About the Anselm Project Bible

The APB is a fresh translation based on the following standard critical editions of the original language texts. No other base texts were consulted for Version 1.0.

• Old Testament: The Unicode/XML Westminster Leningrad Codex (WLC), which represents the Masoretic Text (MT).
• New Testament: The Greek New Testament: SBL Edition (SBLGNT), a modern critical text.

Variant Policy: Version 1.0 of the APB is a direct translation of the specified base texts (WLC and SBLGNT) and does not, as a rule, depart from them to adopt alternate readings from other manuscripts.

Future Considerations (V2): While other important textual traditions like the Septuagint (LXX) and the Dead Sea Scrolls (DSS) were not used for textual adoption in V1, their use in providing translational insight and variant notes is a key consideration for future development.

• Ketiv/Qere: The translation follows the Masoretic scribal tradition of prioritizing the Qere (what is to be read) in the main text. For example, in Psalm 100:3, the APB reads "...it is he who made us, and we are his," following the Qere ("and His") rather than the Ketiv ("and not").
• Psalm Titles: Titles given in the Hebrew text (e.g., "A psalm of thanksgiving") are included as part of verse 1, following a common English convention.
• Psalm Numbering: The APB uses the psalm numbering from the Masoretic Text (WLC) exclusively and does not include alternative numbering from the Septuagint/Vulgate.
• OT in the NT: The APB faithfully translates its base texts as-is, without attempting to harmonize them. When the NT (SBLGNT) quotes the OT using a Septuagint reading, the APB translates that reading. When translating the original OT passage (WLC), it translates the Masoretic text. (e.g., Hebrews 10:5 reads "...a body you prepared for me," while Psalm 40:6 reads "...you opened my ears." V1 does not add a cross-note).
• Deuterocanon: The Deuterocanonical books (Apocrypha) are not included in Version 1.0. They may be considered for inclusion as a separate volume in a future update. I have taught their contents in the past as part of my academic work, and love their contents, but remain dedicated to the standard 66 book canon.

The APB is not the product of a single, rigid "essentially literal" or "functional" philosophy. Instead, it is an "emergent" translation, the result of an experiment in AI methodology.

Each pericope was processed by an "AI committee." A translator AI generated candidates, and a separate "judge" AI, given all the linguistic data and context, selected the best rendering. This process was repeated for all 2,022 pericopes, resulting in a text that seeks to be faithful to the data provided at every point, rather than to a single, human-defined rulebook.

• Divine Name: The tetragrammaton (YHWH) is consistently rendered as "LORD" in all-caps. The Hebrew term Adonai is rendered as "Lord" in title-case.
• Inconsistency: As a product of the "AI committee" approach, some minor inconsistencies in renderings may occur (e.g., "Bathsheba" in one passage and "Beth-sheba" in another). This is an emergent property of the V1 methodology.
• Idioms: The AI was instructed to preserve the original Hebrew and Greek idioms, translating them literally or with their most traditional English equivalent (e.g., "Gird up your loins" in Job 38:3; "my horn is exalted" in Psalm 92:10).
• Measures & Currency: Ancient units of measurement, weight, and currency are transliterated directly into the text (e.g., "cubits," "shekels," "denarius"). No in-text conversions or footnotes are provided in V1.

The AI was instructed to be reasonably literal, which resulted in a formal register that preserves gendered language present in the original texts.

• The Greek adelphoi is consistently rendered as "brothers."
• The Greek anthrōpos is rendered as "man" where context implies a generic individual.
• Specific gendered terms, such as moichalides ("adultresses") in James 4, are preserved and not neutralized into "adulterous people."

Section headings (pericopes) are drawn from a pre-generated library (PERICOPES.json) that divides the text into 2,022 logical units. The AI was instructed to process the text within these defined boundaries. The layout of poetic sections (e.g., Psalms, Proverbs) is also handled programmatically based on the underlying text structure.

The APB is designed as a digital-first text. The web reader provides users with direct access to the "artifacts" of the translation process for deeper study. For a given passage, users can view:

• Original Language Text: The base WLC or SBLGNT text.
• Morphology: Grammatical analysis of the original-language words.
• Alternate Translations: Other high-quality candidates considered by the AI.
• Translation Rationale: The "judge" AI's explanation for its final choice.
• AI Uncertainty Score: A metric indicating the AI's "confidence" in its translation.

In addition to translation-specific data, the APB reader also provides:

• Cross-References: A list of algorithmically-identified parallel and thematic passages.
• Discourse Notes: AI-generated analysis of the passage's literary structure, flow, and main themes.

The "AI committee" process used a mixture of models:

• Primary Translator: An advanced version of chatgpt-5 was used for the base translation generation.
• Review & Judgment: In cases of high disagreement or low confidence, a "review panel" consisting of chatgpt-5-pro, gemini 2.5 pro, and grok 4 provided alternate candidates. A final "judge" (chatgpt-5-pro) made the definitive selection based on this evidence.

To prevent the AI from "remembering" copyrighted modern translations (e.g., NIV, ESV) from its training data, a strict data-first approach was used.

The AI was not given a simple prompt like "Translate John 3:16." Instead, it was fed a structured JSON object containing the raw SBLGNT text, its lemma (dictionary form), and its full morphological analysis. The AI's task was to construct a translation from this data, forcing it to act as a linguist rather than a recall engine.

Terminology: No predefined glossary or translation memory was used in V1. Terminology was "free-generated" by the translator AI based on the lemma data, and then selected by the judge AI, creating an emergent and context-sensitive lexicon.

Audit Trail: The entire process is transparent. The JSON "artifact" file (e.g., Matt_1_1-17__artifacts.json) generated for each pericope serves as a complete audit trail, storing the inputs, candidate translations, judge's rationale, and the final selected text.

Processed 2,022 pericope token files for the full V1 translation.

Input Tokens: 27,158,000

Output Tokens: 123,567,000

Total: 150,725,000 tokens