Ada Lovelace, her Objection, Turing Tests and Universal Computing
Ada Lovelace Day falls on the 2nd Tuesday of each October. While it most broadly celebrates and promotes the achievements of all women in STEM fields, it particularly honours Ada Lovelace herself. Lovelace was a 19th-century mathematician who is considered to be the very first computer programmer and is likely the first person to envision the incredible potential of universal computation.
To understand the implications and possibilities of a modern invention it is necessary to have both the technical proficiency and expertise to correctly understand its workings, aligned with a creative ability to conceptually analyse its potential in the abstract, or even the metaphysical. To do this requires two sets of very different skills, to which few can claim to possess a level of mastery.
As a student of Digital Humanities, a discipline which focuses on the intersection between the use of computational tools and their application in traditional humanities like literature and philosophy, I am fascinated by Ada Lovelace’s role in the development of the design for the first universal computer and in particular her vision in predicting its potential. In honour of this worldwide event, here I have explored her work, her thoughts on the concepts of universality and computation, what Alan Turing referred to as Lady Lovelace’s Objection and the modern test for computer intelligence that bears her name.
The Lady of Number.
Lovelace (whose father was Lord Byron, the famous and scandalous romantic poet) was a colleague of the visionary inventor Charles Babbage, considered by many to be a father (or grandfather) of the modern computer system. He was well known in his own time as an accomplished mathematician, philosopher, inventor and mechanical engineer and he had great esteem for Lovelace’s mathematical abilities, referring to her as ‘The Lady of Number’. His Analytical Engine, first proposed in 1837, was a design for a general-purpose, programmable, steam-powered computing machine that incorporated arithmetic logic, conditional branching, loops and integrated memory, and would have used punch-cards and a printer to input and output data. Although it was never completed, Lovelace recognised that the design was revolutionary:
“The Analytical Machine does not occupy common ground with mere ‘calculating machines.’ It holds a position wholly it's own, and the considerations it suggests are more interesting in their nature… we mean any process which alters the mutual relation of two or more things, be this relation of what kind it may. This is the most general definition, and would include all subjects in the universe.”(my emphasis)
Today this would be referred to as Turing complete, or computationally universal (all modern computer systems are Turing complete); this was a totally unprecedented concept at the time.
Working closely with Babbage during the development of this machine (which pre-dates the first example of a working general computer by over a century), her knowledge of symbolic logic and her intimate familiarity with its workings allowed her to develop a deep understanding of its extraordinary capabilities, including processing algorithms and performing complex general computations. “The engine can arrange and combine its numerical quantities exactly as if they were letters or any other general symbols”, she wrote. Far from being merely a complicated calculator that could simply manipulate numbers, the engine could represent numbers as symbols or other abstract items and thus theoretically compute anything the programmer wanted. This insight is at the very heart of the theory of classical computation.
It appears that the Engine was so far ahead of its time that not even Babbage himself understood the conceptual potential of the machine in the way that Lovelace did. His original aim was to develop an efficient and reliable way to mechanically produce mathematical tables, (essentially a type of calculator). Lovelace’s work in developing a way to calculate a sequence of rational numbers (Bernoulli numbers) for the machine is often considered to be the first complete execution of a computer program, which she wrote “as an example of how an implicit function may be worked out by the engine, without having been worked out by human head & hands first”.
But she also understood its potential for universal applications: the machine could be programmed to do virtually anything:
“We might even invent laws for series or formulæ in an arbitrary manner, and set the engine to work upon them, and thus deduce numerical results which we might not otherwise have thought of obtaining.”
Lovelace had the ability to connect the logical and scientific with the abstract and metaphysical; a concept she may have referred to as ‘poetical science’ and in her writings about the machine she was able to explore the true implications of its potential. It was perhaps this particular talent that allowed her to see what Babbage could not. Her identity as the daughter of Lord Byron almost certainly had a bearing on her decision to shun poetry in favour of mathematics (a decision encouraged by her mother), yet there is undoubtedly a firm understanding of the poetic nature of science in her writings.
Sadly, before she ever got the chance to test her hypotheses and utilise the Engine’s capabilities, she died of cancer at the age of 36. Babbage suffered from a loss of focus (and funding) in later years and the Engine was never completed, and many of its innovations remained undiscovered until well into the twentieth century.
Lady Lovelace’s Objection (Note G):
While being the first to recognise the phenomenal potential of the Analytical Engine and universal computing, Lovelace noted what she saw as an intrinsic limitation: “The Analytical Engine has no pretensions to originate anything. It can do whatever we know how to order it to perform.” On the question of whether this singular machine could develop the capability to think on its own and simulate a human mind, she was unequivocal: “Only when computers originate things should they be believed to have minds”. This objection to the concept of a thinking machine on the basis of a capability to originate has continued to interest both computer scientists and behaviourist philosophers to this day. Lovelace believed that any valuable output of the Analytical Machine (and therefore any universal computer) was down to whoever programmed the device and not in any way to the machine itself (an overwhelming consensus amongst computer scientists up to very recently).
When Alan Turing, who founded modern computer science in the mid-twentieth century, discovered Lovelace’s notes in the course of his research and addressed the position (which he dubbed ‘Lady Lovelace’s Objection’) in the course of his own considerations of thinking machines.
Turing believed that any universal computer should in principle be capable of human-like thought (or artificial intelligence) on the grounds of universality (he recognised the Analytical Engine as a genuine universal computer). Any function of the human brain (a physical object), could be replicated by machinery and computer code. He defended the proposition in a 1950 paper in which he also proposed the famous Turing Test, designed to decipher whether or not a machine was capable of giving the impression of human-like cognition. Could a machine, the test posited, communicate convincingly enough with a person to convince a human judge that their conversation was with another human being? Turing re-framed the question of ‘Can machines think?’ (a question he considered so vague as to be meaningless), to a question of ‘Can machines give the appearance of thought?’
The test has often been cited as a benchmark by which we can measure technological advances in the area of AGI (artificial general intelligence) i.e. machines that think like humans. Each year the Loebner Prize awards a cash inducement to the most convincing AI computer program that participates in a Turing-style test, yet the top prize for a program that successfully fools the judges remains unclaimed. Turing was certain that his test would be defeated with only a small improvement on the hardware available at the time and certainly by the year 2000. It has proven to be more difficult than expected.
However, the test is often criticised for limiting the scope of investigation to communication or speech, which can in certain circumstances be convincingly imitated by modern chat-bots — programs that converse without demonstrating understanding or evidence of real thought. Indeed, most applications that have been specifically designed to beat the Turing Test are exercises in deception: they are preloaded with thousands of set responses, often gathered from masses of conversation data to simulate a genuine human interaction, rather than demonstrate cognitive capacity. The Turing Test is, in many ways, a poor test for general intelligence.
Turing overcame Lady Lovelace’s objection by simplifying it to a question of whether a machine was capable of surprise, which he maintained was certainly the case. In more recent years this response has been criticised as being ‘mysterious’ and superficial. A program might surprise you when it gives an unexpected answer, but if that is caused by a bug in the code or a hardware malfunction, it is hardly evidence of intelligence. A case of true originality (or creativity) is something much more sophisticated than that.
True creativity is not simply a case of coming up with an appropriate sentence in a given conversational situation; it is the origination of something novel and valuable (and according to some definitions it will also generate surprise).
In response to some of these criticisms and others, a number of different tests for human-like intelligence have been formulated, one of which is the Lovelace Test, proposed in 2001 in honour of Lady Ada. This test revisits her original objection — ‘Can a machine originate a valuable novelty’ -
1: which it hasn’t been specifically programmed to do,
2: where the process is repeatable and reproducible, and
3: can the programmer explain how the code managed it?
Computer programs are certainly capable of creating unexpected novelties, but that in itself is not evidence of human-like thought. If the novelty is created randomly without any deterministic input from the machine we are just witnessing creativity by brute force, or by fluke. A machine that is powerful and fast enough, which is allowed to run for long enough, will eventually generate something useful. This is not surprising, nor is it reliably reproducible. There is nothing to suggest that the machine will re-produce an equally impressive example given the same time and resources again.
If the programmer can successfully explain how the code produced the solution, then it has also failed the Lovelace test of intelligence. The creative output clearly lies within the code and therefore the credit belongs to the coder. This goes along with Lovelace’s assumption about the Analytical Machine. However, if they can’t arrive at a plausible solution, then the code has passed the test and should be said to be a truly creative program.
When Ada Lovelace was working on the first complete computer program in 1843 she could not have conceived of deep-learning algorithms, artificial neural networks, clustering, structured prediction or generative adversarial networks. These developments have dramatically changed our understanding of computers as thinking machines, and some researchers even believe that the Lovelace test has in fact already been defeated. Creative applications that generate art, compose music and play games like human beings, making decisions imperceptible to humans within the impenetrable ‘Black Boxes’ of their code, have altered our perception of what is possible for a machine to originate.
Many scientists and philosophers now believe that it is a case of when, not if, AGI will become a scientific reality. Yet it is remarkable that Lovelace’s observations on universal computing and artificial intelligence, which were based on a fully mechanical, steam-powered machine that transmitted information through rods instead of electrical wires and was never even finished, should remain so influential today. Her paper on the Analytical Engine is an incredible work; it would have been understood by very few of her contemporaries, yet remains relevant almost two centuries later.
For more information on the Ada Lovelace Day organisation and their work in promoting and encouraging the work of women in STEM, check out https://findingada.com/.
All views are my own and not shared by Oracle. Please feel free to connect with me on LinkedIn
