The cycle that began with [YOU] on AI practices poetical science without always naming it. The discipline of holding the machine's output in one hand and its meaning in the other—neither collapsing into the hype that mistakes the weave for a mind nor into the contempt that mistakes the weave for nothing—is exactly the combination Lovelace modeled. Every chapter that asks what a capability means for human creativity, accountability, or attention is applying her method to machines she could not have imagined but would immediately have recognized as descendants of the engine she described.
The cycle's warning about fluency decorrelated from authority is a poetical-science diagnosis: it requires both technical understanding of how neural networks produce text and humanistic understanding of what authority has historically meant in human discourse. Neither alone produces the diagnosis. The engineer who knows the architecture does not automatically see the cultural consequence. The humanist who sees the cultural consequence does not automatically understand why it arises from the architecture. Poetical science is what holds the two in view simultaneously.
The phrase appears in Lovelace's private correspondence rather than in the published Notes, but it names the method the Notes demonstrate. Her background was unusual: a mathematician's training imposed by a mother who feared her father Lord Byron's poetic temperament, combined with a temperament that was, despite the training, recognizably her father's. She did not experience these as opposites to be resolved but as collaborators to be disciplined—the scientific rigor checking the imaginative reach, the imaginative reach seeing past what the rigor alone could notice.
The asymmetry of vision between Lovelace and Babbage is the clearest evidence that poetical science is a real method rather than a convenient label. Babbage knew the machine as only its inventor could—every wheel, every card, every mechanism. He saw it as a supreme calculator, a machine to free humanity from arithmetic drudgery. Lovelace, who could not have built it, saw it as a general symbol-processor that could act on music and language and any formalized domain. The difference was not knowledge of the mechanism but imagination about the idea. Poetical science is whatever produced that asymmetry.
The method has a clear danger, which Lovelace's own career illustrates: the imaginative reach can outrun the evidence and produce grandiosity rather than insight. Her plans for a “calculus of the nervous system” never materialized. She wrote of her own powers in terms that mixed genuine self-knowledge with unchecked romance. Poetical science requires both the poetical faculty and the scientific discipline that keeps it honest—the willingness to check the calculation even when the vision is beautiful.
The inseparability of mechanism and meaning. Poetical science rests on a foundational claim: that you cannot fully understand what a machine is without understanding what it means, because meaning—what the machine does to us, for us, against us—is part of what the machine is. The flowers are not in the loom; they are in the relation between the loom and the human eye. To study the loom without the eye, or the eye without the loom, is to miss the thing entirely. Applied to large language models: a purely technical account of how they produce tokens explains the mechanism; a purely cultural account of their effects explains the response; only both together explain what they are.
Imagination as a precision instrument. Poetical science is not imagination running free; it is imagination disciplined by rigor and aimed at a specific target—the meaning latent in a technical structure. Lovelace used it to see, in a machine of brass gears, the universality that Babbage had not noticed was there. The imaginative act was precise: she asked what it would mean for numbers to stand for anything, and she followed the answer wherever it led. This is the kind of imagination that produces not poetry but prophecy—accurate prediction about the future of a technical system based on understanding its logic more completely than its makers.
The accountability of the method. Because poetical science makes claims about meaning rather than only about mechanism, it is accountable to both kinds of test. The technical claims can be checked against the engineering; the humanistic claims can be checked against the human consequences. Lovelace modeled this accountability by catching Babbage's error in the Bernoulli program and by insisting, throughout the Notes, on distinguishing what the engine demonstrably could do from what she speculated it might. Her objection is a masterpiece of poetical science: a claim about meaning (the machine cannot originate) grounded in a technical analysis (it can only execute what is specified).
The warning for the AI age. The development of AI is dominated by people of formidable technical skill who are not always equipped to ask what the systems mean—and by humanistic critics who are not always equipped to understand how the systems work. The result is a familiar pattern: engineers who know exactly how the architecture produces its output and are regularly surprised by what it means for human attention, creativity, and self-understanding. Poetical science is the name for the capacity that bridges this gap, and Lovelace's Notes are the existence proof that it is achievable.