How Logos uses AI

Fantastic. I love all the links included, allowing one to familiarize oneself with various aspects of AI within Logos while still keeping the article organized and at a manageable length.

Well done, Mark.

Well written and very helpful. Thanks for taking the time to do this.

1st: Thanks for the article.

In the article it says:

The Bible is the most extensively written-about book in history, which means that rather than searching the Bible in a traditional sense, we can use the vast amount of Bible-related training data in AI models to help us locate the most relevant verses, pericopes, or chapters with much more nuance than would be possible with a traditional search.

If I'm interested in a catholic (or protestant or orthodox or… ) view of justification wouldn't it be nice to have a drop-down menu or so to choose the denominational lense for the query?

I genuinely appreciate Logos for pioneering this use of technology and maintaining a strategy grounded in clearly articulated principles. I look forward to seeing where this pathway leads!

@Mark Barnes (Logos) said:

For Smart Synopsis, Logos does something more sophisticated: it takes your search query and top five search results, sends those results to the AI model along with your original query, and asks the model to extract and combine the most relevant portions into a coherent synopsis. This means Smart Synopsis is not just the AI’s guess at an answer—it’s a synthesis drawn directly from the best content in your library, with clear citations so that you can check the sources yourself.

I have not used AI in Logos much … but it seems obvious here that a big assumption is being made that the top 5 search hits happen to be from the best content in your library ?

I have not used AI in Logos much … but it seems obvious here that a big assumption is being made that the top 5 search hits happen to be from the best content in your library ?

I'm using the word "best" as shorthand for "most relevant" rather than as shorthand for "most respected."

Right, but even in this case a big assumption is being made. Ideally, the statement would be true. But it certainly is not always. The most relevant content from one's library does not always appear within the top 5 search results.

If I'm interested in a catholic (or protestant or orthodox or… ) view of justification wouldn't it be nice to have a drop-down menu or so to choose the denominational lense for the query?

We've deliberately kept the UI for Smart Bible search as minimal as possible to give you as much freedom as you need to create your search. You can search for catholic views of justification or something similar to get (for example) James 2:24 very near the top of the results, or reformed views of justification giving Ephesians 2:8-9 very near the top.

Thanks, Mark. I would love it if the user had some options when using AI for synopsis or summaries to decide things like how long the answers should be. Since we are credit-limited, this should not be an issue - if I prefer longer replies (and I do), I would simply be able to use fewer requests, which I would be fine with.

Very helpful article - thank you @Mark Barnes (Logos).

This is a very helpful summary. Thanks.

@Aaron Hamilton

The most relevant content from one's library does not always appear within the top 5 search results.

That has been my experience for sure, as one who has not spent the time required to organize and prioritize resources.

So it appears that prioritizing your resources well would not just result in better searches, but also information being sent to the AI?

@John A user's prioritization of resources does not currently affect Smart search in any way.

We've deliberately kept the UI for Smart Bible search as minimal as possible to give you as much freedom as you need to create your search. You can search for catholic views of justification or something similar to get (for example)

James 2:24 very near the top of the results, or reformed views of justification giving Ephesians 2:8-9 very near the top.

Ok, I see. Thanks for the screenshots, too.

But why do I get slightly different results for the question about the Papal supremacy in German vs English?

And a follow up question: What about a "denomnational" lense in Factbook?

But why do I get slightly different results for the question about the Papal supremacy in German vs English?

It's impossible to say. Perhaps German speakers tend to refer to these verses more often than English speakers, and the AI has noticed that? But that's a guess, and it's not provable either way.

What about a "denomnational" lense in Factbook?

We're considering more lenses. Right now, we don't have a lot of denominational tagging outside commentaries, which is why only the commentaries section includes denominational grouping.

Thanks. Nicely done, Mark.

very cool. 🤓🔬➕➖➗

Well done Mark Barnes, thanks.