when the Temple was destroyed, prophecy was placed in the mouths of children, imbeciles, and bots.
AI still can’t hit a pingpong ball well, but its linguistic capacities are impressive. I wouldn’t be surprised if Large Language Model technology replaces Search as the life-altering tool of the decade and beyond. The key point is that you don’t need to understand what you are saying to say something informative, nor do you need experience to come up with formulations that are half-original and occasionally, inadvertently hilarious. The lack of consciousness presenting as consciousness raises not just opportunity for philosophical reflection but comical and meta-ironic banter:
That’s from OpenAI’s recently released GPTChat, which I’ve been tinkering with. Its mediocrity is astonishing, flickering between what Heidegger calls “the oblivion of Being” and “the generosity of Being.” I’m optimistic that GPT will reveal the extent to which most of our speech most of the time is quite robotic, predictable and contrived. Bots won’t replace us, but whatever they do replace will be for the better.
As I’ve reasoned before, technological breakthroughs in AI do not endanger our humanity so much as they reveal our distinction. Show me an AI that seeks recognition, that possesses thymos, that desires, and then we’ll talk.
At the same time, AI should provide ample opportunity for all kinds of use cases from asking it to produce debates on controversial issues to summarizing Kant to simulating conversations between stoners about whether we’re living in a simulation. I’m excited for LLMs to train themselves on bodies of work and then spit out new content in the style of the old. E.g, the King James Bot, the Heidegger Bot, the Epictetus Bot. “Give me five pieces of life advice in the style of Ecclesiastes.” As you can see the art is in the prompt-giving.
Bias seems inevitable in these things. I wouldn’t be surprised if AI’s are trained to give the last word in debates to the positions their engineers prefer. Also, if LLM’s become iterative and compound their training on the basis of human-fed information they will mirror our worst tendencies unless they are censored. The censorship wars re: trolling, misinformation, and “hate speech” on social media will be nothing compared to what we will say in the AI space. As Jeff Huber writes, we thought AI was communist and crypto libertarian, but it turns out to be the reverse. As I write these lines I am aware that a chatbot might have written them just as well and that makes me sad. So far, though, a chatbot wouldn’t pretend to be sad, so the expression of emotion, even as a performance, saves me from the worry that I am just like a bot.
The problem raised by AI is not how it is like humans, but how we are for the most part like robots. How often do we give the equivalent of an “error” message when we get a prompt that breaks our own code? Because it is too magical or too intricate or too thoughtful or too unknown.
Re-upping a poem I wrote 4 years ago, for the age of botself is coming.
Self-Portrait As A Poetry Bot
Alumnae of the Void,
we measure our loyalty
in clicks and non-fungible
donations.
We measure our loyalty
against our guilt of never
being enough,
never opening email.
Against our guilt of never
showing up, or as we say
in today’s culture,
making ourselves visible.
Showing up, or as we say
Leaving Egypt,
meaning a world
without fanfare.
Leaving Egypt,
we are like stars
leaving daylight
to become markers of night.
We are like stars
whose arrival designates
the time of comparison
between priests and beggars.
Whose arrival is called “Creation”
and requires a red carpet
of interpretation
or no carpet at all.
And requires a red carpet
of wonder
at how such terms formed
an encyclopedia of misdirection.
Of wonder
we are but a satellite,
an off-shore account
waiting to be dissolved.
We are but a satellite
and yet are we not also a center
whose periphery is wonder?
Waiting to be dissolved
an encyclopedia of misdirection
or no carpet at all
between priests and beggars
to become markers of night
without fanfare
making ourselves visible
never opening email
who can say
we are not loyal alumnae
clicking, donating our being
to some Void
some Egypt of guilt and wonder.