23 Aug 2023
Artificial Intelligence and Rare Books
Dear readers, the silence of this blog has lasted too long!
Since our latest exchanges, a lot of water has flowed under the bridge (notwithstanding the current drought in Europe). Yes, the dramatic days of sanitary confinement are long gone and The Love of Books in the Time of Covid-19, a section of this blog designed to make my (and your) isolation less painful, is already an old memory. But there is also the unparalleled frenzy of the modern world, which quickly replaces one anguish with another and which, after having sounded the tocsin in the four corners of the globe, has already moved on to many other things.
"Tempus fugit". One can wonder if it is time that is running away , or if it is not rather us.
As I speak, it is the so-called "Artificial Intelligence" (AI) that is in the news. The most famous of them, whose slightly robotic name seems to be feared to become as immortal as Plato's, Shakespeare's or Einstein's, has just been declared persona non grata in Italy - after having been banned from four countries well known for their unparalleled role as Usual Suspects in international politics: China, North Korea, Iran and Russia. Italy must be credited with courage - bordering on temerity - for joining such league. But it is not alone, in our democracies, in worrying about the progress of artificial intelligence. Almost at the same time, some well-known billionaires, and others less well known, have just signed a petition to demand the "temporary suspension" of artificial intelligence developments, on the grounds that it would threaten the balance of the world. One can only dream when one thinks of the place these gentlemen have taken in the flood of technologies that have purely and simply revolutionized our lives (and not always for the better) over the last thirty years.
This blog is not intended to be a militant forum and I hope that those of you who have a different perception of what I am talking about in the previous paragraph will forgive me for not being able to be their champion on this subject. Fortunately, my purpose is not to be polemical: I'd rather like to address a question on which I have a little more experience: the place of artificial intelligence in our world of bibliophiles.
The history of bibliographic science is that of a slow blossoming, allowing the passage from an almost indiscriminate list of books to the elaboration of extremely well-documented directories covering, in a more or less specific manner, a thousand aspects of old and rare books. Authors, themes, places and dates of printing, printing workshops, print runs and papers, illustrators, translators, bindings, origins, etc. The list is long, so vast is the universe of printed books since their origin.
For example, are you familiar with the National Union Catalogue? It is an extraordinary publishing project that lists all the books printed before 1956 that are in public and university preservation libraries in the United States. I remember when we used to consult it in the bookstore with a bulky, prehistoric microfiche reader that reproduced the entire seven hundred and fifty-four folio volumes of the printed edition.
Computer technology has made an invaluable contribution to this Benedictine effort to classify books. Today, thanks to the development of the Internet, not only can we consult many national union catalogs, offering a broader perspective than the National Union Catalog, but we can often search the contents of the books themselves. And now the irruption of artificial intelligence is expanding the possibilities induced by computing exponentially, bringing extremely significant advances to these research tools.
Indeed, character recognition, the detection of complex linguistic patterns and the shuffling of gigantic amounts of information, allow us today to make sensational discoveries that would have been impossible only a few years ago.
Two months ago, it was announced that an anonymous manuscript play kept in the National Library of Madrid was attributed to the great Spanish playwright Lope de Vega. This result was obtained with the help of several artificial intelligence tools, which were able to decipher the manuscript and compare it to their database of linguistic models.
At my humble level, I can only be pleased to see that the knowledge of old and rare books is acquiring new dimensions that open new doors. What can the opponents of artificial intelligence say about this? I find it a great pity, for my part, to think that with our conventional bibliographic tools this manuscript of Lope de Vega would still be sleeping, ignored by all, at the bottom of a reserve.
I have a little personal anecdote on the subject.
A few months ago, the Syndicat de la Librairie Ancienne et Moderne honored me with a "bookseller portrait". I was invited to describe my career path and my aspirations in the exciting world of rare book selling, and so I devoted a first paragraph to evoking my life with antiquarian books "since childhood".
Antiquarian books have accompanied my life since childhood. If the bookshop initially founded by my parents in 1969 in Paris was a store located on rue Gay-Lussac, they soon afterwards made the choice to work at home, which filled our successive houses (my parents having moved many times) with old bindings, brochures, bundles of documents and manuscript jumble of all kinds. This did not make me a bibliophile in short pants, for I was first a reader and my curiosity towards old books was only awakened in adulthood, but their silent presence at my side from an early age had the effect of establishing a kind of natural familiarity between us. Continuing my life among books was neither a choice nor a vocation, but rather what I would call "a way of being".
So what does this have to do with artificial intelligence, you might ask? Well, here it is:
I was recently alerted by Google that a rare book kept at the Bibliothèque Municipale de Lyon was associated with my name in the Google Books records. As you know, this powerful company has undertaken to digitize a large number of books kept in public collections around the world, making their content available to the public. Thanks to a very advanced technology of character recognition, it also allows to index the content of these books, and even the handwritten inscriptions they contain.
The picture below was taken from Google Books, which spotted my signature on one of the endpapers of this book. Did the artificial intelligence of Google Books guess that I was just a child when I wrote my name in pencil on this ancient book? Does it have an imagination? Can it see me as I do? Sitting on the floor, sticking out my tongue and writing my name on an ancient book borrowed from my parents (sacrilege!), an ancient book that my parents would later sell, without realizing my misdeed, to the Lyon Municipal Library. Its curator at the time, Mr. Parguez, was one of their most faithful customers... Can artificial intelligence tell such a story? I can't help but doubt it.
What I can tell you without doubting it for a single second, in any case, (paraphrasing Guillaumet rescued from the Andes, for those who have read Saint-Exupéry), is that the emotion I felt while discovering this clumsily written line... no machine will ever be able to feel it.
And I can't help but smile when I think that being in my fifties, after more than thirty years in the business and thousands of rare books whose paths have passed through my hands, Google Books associates me with only one ancient book, the one on which I wrote my name when I was barely five years old! How ironic...
What about you? Artificial intelligence and rare books, what do you think?
A few links :
An anonymous manuscript theatre play ascribed to Lope de Vega
SLAM's bookseller portrait
The Google Books digitized copy of the Library of Lyon
BLOG COMMENT
Rare books, even when I was a child!
Ciao and congratulations!
Leandro
Ciao and congratulations!
Leandro
An extraordinary anecdote to conclude a splendid and thought-provoking essay. What is most fascinating, from an AI perspective, is not that Google Books found Julien's childhood autograph, but that they intuited a connection, and then acted on this by alerting Julien to their discovery.
---Bill C.
---Bill C.
As a matter of fact Google was more obedient than intuitive since my gg account is set to alert me when my name is quoted online... but maybe intuition is the next step?
It was, I think, the French Jesuit paleontologist and philosopher Father Pierre Teilhard de Chardin who, in his The Phenomenon of Man, first imagined the global spread of communication and its ultimate interconnection of all of us. He called this the Noösphere. A consciousness encompassing the whole Earth. It seems likely that we are arriving at that point.
A very primitive form of such a thing already manifests itself in the field of rare books: I can (without much more effort than to simple think it) consult the Internet (proto-Noösphere?) and find available copies, library copies, commercial availability, etc., of almost any book. There are still some books hidden away -- but they are quickly being found and subsumed into the global awareness.
The question of AI is aptly raised by Julien. We are only now seeing what it possible. What will AI be capable of when it eventually has access to all the books that have been digitized? It will "read" them all (recognize the characters, learn the words, the sentences) an gain command of their contents. Imagine the connections it will be able to make, the influences it will ferret out, the lines of influence it will identify! It is the identification of Julien's autograph but writ large. I cordially dislike soi-disant "futurists" and their predictions -- but this seems inevitable. It will be the golden age of book collecting, or its death.
Apropos of this issue (and I mean "issue" not "problem") is a book reviewed this weekend in the Wall Street Journal. A gloomy view, but a very interesting read. And the book appears to worthy of reading. Here is a link to a PDF of the review:
The Gutenberg Parenthesis’
A very primitive form of such a thing already manifests itself in the field of rare books: I can (without much more effort than to simple think it) consult the Internet (proto-Noösphere?) and find available copies, library copies, commercial availability, etc., of almost any book. There are still some books hidden away -- but they are quickly being found and subsumed into the global awareness.
The question of AI is aptly raised by Julien. We are only now seeing what it possible. What will AI be capable of when it eventually has access to all the books that have been digitized? It will "read" them all (recognize the characters, learn the words, the sentences) an gain command of their contents. Imagine the connections it will be able to make, the influences it will ferret out, the lines of influence it will identify! It is the identification of Julien's autograph but writ large. I cordially dislike soi-disant "futurists" and their predictions -- but this seems inevitable. It will be the golden age of book collecting, or its death.
Apropos of this issue (and I mean "issue" not "problem") is a book reviewed this weekend in the Wall Street Journal. A gloomy view, but a very interesting read. And the book appears to worthy of reading. Here is a link to a PDF of the review:
The Gutenberg Parenthesis’
Excellent input, dear "LAMBERT." Your insight into the future developments of AI is remarkably accurate and vividly portrays the extraordinary possibilities on the horizon.
When it comes to book collecting, a broader perspective is essential. It's not just about academic knowledge and a textual approach; we must also consider various factors related to the materiality of books and the multifaceted lenses through which they can be viewed. AI will undoubtedly provide new tools that can enhance collectors' approaches in certain areas. However, I believe that many aspects of book collecting will remain firmly within the realm of human expertise.
When it comes to book collecting, a broader perspective is essential. It's not just about academic knowledge and a textual approach; we must also consider various factors related to the materiality of books and the multifaceted lenses through which they can be viewed. AI will undoubtedly provide new tools that can enhance collectors' approaches in certain areas. However, I believe that many aspects of book collecting will remain firmly within the realm of human expertise.
What a nice little story! Thank you for sharing it with us, Julien!
Part I: The AI promise!
In my view AI will change serious book collecting a lot. Those of us, who revel in establishing the print history of a certain book, edition, issue will see their endeavors enriched by completely new capabilities of technology. No more hunting for small changes in texts, spelling, punctuation. No more fear of missing out on minute detail during collation that might hold the key for establishing or refuting hypotheses of the way a certain text evolved over time. The accessibility of large data of digitised books will also allow us to solve many puzzles of provenance that hitherto had no chance of ever being solved. Think of the ability to get a better understanding of the composition of libraries of the past.
The downside of this? Our own knowledge will be seriously devalued. Those of us with the ability to remember books, auctions, marks of provenance will find their expert status seriously diminished. Value will be less in memory, and more in the ability to ask the right questions to AI. May be this has a good side too. Younger collectors will find it much easier to enter the discussion.
And the book market? It seems even more concentration on desireable copies of a certain book will be the consequence. But may be more books will deemed to be interesting as we start to learn even about the lesser authors and works.
As you can see I am an optimist. Change has many advantages. If you belong to those that feel the way to get there is more interesting than the destination, you will have a wonderful journey ahead of you. For those of us, who try to preserve the status quo by collecting the past, things look dreary indeed.
Part II: AI Reality today!
But mind you, all of this is far out. Reality is and will be for years to come different. Here is an example. I have recently acquired a book on economic thought dated 1691. The title page had an ownership inscrpition, which read "Rich. Rooth"... after three hours of research I understand that this has been Irishman Richard Rooth (1657-1726), who was a friend of the Quaker William Penn (1644-1718). Penn had secured Rooth the governorship of Kinsale, Ireland, as a successor to this father. Both fathers had served in the Royal Navy.
Then I asked ChatGPT, what it is that we know about him. Here is the answer:
"Richard Rooth (1657-1726) was a French economist and writer known for his work during the late 17th and early 18th centuries. He is often associated with early economic thought and writings on economic policy. Here are a few key points about him:
Economic Thought: Richard Rooth is known for his contributions to economic thought, particularly his views on taxation and public finance. He wrote about economic issues during the reign of Louis XIV in France.
"La France Sous le Règne de Louis Boisguilbert": One of his notable works is "La France Sous le Règne de Louis Boisguilbert," which can be translated as "France Under the Reign of Louis Boisguilbert." This work is an important historical source for understanding economic conditions in France during that era.
Influence: Rooth's ideas were influenced by the writings of Richard Cantillon, another early economist. He explored the role of taxation and its impact on the economy, including the burden of taxes on different segments of society.
Legacy: While not as well-known as some of the later economists of the Enlightenment period, Richard Rooth's work contributes to our understanding of economic thought in the late 17th and early 18th centuries. His writings shed light on economic conditions and debates of his time."
Needless to say that all of that is pure nonsense... however also interesting to note is that the AI seemed to have used previous questions I had asked about Pierre LePesant Boisguilbert.
Part III: The short answer!
Will it change collecting? Yes... but probably more for the next generation...
Part I: The AI promise!
In my view AI will change serious book collecting a lot. Those of us, who revel in establishing the print history of a certain book, edition, issue will see their endeavors enriched by completely new capabilities of technology. No more hunting for small changes in texts, spelling, punctuation. No more fear of missing out on minute detail during collation that might hold the key for establishing or refuting hypotheses of the way a certain text evolved over time. The accessibility of large data of digitised books will also allow us to solve many puzzles of provenance that hitherto had no chance of ever being solved. Think of the ability to get a better understanding of the composition of libraries of the past.
The downside of this? Our own knowledge will be seriously devalued. Those of us with the ability to remember books, auctions, marks of provenance will find their expert status seriously diminished. Value will be less in memory, and more in the ability to ask the right questions to AI. May be this has a good side too. Younger collectors will find it much easier to enter the discussion.
And the book market? It seems even more concentration on desireable copies of a certain book will be the consequence. But may be more books will deemed to be interesting as we start to learn even about the lesser authors and works.
As you can see I am an optimist. Change has many advantages. If you belong to those that feel the way to get there is more interesting than the destination, you will have a wonderful journey ahead of you. For those of us, who try to preserve the status quo by collecting the past, things look dreary indeed.
Part II: AI Reality today!
But mind you, all of this is far out. Reality is and will be for years to come different. Here is an example. I have recently acquired a book on economic thought dated 1691. The title page had an ownership inscrpition, which read "Rich. Rooth"... after three hours of research I understand that this has been Irishman Richard Rooth (1657-1726), who was a friend of the Quaker William Penn (1644-1718). Penn had secured Rooth the governorship of Kinsale, Ireland, as a successor to this father. Both fathers had served in the Royal Navy.
Then I asked ChatGPT, what it is that we know about him. Here is the answer:
"Richard Rooth (1657-1726) was a French economist and writer known for his work during the late 17th and early 18th centuries. He is often associated with early economic thought and writings on economic policy. Here are a few key points about him:
Economic Thought: Richard Rooth is known for his contributions to economic thought, particularly his views on taxation and public finance. He wrote about economic issues during the reign of Louis XIV in France.
"La France Sous le Règne de Louis Boisguilbert": One of his notable works is "La France Sous le Règne de Louis Boisguilbert," which can be translated as "France Under the Reign of Louis Boisguilbert." This work is an important historical source for understanding economic conditions in France during that era.
Influence: Rooth's ideas were influenced by the writings of Richard Cantillon, another early economist. He explored the role of taxation and its impact on the economy, including the burden of taxes on different segments of society.
Legacy: While not as well-known as some of the later economists of the Enlightenment period, Richard Rooth's work contributes to our understanding of economic thought in the late 17th and early 18th centuries. His writings shed light on economic conditions and debates of his time."
Needless to say that all of that is pure nonsense... however also interesting to note is that the AI seemed to have used previous questions I had asked about Pierre LePesant Boisguilbert.
Part III: The short answer!
Will it change collecting? Yes... but probably more for the next generation...
Dear "COPERNIC",
thank you so much for your detailed note and for sharing with us your own experience with AI !
I agree with most of your thoughts. In particular, I find extremely significant that despite the unbelievable improvements of this technology over the last years, it still remains unable to dig efficiently into the huge amount of information the study of books requires to handle. It still lacks the ability of priorizing and classifying properly, depending on the context, the information it collects, and uses little tricks (like referring to the user's previous requests vaguely related to the same subject) to make its speech look plausible, without any safety proceeding to dismiss errors and factual unconsistency.
As a result it often produces complete nonsense when questionned about our specialized field of rare books. What is funny and worrying at the same time with our common friend, Mr. ChatGPT (or Mrs., but that is another question), is the self-confidence it sprays in his responses, while making totally wrong statements. It's able to speak for hours but it's just a machine, after all, and you should'nt expect him to always make much sense. Why is it worrying? Well, just because unexperienced rare book lovers will perhaps tend to believe in excess what it says.
I do believe that human knowledge, experience and reflection are still necessary to investigate rare books efficiently. Our reference library, here at the book shop, gathers litterally thousands of volumes. It's not all about being able to browse billions of words and relating them in a given context. It's perhaps even more important to be able to check the right reference with the proper perspective, and to draw a line that connects one source of information with another. This requires a different approach than just searching a chain of characters within tons of printed pages, and AI doesn't seem ready yet to perform this search without human supervision.
That being said we have certainly beautiful tools to be brought to us in the future by AI and serious collectors of rare books, either private or institutional, will definitely see a wider range of tools helping them to raise a deeper understand and knowledge of their material.