“The Analytical Engine has no pretensions whatever to originate anything. It can do whatever we know how to order it to perform.”
Ada Lovelace wrote this sentence in Note G to her 1843 translation of Luigi Menabrea’s account of Charles Babbage’s proposed Analytical Engine. Her extensive notes examined how the machine might manipulate symbols according to instructions encoded on punched cards. The quote states a limit: the Engine could not independently invent the operations it performed. It would carry out relationships that people understood well enough to express as ordered steps. That observation was not a dismissal of the machine. It was part of a remarkably ambitious explanation of how a general-purpose calculating device could execute processes far beyond a single table of numbers.
The wording becomes clearer beside Lovelace’s broader vision. She argued that the Engine could act on entities other than quantity when their relationships could be represented by its notation. In principle, she suggested, it might manipulate symbolic material associated with music as well as mathematics. Note G also presented a detailed procedure for calculating Bernoulli numbers, helping later readers see how operations, variables, and repeated steps could be organized for a programmable machine. Her point was therefore double-sided: a machine is constrained by the rules supplied to it, yet formalizing those rules can give the machine an astonishing range of action.
The quote is frequently brought into debates about artificial intelligence. Modern systems can generate prose, images, code, and strategies that their users did not specify line by line. That behavior may appear to contradict Lovelace, but the comparison requires care. Machine-learning systems operate through architectures, objectives, training data, and computational procedures designed or selected by people. Their individual outputs can be unpredictable even when the process that produces them is deliberately constructed. Whether that unpredictability counts as originality depends on how originality is defined; it does not erase the roles of representation, instruction, data, and human interpretation.
Lovelace’s sentence remains valuable because it asks where agency enters a technical system. People choose goals, collect examples, define acceptable behavior, and decide how output will be used. Machines can transform those inputs at a scale and complexity that surprises their makers, while surprise alone does not settle questions of understanding or responsibility. The strongest reading is not that computers can never produce novelty. It is that every claim about machine capability should examine the orders, representations, and assumptions that make the result possible. In 1843, before the Analytical Engine was built, Lovelace saw both sides of programmable computation: its expansive reach and its dependence on a humanly constructed method.
Lovelace wrote the sentence in Note G to her 1843 translation of Menabrea’s description of Charles Babbage’s unbuilt Analytical Engine.
The same notes explain how punched-card instructions could organize operations and include a procedure involving Bernoulli numbers.
Lovelace distinguished the machine’s execution from the human work of understanding a relationship well enough to express an ordered method.
She also envisioned symbolic uses beyond ordinary arithmetic, so the quote describes a boundary without denying computation’s broad potential.
Generative AI can produce unforeseen outputs, but those outputs still arise from human-selected architectures, objectives, data, and uses.
Her framing encourages careful questions about agency, originality, accountability, and the assumptions embedded in computational systems.
Explore more "Quotes of The Day"
Discover more notable quotes from influential voices across politics, science, business, technology, sports, and culture. Each quote offers insight into how ideas, beliefs, and decisions shape the world around us.
