NotebookLM shows us how AI actually should be implemented in Logos
The use of AI in Logos has been controversial, and with good reason. I believe I have found the model that Logos should follow for this implementation that will make the maximum use of the tools and our library without giving over too much of the thinking to a computer. Having AI write material for your sermon is a failure to fulfill your office; this is the wrong direction for Logos to go. However, serving as a research librarian assistant is the entire point of Logos, and AI can do this.
I have been playing around with NotebookLM, which is based on Gemini. https://notebooklm.google/
It is free to use, and I encourage anyone who is curious to export out a few .txt files from your library on a given topic and let this chew on them. The results speak for themselves. You can ask questions, and will get well resourced and cited responses to the questions, with the ability to immediately see the evidence for the assertions by hotlinked footnotes. It adheres very closely to the material you give it. In all my experiments it has been astonishingly accurate. Whenever I am unhappy with an answer, I discover that it is citing materials that I had given which I disagree with, it isn't the AI's fault. If you give it scholars of a certain type, the answers will be based entirely on those sources. It is a reliable research assistant that shows you its work.
Now, for Logos to incorporate this fully into the library tools would be a dream come true, especially if we could classify our sources by reliability, denominational perspective, and quality. Perhaps collections could be improved in order to make research clusters. I promise you that if you try notebookLM you will understand immediately why this needs to be the model for Logos AI. (Perplexity is another good example to look at).
However, in the short term this could also be accomplished by simply allowing us to have better export features for guide or workflow results. Give us a way to produce a "packet" on a given text, making a document that puts in all the clippings from commentaries, the language resources and so on into a document that fits within the token limit of a tool like this. Right now I can do this manually, but it ought to be possible to do this automatically. These reading packets could be somewhat like how the clippings list works when applied as a filter, one "document" that you can read from top to bottom which pulls from resources according to a certain criteria.
You really need to try it for yourself, but here is an example for a notebook I made where I took material from a few Lutheran Dogmaticians and asked it questions (note the citations don't copy/paste so you just have to trust me that it has over 20 citations to the material I gave it, for each assertion there is at least one source). You can ask follow up questions, ask for study guides and more. Here is an output:
I asked: What role should the church fathers play in determining doctrine? Give examples of what Lutheran Dogmaticians have stated on this subject.
NotebookLM:
The Role of the Church Fathers in Lutheran Dogmatics
The church fathers serve as witnesses to the truth found in Scripture, offering valuable insights into the historical interpretation and development of Christian doctrine. They are not, however, to be considered authorities on par with Scripture in determining doctrine. The Lutheran Confessions, echoing a core principle of the Reformation, emphasize that Scripture alone (sola scriptura) holds ultimate authority in matters of faith and doctrine.
Luther viewed the Church as "the infallible ground of the truth" but recognized that the Church throughout history is fallible. The church fathers, as part of this historical Church, can err and their interpretations should always be compared to Scripture. Therefore, the Lutheran approach prioritizes the clear and accessible Word of Scripture over any external authority, including the church fathers.
Statements from Lutheran Dogmaticians:
- Chemnitz: Warns against the practice of formulating articles of faith based on the church fathers. He emphasizes that the church, while guided by Scripture, is not a monolithic entity with a fixed structure determined by bishops or historical precedent. He highlights the dynamic nature of the church, rooted in the Word and responsive to specific contexts.
- Gerhard: Argues for the distinction between theology derived from God's Word (objective theology) and the inward application of that theology by the Holy Spirit (concrete theology), both of which are essential for a proper understanding of Scripture. He considers the church fathers, along with human teachers of the Church in general, as valuable "witnesses of truth," demonstrating the historical continuity of Scriptural teaching. However, he maintains that theologians should derive their doctrine solely from Scripture and not rely on the pronouncements of these teachers.
- Quenstedt: Emphasizes the divine origin of Scripture and its inerrancy, highlighting that it is a direct result of the Holy Spirit inspiring the biblical writers. He recognizes the need for theological prolegomena, a foundational framework for understanding and interpreting Scripture. Quenstedt stresses the importance of Scripture alone as the source of theology, rejecting the Roman Catholic position that tradition holds equal authority.
The Lutheran tradition acknowledges the importance of the church fathers as historical guides while affirming the primacy of Scripture in determining doctrine. This approach avoids elevating any individual or group, even esteemed figures from church history, to a position of authority alongside God's Word.
Comments
-
I agree! NotebookLM capabilities combined with Logos books would make an extremely powerful tool. I am already exporting my Logos books as PDFs so that I can use them in NotebookLM. Currently NotebookLM allows for 50 sources of 500,000 words (this being roughly 1000-1300 pages). This amounts to around 60,000 pages of source material per notebook! By strategically grouping books on specific topics or passages, I'm seeing fantastic results for my research. This integration should undoubtedly be the future of Logos AI.
Also, the hallucinations which are so concerning in AI in general are almost non-existent in NotebookLM because google has somehow limited the scope of usable information to your sources.
1 -
I have been playing around with NotebookLM which had me also thinking how great Logos would be if I could interact with library resources like this. There is a semblance of it available in the Smart Search. You can filter your search on a particular book and ask questions using Smart Search but it is not the rich experience you get from NotebookLM. I am hoping Logos go down this track. The beta Synopsis in search on the Web App is another step in this kind of direction. When I first came across AI I realised that the day would come when I could ask questions of my books and they would answer them. To get a synopsis of a book, or a Briefing, as NotebookLM puts it, is much more useful than a simple summary. I am looking forward to what Logos will do.
👁️ 👁️
1
