Preface
AICNT (AI Critical New Testament)
Introduction
The AI Critical New Testament (AICNT) is a critical edition of the New Testament translated with AI for maximum transparency into the text.
A critical edition is a version of the text that is carefully edited and compiled based on extensive review of the earliest available manuscripts in contrast to later textual traditions and in reference to other benchmarks including other critical editions. The primary objective of a critical edition is to reconstruct, as closely as possible, the original text of the New Testament writings as they were first authored. A critical edition also serves to document variants and the corresponding authorities that support particular readings through a “critical apparatus,” a system of notes and annotations accompanying the main text.
Critical editions of the New Testament are typically in Koine Greek, and are not adapted for Bible students who rely solely on English resources. The AICNT pioneers a new approach by offering a critical edition in English that leverages AI. This method ensures an objective and transparent output, minimizing human biases and interpretative proclivities. The innovative approach opens up a rich treasury of information to a broad audience. It enhances accessibility and understanding of critical biblical scholarship for those who may not be proficient in the original languages.
The English text of AICNT (including variants) is completely rendered by AI according to a methodology that delivers excellent accuracy and readability, with additional protocol to ensure an unbiased and transparent output. The strategic utilization of AI for this translation overcomes many shortcomings associated with relying on human translators by reducing the potential for human error and subjective bias.
The AICNT is tailored for readers desiring a literal and transparent translation of the New Testament that aligns with the earliest known Greek manuscripts, with additional transparency into textual changes that occurred over time.
The specific objectives of the AI Critical New Testament (AICNT) are as follows:
1. Base Translation on Greek Text: the translation is directly based on the Greek text as attested by benchmark critical editions relied upon for modern Bible translations.
2. Integration of Textual Variants: the translation incorporates, yet distinguishes, variant text-types of early manuscripts and later textual traditions, including the Byzantine text-type and the Textus Receptus.
3. Textual Variants Identification: use of braces {}, brackets [ ], and double brackets [[ ]] within the text to identify and differentiate textual variants as follows:
(a) Text with alternate reading: braces {}
(b) Text not unanimously attested by early manuscripts: single brackets [ ]
(c) Text regarded as not originally part of the text: double brackets [[ ]]
4. Extensive Footnoting for Textual Variants: the critical apparatus includes extensive footnotes documenting textual variants, including changes and additions in later manuscripts, differences between early manuscripts, and variances in critical editions.
5. Notation of Cited Witnesses: incorporation of conventional notation for frequently cited witnesses in the critical apparatus, including:
(a) Earliest manuscripts (up to the end of the 5th Century)
(b) Traditional text-types (BYZ and TR)
(c) Benchmark critical editions (NA28, SBLGNT, THGNT).
6. Use of BDAG Lexicon: the translation utilizes the authoritative BDAG lexicon for translating Greek words into English.
7. Accurate and Neutral AI Rendering: utilize AI (GPT-4) to achieve an accurate, readable, transparent, and unbiased translation.
8. Avoid Editorializing: the translation avoids editorializing through suggestive section headings and other types of suggestive formatting.
9. Optimal Paragraph Breaks: frequent paragraph breaks are implemented to improve readability and comprehension, balancing between large blocks of text and smaller text segments.
10. Clarifying Footnotes: provision of footnotes for clarity, including abridged BDAG definitions, cross-references, and notes pertaining to the neutrality of the rendering.
By accomplishing these core objectives, the AI Critical New Testament equips English readers with a critical New Testament of unparalleled depth and transparency.
The AICNT is a window into the earliest New Testament manuscript tradition via the critical apparatus combined with an accurate, readable, and neutral AI rendering. Thus, with this revolutionary and innovative tool, readers can ascertain the text of the New Testament for themselves.
Part I: Critical Edition, Rationale & Methodology
1. Understanding Textual Changes
Over the centuries, the text of the New Testament has undergone various kinds of changes as it was copied, translated, and transmitted. These changes range from unintentional errors like misspellings, omissions, and duplications to more deliberate alterations such as harmonization between Gospel accounts, adding words for clarification, streamlining the text, or the inclusion of marginal notes into the text.
Variants that enhance or embellish the story are often termed "expansions of the text" or "additive variants." These are instances where scribes might add phrases, sentences, or even entire verses to make the narrative more detailed, dramatic, or theologically rich. Such changes could include additional dialogue, descriptive elements, or interpretative clarifications. Expansions of the text may be aimed at clarifying or elaborating upon specific points for the reader or listener. In other instances, they might arise from the incorporation of marginal notes into the main body of the text.
Among the variants in New Testament manuscripts, the most significant are those with theological or doctrinal implications. Such variants are not merely scribal errors or stylistic modifications; they can impact how a passage is interpreted and understood in the context of Christian beliefs.
The New Testament as we know it today is based on many ancient manuscript copies. Later texts exhibit the cumulative effect of over a thousand years of textual changes. The later manuscript witnesses of the 12th-16th centuries have thousands more words in the Gospels than do 3rd-5th century manuscripts. Thus, the text has numerous variations between copies of the same book. Textual criticism is the practice of comparing these various copies to figure out the original text.
The sheer number of manuscripts, spanning different periods and geographical locations, has resulted in a complex textual history with multiple text-types like Alexandrian, Western, and Byzantine. Textual critics employ a range of methodologies to evaluate variations to reconstruct the most authentic version of the New Testament text. Studying and comparing these manuscript families and individual variants is crucial for understanding the New Testament's historical development, interpretation, and reliability.
2. What is a Critical Edition?
A critical text or edition is based on the best scholarly estimates of the original text as best reconstructed from the manuscript witnesses. A critical edition is a publication of a text designed to present the most accurate and authoritative version of that text. It is based on exhaustive research involving the collation of various manuscripts, fragments, and other primary sources. The edition typically includes a base text, considered the most reliable representation of the original, along with a critical apparatus that annotates textual variants, manuscript differences, and further justification for decisions.
In addition to the purpose of a critical edition of offering a text as close as possible to the original, they aim to serve as a reliable reference and provide insight into the history of textual transmission. These editions are essential tools in biblical scholarship. There have been numerous critical texts over the last 200 years. Many have been revised and updated as additional manuscript evidence has been discovered. Some differ in methodology or criteria for selecting between textual variants.
Critical editions typically include a critical apparatus, an organized system of notations to represent, in a single text, the complex history of that text in a concise form. The apparatus typically includes footnotes with standardized abbreviations for the source manuscripts that indicate what manuscripts attest to specific readings (variants).
For a concise primer on textual criticism and how modern critical texts are determined, Introduction to the Greek New Testament, by Dirk Jongkind (Tyndale House Cambridge, 2019), is recommended. For a more detailed introduction to critical editions and the theory and practice of modern textual criticism, recommended books include:
Aland, Kurt, and Barbara Aland, The Text of the New Testament: An Introduction to the Critical Editions and to the Theory and Practice of Modern Textual, Translated by Erroll F. Rhodes, 2nd ed. (Eerdmans, 1995)
Metzger, Bruce M., and Bart D. Ehrman, The Text of the New Testament: Its Transmission, Corruption, and Restoration, 4th ed. (Oxford University Press, 2005)
Parker, D. C., An Introduction to the New Testament Manuscripts and their Texts (Cambridge University Press, 2008)published by Oxford University Press (3rd Ed. 1992).
3. Braces and Brackets
In this project, we have endeavored to document all significant differences between the text of the early manuscripts, the current critical editions, and the later text-types. Text considered by the critical editions to be later improvements and additions is distinguished from the undisputed text with brackets.
Braces {}, brackets [ ], and double brackets [[ ]] are a means within the body of the text to segment and group textual variants and to highlight and differentiate text as follows:
{} Braces, called curly brackets, indicate where textual variants occur with alternate wording. The variant is selected for the text that is most likely the original reading based on the critical editions. The readings of other manuscripts are noted in the footnotes.
[ ] Single square brackets pertain to text included in at least one of the critical editions but absent in one or more early manuscripts. Since the text in single brackets is not unanimously attested by all the early manuscript witnesses, this indicates some uncertainty about the text being in the original autographs.
[[ ]] Double square brackets indicate words or series of words deemed later additions and interpolations by the Benchmark Critical editions. This material is either indicated with double brackets in the critical editions or only included in the notes of the critical editions, indicating that critical editions do not consider the text original. Double brackets separate supplementary or secondary content (later additions) from primary content (earliest text).
Brackets also indicate varying levels of certainty concerning the text's original wording. By using this formatting in the English translation, readers are given a clear indication of the extent of significant textual variants exhibited through centuries of reproduction. Readers can easily distinguish the critical text from manuscript changes that prevailed later in development.
4. Types of Variant Footnotes
There are several types of footnotes used for documenting variants. The following summaries are provided to clarify how to understand particular footnotes.
Some manuscripts read “x” – This statement pertains to text in braces {} and indicates a variant of a different word or combination of words between manuscripts. The best-attested reading is the wording in the base text, and the alternative reading (considered a later variant) is provided in the footnotes. Usually, it is evident that the alternate reading is that of later manuscripts, but occasionally, there is also early manuscript support for the alternate reading. Also, sometimes more than one alternate reading is noted.
Later manuscripts add – This statement pertains to text in double brackets [[ ]], which lacks support from any early witnesses. Usually, the only support for this reading among the frequently cited witnesses is BYZ and TR.
Some manuscripts add – This statement pertains to text in double brackets [[ ]], which is not considered original but is incorporated in at least one early manuscript of the frequently cited witnesses. This text is double bracketed because, although there may be some support in early manuscripts, none of the benchmark critical editions accept it as original.
Absent from some manuscripts – This statement pertains to text in single brackets [ ] absent in one or more early manuscripts. Although critical editions have incorporated the text based on significant early attestation, there remains some uncertainty about whether the text is original.
- When the footnote includes witnesses supporting the base text and those pertaining to a variant, the contrasting statements and their corresponding witnesses are separated with ‖.
Footnotes serve as a valuable tool for further providing additional information about variant readings, including which particular manuscripts and additions affirm them. The variant footnotes include notations indicating the frequently cited witnesses that support the reading.
Typically, a statement is made, followed by notation indicating the textual witnesses that support the particular statement. Witnesses provided generally are only those frequently cited witnesses listed in the section entitled “Frequently Cited Witnesses.” Later and less relevant manuscripts are typically not noted so as not to overwhelm the general reader with information. When the word “manuscripts” is used in the footnotes, it typically refers to multiple manuscripts, even when only one witness is cited. In some cases, there may be a single manuscript that the note pertains to.
5. Exclusions to Documented Variants
The aim of the footnotes in this translation is not to document every minor variation. Instead, the focus is on documenting those variants that result in a significantly different English rendering. Variants that are generally not noted in the AICNT critical apparatus are as follows:
Orthographic variations: Differences in the spelling of a Greek word that results in an identical rendering are generally not noted.
Abbreviations: In various manuscripts, words are sometimes abbreviated, including some common frequent words (e.g., man/human) and, more often, Nomina Sacra (sacred names or titles such as “God” or “Lord”). Variants pertaining to the use of abbreviations are not noted.
Articular variations: Differences where some manuscripts include the definite article “the” and some exclude it are generally not noted unless it may have a significant interpretative implication.
Conjuctional variations: Differences in the use of conjunctions “and” “but” or “for” are not always noted.
Tense variations: Differences in verb tenses in cases where the different tenses of the same word result in an identical or nearly equivalent rendering are generally not noted.
Word substitution: When synonyms or near-synonyms are used in place of certain words, these differences are not noted unless they result in a different rendering.
Verse divisions: Differences resulting from dividing text in verses are generally not noted. An example is when the text at the end of a verse in one edition is transposed to the beginning of the following verse in another. Verse divisions are more a matter of textual presentation and organization than textual integrity. The AICNT follows the verse divisions of the critical editions used as a benchmark.
Documenting the exclusions mentioned above would result in extensively more footnotes without adding much value for the general reader.
6. Manuscript Text-Types
The terms “Alexandrian,” “Western,” and “Byzantine” refer to the three primary text-types or families of manuscripts of the New Testament. These text-types are named after the regions where they were most commonly found or associated with and exhibit distinct textual characteristics. The following is an outline of these three major text-types.
Alexandrian Text-Type: Initially associated with Alexandria, Egypt. Generally dated to an earlier period, from the 2nd to 5th centuries. Known for a more concise and grammatically rigorous text. It is generally considered by many scholars to be closer to the original autographs. Often favored by scholars for its early dating and perceived textual purity.
Western Text-Type: In early textual criticism, this text-type was associated with the Western part of the Roman Empire, particularly centers like Rome and Carthage. It is now established that this text was found in the East as well as in the West, with correlation to Old Syriac manuscripts and quotes from Eastern church fathers. Extends from the 2nd to the 9th centuries, although most manuscripts are from later periods. Known for significant paraphrasing, expansion, and occasional omissions. It often includes additional explanatory material. It is generally not favored for establishing the original text due to its loose handling of the text. Still, it provides valuable insights into early interpretation and the history of the text.
Byzantine Text-Type: Associated with the Byzantine Empire, centered around Constantinople (modern-day Istanbul, Turkey). Mostly dated to the 5th century and later, although some may exist from earlier periods. Known for a smoother, more polished text that sometimes includes explanatory phrases. It is the most numerous among the manuscript families. This text became the prevailing ecclesiastical form of the New Testament and eventually constituted the basis of the Textus Receptus. Generally considered less reliable for critical text reconstruction due to its later dating and tendency toward textual expansion.
Other minor subtypes, including the Caesarean text-type and text families such as f1 and f13, exhibit a distinctive relationship. However, the three major text-types outlined above are the most significant when considering the textual history and transmission of the New Testament. Manuscript families are important factors in New Testament textual criticism.
7. Categories of New Testament Manuscripts
Aland and Aland developed a system for categorizing the degree to which various text-types are exhibited in New Testament manuscripts.a The five categories are as follows:
Category I: “Manuscripts of a very special quality which should always be considered in establishing the original text.” These manuscripts have almost no Byzantine influence. Although most texts in Category I agree with the Alexandrian text-type, they are not necessarily Alexandrian.
Category II: “Manuscripts of a special quality but distinguished from manuscripts of Category I by the presence of alien influences.” These are generally Alexandrian texts with some Byzantine intrusion.
Category III: “Manuscripts of a distinctive character with an independent text, usually important for establishing the original text, but particularity important for the history of the text.” These are manuscripts with a large Byzantine component and many non-Byzantine readings.”
Category IV: “Manuscripts of the D text.” The D text is the principal witness of the Western text-type and stands apart from the Alexandrian and Byzantine text-types.
Category V: “Manuscripts with a purely or predominately Byzantine text.” This is often associated with the majority text-type more developed than the texts of other categories.
8. Frequently Cited Witnesses
Frequently cited witnesses are noted in the critical apparatus footnotes when applicable for indicating support for various types of textual variants shown in the footnotes. The frequently cited witnesses fall into three categories: (a) early manuscripts, (b) text-types of later tradition, and (c) benchmark critical editions.
a. Early Manuscripts
Particular Greek manuscripts are noted using symbols and numbers, according to the system of Gregory-Aland. However, the general symbols are maintained here without more complex qualifying symbols (additional syntax) seen in some critical editions, which may not be apparent to the general reader.
Citations herein only pertain to the original version of the manuscripts. Citations pertaining to later corrections made by a second, third, or fourth hand are omitted. To simplify the notation in the apparatus, superscripts used in NA28, including “*” (original reading of a manuscript) and “vid” (apparent reading but not certain) are also omitted. Moreover, for the sake of simplicity, parentheses are not used as in other critical editions to indicate a particular witness that attests to a reading but with slight differences. The goal of the AICNT critical apparatus is to provide only the most essential information in the most intuitive way possible.
These frequently cited witnesses first include papyrus fragments and uncial manuscripts, up to about 500 CE. These are representatives of the most primitive forms of the text. In addition to early Greek texts, there are early texts translated into other languages from this same period, most notably Old Latin and Old Syriac manuscripts.
𝔓: Papyrus fragments. This is the earliest type of New Testament manuscripts. Below is a list of identification numbers of those cited Papyri dated up to about 500 CE, as reported by Aland:b
𝔓 ≤ 200 CE: Category I: 𝔓 32, 46, 52,c 64+67, 66, 77, 90, 104
200 CE < 𝔓 ≤ 300 CE: Category I: 𝔓 1, 4, 5, 9, 12, 13, 15, 16, 18, 20, 22, 23, 27, 28, 29, 30, 37, 39, 40, 45, 47, 49, 53, 65, 70, 72, 75, 78, 80, 87; Category IV: 𝔓 48, 69
300 CE < 𝔓 ≤ 400 CE: Category I: 𝔓 10, 24, 35; Category II: 𝔓 6, 8, 17, 19, 51, 57, 62, 71, 81, 82, 85, 86; Category III: 𝔓 38
400 CE< 𝔓 ≤ 500 CE: Category II: 𝔓 14, 56; Category III: 𝔓 54, 63
ℵ(01): 4th century Greek uncial, Codex Sinaiticus. This is an Aland Category I text. It exhibits numerous singular readings and errors and was overrated by some early textual scholars (Tischendorf), as it is “distinctly inferior to B(03).”d Jongkind regards it as “a copy of a good text, but it is not a good copy of a good text.”e
A(02): fifth-century Greek uncial, Codex Alexandrinus. It is the most complete Greek Bible from before CE 1000, and it contains most of the New Testament with only two parts missing. A(02) is of uneven value, inferior in the Gospels, where it is classified as Aland Category III, and good in the rest of the New Testament, where it is Aland Category I.f
B(03): fourth-century Greek uncial, Codex Vaticanus. This is an Aland Category I text. The end of it from Hebrews 9:14 to Revelation is lost. Many scholars have considered Codex Vaticanus superior to all extent texts. Sir Frederic Keyton wrote, “Codex Vaticanus [is] the most valuable of all the manuscripts of the Greek Bible.”g Professor Kurt Aland said, “Among the uncials, B has a position of undisputed precedence in the Gospels.”h For this reason, Reuben J. Swanson used Codex Vaticanus as the exemplar text in several reference volumes featuring a critical apparatus for documenting variant readings of New Testament books.i
C(04): fifth-century Greek uncial, Codex Ephraemi Rescriptus. This Aland Category II text contains most New Testament books but with sizable portions missing. It originally contained the whole of both the Old and New Testaments.
D(05): fifth-century Greek uncial, Codex Bezae. This is an Aland Category IV text, defined as “Manuscripts of the D text,” that contains the Gospels and Acts. It is the principal example of the “Western” text-type. Due to numerous peculiarities, Codex Bezae has been the most controversial New Testament uncials. Prominent scholars regard it as a highly revised text, but one used as its source text is an outstanding example of a primitive text.j Scholars have expressed the three primary views regarding D(05):
(a) It is heavily revised as influenced by the Old Latin text.k
(b) It is heavily revised as influenced by the Old Syriac text.l
(c) It is the earliest form of the text (most notably Luke and Acts) before efforts to streamline it.mno
W(032): fifth-century uncial, Codex Washingtonianus. This is an Aland Category III text that only contains the Gospels and is frequently characterized by independent readings. W(032) and 𝔓45 are the principal early examples of the hypothetical Caesarean text-type, used to characterize manuscripts with a pattern of variant readings not found in the other more widely recognized text-types. The term Caesarean is used because the particular variants match quotations found in the third-century works of Origen written in Caesarea.
About 50 other Greek uncial fragments are dated up to around 500 CE. These early uncials are grouped below according to the five Aland categories:p
Category I uncial manuscripts: ℵ(01), A(02) (excluding Gospels), B(03), 057, 0162, 0189, 0220, 0254
Category II uncial manuscripts: C(04), I(016), T(029), 048, 071, 076, 088, 077, 0172, 0173, 0175, 0181, 0185, 0201, 0232, 0240, 0244, 0247, 0270, 0274
Category III uncial manuscripts: A(02) (gospels) B(03) (gospels), W(032), 058, 059, 062, 068, 069, 072, 0160, 0163, 0165, 0166, 0169, 0170, 0176, 0182, 0186, 0188, 0206, 0207, 0212, 0213, 0214, 0216, 0217, 0218, 0219, 0221, 0226, 0227, 0228, 0231, 0236, 0242, 0252, 0261
Category IV uncial manuscripts: D(05), 0171
Category V uncial manuscripts: Q(026), 061
Because most of these early uncials only contain parts of a limited number of chapters, they are rarely noted unless they play a pivotal role in attesting to a variant reading. imt4 Early Syriac Manuscripts
Early Latin and Syriac Manuscripts
A handful of Latin and Syriac manuscripts date to the same period of before 500 CE as the early Greek codices of ℵ(01), A(02), B(03), C(04), D(05), and W(032) noted above.
There are about fifty surviving manuscripts (or fragments). Metzger describes the most significant.q Those with substantial text dating from the early period of 350 to 500 CE are identified by the notation of a, b, d, e, ff2, h, i, and k.r Of these, a, b, d, e, and ff2 are characterized as exhibiting a “Western” text-type having similarities to D(05). Notation includes:
a k: indicate individual Latin manuscripts up to the end of the 4th century CE.
b e d ff2 h i: indicate individual Latin manuscripts up to the end of the 5th century CE.
it: (Itala) Used when a reading is supported by the “all or majority of Old Latin witnesses as a group.”
lat: Used when a reading finds support in the Vulgate and a part of the Old Latin tradition.
latt: Used when supported by the entire Latin tradition.
Early Syriac manuscripts from about or before the fifth century are noted as follows:s
sys: Syrus Sinaiticus (Old Syriac manuscript of about 400 CE)
syc: Syrus Curetonianus (Old Syriac manuscript being a revised form of Syrus Sinaiticus, 400's CE)
syp: Syriac Peshitta (400's CE)
syph: Syriac Philoxenian version (dated about 507/508 CE)
sy: The entire Syriac tradition
The Old Syriac versions of Sys and Syc often agree with the “Western” text of D(05) Codex Bezae.
b. Text-Types of Later Tradition
Secondly, the frequently cited witnesses include the Byzantine text-type and the Textus Receptus, representatives of enhanced textual traditions established in later periods of textual transmission.
BYZ: The Byzantine text-type (its earliest form and not necessarily the majority of manuscripts). Used herein, BYZ is in reference to Robinson, Maurice A., and William G. Pierpont, The New Testament in the Original Greek: Byzantine Textform (Bellingham, WA: Logos Bible Software, 2005). “BYZ” has been used herein and is sometimes indicated by “RP” in other critical editions. This text-type corresponds to Aland Category V, defined as “Manuscripts with a purely or predominately Byzantine text.”
TR: Stephen’s 1550 edition of the Textus Receptust served as a basis for the Bible's King James Version (KJV). This text incorporates much later changes.
c. Critical Editions
Finally, the frequently cited witnesses include the following critical editions. These critical editions are described in more detail in a subsequent chapter entitled “Benchmark Critical Editions.”
NA28: Nestle-Aland 28th Edition.
NA27: Nestle-Aland 27th Edition (NA27 differs from NA28 only in the Catholic epistles and is only noted there.)
SBLGNT: The Greek New Testament: SBL Edition (Society of Biblical Literature).
THGNT: The Greek New Testament, Produced at Tyndale House.
When a witness is shown with brackets [ ] (for example, NA28[ ]), this indicates that the word is bracketed in the critical edition cited.
The present work aims to provide readers with increased transparency into variations between early manuscripts and the changes and additions of later manuscripts. It also compares four modern critical editions where they differ. Those variants not attested by any critical editions mentioned are assumed to be later changes or additions. These usually correspond to Byzantine (BYZ, 5th century and beyond) or a later text-form known as Textus Receptus (TR, 16-17th century).
In cases where two or more critical editions disagree, the footnotes are accompanied by all the frequently cited references in what is likely the original reading. This gives the reader a vital data set in cases where significantly different yet well-attested readings exist. In the cases of conflict between critical editions, the goal is to indicate how well-attested a reading is, yet without overwhelming the reader with details.
9. “Western Non-Interpolations”
The Old Latin (a b d e ff2 it) and Old Syriac versions (sys, syc) before 500 CE. often agree with the “Western” text of D(05) Codex Bezae. Because these are translations, they are not always accurate witnesses for the exact wording of the original Greek but are especially significant when they exhibit an absence of text as compared to other manuscripts, which may call into question whether the absent words, phrases, or even entire verses were part of the original text (autographs).
Some scholars have suggested that the combination of Old Latin and Old Syriac with D(05) indicates the original New Testament text, especially when the text is shorter than other forms of the tradition.u That is, when the absence in Old Latin and Old Syriac text corresponds with D(05), the text of other manuscripts is not original but rather later interpolations (additions). Thus, the omissions in the “Western” text-type have been called “Western non-interpolations.”
F. J. A. Hort affirmed that in many places, the Western text-type D’s omissions were original and that the absence of text attested by the combination of witnesses indicates a more primitive second-century text. Based on the tendency of the text to be expanded over time, he assumed that any significant omissions must represent the original form of the text. Prior generations of textual critics were trained in this perspective.”v
Recent scholars, including Aland,w Metzger,x and Fitzmyer,y have maintained that new findings such as 𝔓75 have refuted Hort’s hypothesis regarding text absent from D(05), the Old Latin, and Old Syriac manuscripts. Aland expressed the opinion that Hort’s line of thinking “can only be regarded today as a relic of the past.”z Scholars have based this view on the reasoning that the papyri discovered over the last century, such as 𝔓75, have not lent additional support to the omissions of D(05).
In response to scholars who rejected Hort’s views, Ehrman rebuts in his “Excursus on Western Non-Interpolations,” and concluded that new findings such as 𝔓75 are consistent with what Hort expected.a
In this edition, rather than omitting “Western Non-Interpolations” or enclosing them in double brackets as Westcott and Hort did, we have enclosed them in single brackets (except when NA28 uses double brackets). Single brackets indicate that the early manuscripts do not unanimously attest to the text.
Based on the manuscript evidence and the nature of the omissions, it is up to the reader to form their own opinion on whether they regard bracketed text as original in light of the witnesses cited.
10. Benchmark Critical Editions
The leading critical editions consulted for this translation as the authorities for the base text include the following:
SBLGNT, SBL Greek New Testament, Society of Biblical Literature & Logos Bible Software (2010)
NA27, Nestle-Aland Novum Testamentum Graece, 27th Edition, German Bible Society (1993)
NA28, Nestle-Aland Novum Testamentum Graece, 28th Edition, German Bible Society (2012)
UBS5, United Bible Societies’ Greek New Testament, 5th Edition, German Bible Society (2014)
THGNT, Tyndale House Greek New Testament , Tyndale House Cambridge & Crossways (2017)
The NA27, NA28, and UBS5 are published by the German Bible Society, which claims “the scholarly editions from the German Bible Society are the basis for almost all modern translations of the Bible.”
NA26-NA28 are reflective of the leading critical editions used for modern Bible translations. For approximately 33 years, between 1979 and 2012, the NA26/27 text was the dominant critical edition. Only the apparatus (notation of variants) was changed in the publication of the NA27 in 1993. With the publication of the NA28 in 2012, the text and the apparatus were revised in the catholic epistles with the implementation of a revised methodology as a first point of departure from NA26/27.
In these revised sections, the NA28 text is characterized as having more of a pro-Byzantine bias than the NA26/27, and it is not always clear how the selection of particular readings can always be justified. Additional sections of NA29 and beyond will be updated according to the revised methodology in future publications. Future editions of the AICNT will compare NA27 and other critical editions with whatever the latest edition of the Nestle-Aland text is.
Scholars such as Dirk Jongkind have noted, “It was only the 26th Edition of, published in the [19] 70s, that really marked a sort of independent look at all the variants. So then we have 26 and 27, which are identical, and then we are now at the 28th edition and have started a new method, highly dependent on statistics, and the text is sort of reflecting a slightly different approach in sections.”b Others, such as Daniel Wallace, have noted that the NA28 has “actually taken a step backward in cooperative effort across ‘denominational’ lines (in a broad sense).”c
The UBS5 critical text is identical to NA28. The difference between the two is the apparatus and some minor punctuation differences. A text equivalent to the NA28 is the Open Greek New Testament (OpenGNT, opengnt.com), which features an open license and differs from the NA28 only in the spelling of 61 words and in the word order for three verses.
SBLGNT (SBL Greek New Testament) is edited by Michael W. Holmes, who did his Ph.D. work under Bruce Metzger. It uses a modern text-critical methodology and guidance from the most recently available articles, monographs, and technical commentaries to establish the text of the Greek New Testament. The SBLGNT differs from the wording of NA28 in about 540 variant units. Often, SBLGNT omits text bracketed in NA28, noting the questionable text in the footnotes.
THGNT (Tyndale House Greek New Testament) is a recent critical text developed by Dirk Jongkind and Peter Williams with a team of scholars from Tyndale House. They “have taken a rigorously philological approach to re-evaluating the standard text.” This includes re-examining paragraph decisions and allowing more recent discoveries related to scribal habits to inform editorial decisions. A notable feature the AICNT shares with THGNT is more frequent paragraphing.
11. Deciding Between Critical Editions
The final rendering of specific verses is consistent with one or more of the critical editions noted (except for double-bracketed text supplemented by later traditions). To provide a neutral translation that most closely exemplifies the original manuscripts, we follow these additional two principles for choosing between variants in instances where the critical editions disagree:
(1) A manuscript that exhibits a shorter and less embellished text is likely more primitive than one that is more lengthy and expressive. Usually, the more concise text is rendered as the base text, with the more lengthy variant(s) documented in the footnotes.
(2) In the cases where variants have a significant theological implication, the text is selected that reflects the most neutral rendering. Theologically neutral readings are more likely original than those with more pronounced theological connotations. Variants bearing heightened theological significance are documented in the footnotes in such instances.
In places where there is a significant difference between the text of the critical editions, the variants are noted in the footnotes along with the respective critical editions (for example, SBLGNT, NA28, THGNT) that attest to each reading. In most instances, NA27 is only noted when it disagrees with NA28 (in the Catholic Epistles); otherwise, it is assumed that NA27 is consistent with NA28. The notation of NA28 also indicates UBS5, which is also not noted separately.
12. Software Resources
Various Bible software tools and resources were used, these include:
Logos Bible Software: Greek texts compared side by side with Text Comparison Tool include:NA28, SBLGNT, THGNT, BYZ (RP), TR(TR1550MR),Codex Sinaiticus, and Codex Bezae. Also used was the critical Apparatus for NA28, SBLGNT and THGNT. Additionaly consulted was A Textual Guide to the Greek New Testament: An Adaptation of Bruce M. Metzger’s Textual Commentary for the Needs of Translators.
Accordance Bible Software: Resources utlized include Greek Pro Collection, CNTTS Apparatus (Revised), The Comprehensive New Testament, Comprehensive NT Notes.
Olive Tree Bible Software: Resources utlized include ESV Greek-English Interlinear, BDAG: A Greek-English Lexicon of the New Testament and Other Early Christian Literature (3rd ed.), and UBS Handbooks for New Testament.
We trust that the prior information will be a helpful guide to your further understanding of the basis of the AICNT in the field of critical scholarship. Specifically, we hope the preceding sections provide a sufficient overview of how the critical text is determined and what additional information the critical apparatus conveys. We are pleased to provide you with critical text, neutrally rendered with GPT-4 and critical apparatus.
13. Methodology for Identifying Textual Variants
The method of identifying textual variants involved three primary steps as follows:
 
a. Text Comparison of Digital Editions
The first method utilized for identifying textual variants was to compare several texts using the Text Comparison Tool in Logos Bible software. These manuscripts include benckmark critical editions noted previously, the traditional text types of BYZ and TR and the early manuscripts of Codex Sinaiticus and Codex Bezae.
Where differences were found between manuscripts the difference text was translated with AI into English. This included entire verses, phrases or words depending on the number of words within the scope of the variant reading. The rendering in English was conducted to determine if the variants result in a different English rendering and to determine what the variant readings in English are.
 
b. Apparatus Review of Benchmark Critical Editions
Reference information pertaining to manuscripts variations was supplemented by the critical apparatus of the benchmark critical editions including the NA28, UBS5, SBLGNT, and THGNT. Of these the NA28 documents the greatest number of variants in the apparatus and includes notation of some early Latin and Syriac references.
A number of limitations were seen in reviewing the apparatus of these critical editions. None of them (including the NA28) were found to document all significant variants determined through textual comparison and the rendering of differences in Step (a). The NA28 often failed to note a variant associated with single frequently cited witness if that manuscript was later corrected. Additionally, the NA28 provides limited information about the early Latin manuscripts.
 
c Review of the CNTTS Critical Apparatus
The final step in identifying and document variants was to review the extensive CNTTS, which is textual database designed to include as much textual information from each manuscript as possible. The intent is to be an exhaustive computer database, with every manuscript reading, no matter how large or small, included. The database offers a great deal more information on any passage than is normally available to the reader through the apparatus of critical editions of the New Testament.
The CNTTS database is developed and maintained by the H. Milton Haggard Center for New Testament Textual Studies (CNTTS) and is under the direction of Dr. William Warren. Work on the database began in 2010 with the latest release of the 3rd edition in 2021 including corrections to numerous readings. At first the data for the MS readings largely depended on various sources but they claim data is now predominantly derived from fresh transcriptions and/or collations by CNTTS researchers.
Hundreds of additional variants were noted in the AICNT based on the CNTTS. The database also contributed supplementing the witnesses noted for already documented variants. The CNTTS also includes data on early Latin manuscripts greatly exceeding the NA28.
Despite their best efforts, the NA28 and the CNTTS may have occasional errors, which result in some inaccuracy in a small number of footnotes herein. In rare cases, references indicated contrary information about the attestation of particular manuscripts. When this occurred, notation pertaining to the particular manuscript in question was left out of the footnote. We are committed to improving the apparatus and correcting any inaccuracies in future editions.
Part II: AI Translation, Rationale & Methodology
14. AI as the Solution for Neutral Translation
The Bible is the best-selling book of all time, as it serves as the primary basis for many believers everywhere to understand God’s revelation. Yet, widely used English versions often reflect the inherent biases of translators and editors. The textual biases are then combined with an additional layer of bias reflected by editorial decisions to lead readers down a predetermined interpretive path. The opacity in these translations can sometimes render them closer to a commentary on the meaning of the text instead of a neutral rendering.
Faced with the challenge of human bias in translating Scripture, the question arises: where can one find an impartial rendition? Recent technological advancements offer an answer in the form of Artificial Intelligence. Specifically, OpenAI’s GPT-4 Large Language Model provides a highly precise and consistent tool for translating biblical Greek when programmed with appropriate guidelines. The AICNT thus serves as a viable resource for those seeking a transparent presentation of what the New Testament manuscripts genuinely convey rather than an interpretation shaped by human beliefs.
Technologies serve as versatile instruments with the capacity to produce both beneficial and detrimental effects contingent upon their application. When deployed with ethical considerations and stringent guidelines, AI becomes a robust mechanism for enhancing human understanding across various disciplines. In the humanities, machine-learning models can assist in translating ancient texts with a level of consistency and impartiality that is difficult for human translators to achieve.
The AICNT is generated directly from the Greek text provided through OpenAI’s API using GPT-4, the most advanced Large Language Model (LLM) available at the time of this publication. All that was given as inputs were the system message (specific instructions), the API settings, and the selected Greek text. All text incorporated in the AICNT, including alternate readings of variant manuscripts provided in the footnotes, was rendered utilizing AI as a translation tool.
The AICNT is generated directly from the Greek text provided. through OpenAI's API using GPT-4, the most advanced Large Language Model (LLM) available at the time of this publication. All that was given as inputs were the system message (specific instructions), the API settings, and the selected Greek text. All text incorporated in the AICNT was rendered from the GPT-4 LLM.
Utilizing its comprehensive knowledge repository and state-of-the-art computational architecture, GPT-4 demonstrated exceptional proficiency in accurately rendering biblical Koine Greek. AI Critical New Testament is developed for those who want a clear rendering of what the New Testament manuscripts say, as opposed to what human translators believe it means.
15. GPT-4 Rendering Methodology
To ensure the highest level of consistency and accuracy in the translated text, the parameters of the AI model were fine-tuned to minimize randomness.
Rather than use the web interface for ChatGPT, the API for Chat Completion, featuring additional controls, was utilized with custom Python code that defined specific parameters to yield a more deterministic and consistent output.
One such parameter is the Temperature variable. Higher values result in more random outputs, whereas lower values yield a more focused and deterministic result. For this translation, the minimum value 0 was utilized to generate the most deterministic output possible.
The custom API script developed for interacting with GPT-4 LLM included a system message to ensure the best quality output. The system message contained the instructions for translating the Koine Greek text provided. This had the instructions only to use the BDAG lexicon and to give a theologically neutral rendering to minimize interpretive bias.
It is well known that AI could give varying results if not prompted correctly with tight controls and specific instructions. For this reason, the previously described measures were taken to maintain tight parameters for the LLM to function. With such rules, the output was optimized for neutrality and repeatability. Once the program was developed, the only user input was the raw Greek text for generating the output rendering.
16. BDAG Lexicon
Regarding the use of the BDAG lexicon as the basis of the translation, BDAG is an acronym pertaining to A Greek-English Lexicon of the New Testament and Other Early Christian Literature, 3rd ed., edited by W. Bauer and based on previous contributions from F. W. Danker, W. F. Arndt, and F. W. Gingrich (Chicago: University of Chicago Press, 2001).
This comprehensive reference work lists thousands of references to classical, intertestamental, and early Christian literature. It is considered the most authoritative and widely used lexicon of the New Testament in the English-speaking world. BDAG contains a thorough and detailed analysis of the meanings and uses of every word in the Greek New Testament and other classical literature. The entries provide a range of information, including etymology, usage in various contexts, historical and cultural background, and cross-references to related words and concepts.
17. Rendered Word Equivalents
When rendering an English translation from the raw Greek text using the controls mentioned above, the GPT-4 Large Language Model (LLM) output sometimes fluctuated between two equivalent English words (highly equivalent synonyms) when translating particular Greek words provided within a sentence or phrase. This was the primary variation observed when using GPT-4 as a translation tool.
In the spirit of transparency, we are informing you the reader of these equivalency pairs. Some examples of these equivalency pairs in the output, with the word that is less frequently used in parentheses, are heavens (heaven), Gehenna (hell), beginning (start), taught (instructed), wealth (mammon), purchased (ransomed), gain (acquire), follow (go after), disturbances (insurrections), expectation (foreboding), and disciplined (punished).
In places where the output fluctuated between two equivalent words, it was observed that both word choices were always reasonable options, and those words with more familiarity to the general reader were usually selected. In cases where a less widely understood word choice is used for improved accuracy (for example, Gehenna), a clarifying footnote is added to give the reader a fuller understanding of the word's meaning.
18. Quality Assurance
When evaluating the output differences in providing GPT-4 with entire chapters compared to a single paragraph or verse, we observed that the number of verses translated simultaneously influenced the output. In general, rendering more text at a time improved readability but reduced accuracy. Rendering a paragraph or sentence at a time typically resulted in the best combination of high accuracy and excellent readability. In instances when the Greek text is more complex or abstract, processing a single verse, phrase, or word at a time was employed to achieve an even more accurate output. Inaccuracies were generally resolved by inputting smaller bits of Greek text for GPT-4 to render at a time.
The implementation of the critical apparatus plays a crucial role in ensuring quality. The process of constructing the critical apparatus and the numerous renderings required to compare variant readings between manuscripts was a verification step that improved accuracy in rendering fewer words where initially multiple paragraphs were rendered at a time. This resolved issues where the rendering of multiple paragraphs together tended to be less accurate in places where textual variants exist between manuscripts. Thus, the accuracy was further improved during the process of meticulously reviewing every textual variant in constructing and examining the critical apparatus.
The master file for the translation was configured in a format suitable for use with various software applications, including those used by translators, to ensure the quality of translations. Such software features several quality checks to check punctuation, capitalization, and grammatical issues, among other things. The process of checking for consistency of translated words between Gospel parallels and re-rendering shorter text lengths with GPT-4 to improve consistency resulted in an additional layer of quality assurance.
Dustin Smith, Ph.D., biblical scholar of Early Christianity at Spartanburg Methodist College (Spartanburg, SC), reviewed the translation for accuracy and readability. Professor Smith conducted a complete review of the rendered text, but did not edit the translation. When an issue was identified, shorter lengths of Greek text were rendered to resolve the inaccuracy. In the vast majority of cases, this resolved the issue without the need for a footnote. In some cases, a footnote was added to provide clarification.
19. Clarifying Footnotes
Various types of footnotes were added to give clarification and greater transparency. This includes:
(1) Providing Scriptural references for the verses that quote the Old Testament and other texts. “LXX” in the footnote indicates references particular to the Greek Septuagint
(2) Providing the abridged BDAG definition in places where the translated Greek word may have a wide range of meanings or may not be fully expressed by the English word rendered
(3) Providing an alternative reading of a specific word or phrase. This is in cases where the rendered word or phrase might pose some difficulty in comprehension, such as an idiomatic or cryptic expression, or where the word or phrase can be expressed or understood differently
(4) Providing the alternate GPT-4 rendering in those few cases where GPT fluctuated between two significantly different renderings due to syntactical or textual ambiguity
(5) Providing a clarifying explanation about the theological neutrality of particular verses.
20. Formatting
The AICNT does not use capital letters for pronouns that refer to God. This aligns with the original texts, as Hebrew does not differentiate between upper-case and lower-case letters. Additionally, the initial Greek manuscripts were composed entirely in capital letters.
Especially regarding explanatory section headings, the formatting exhibited in many popular translations adds a layer of subjectivity and editorial bias that is avoided here. Since section headings are not used, this translation relies on more frequent paragraphing (similar to the Tyndale House Greek New Testament) to help the reader navigate and search through chapters more efficiently.
AICNT abstains from particular formatting conventions of modern translations used to distinguish and classify types of text, such as quotations or poetry. Italicized words are only used in cases where the AI model indicated that it added helper words in English that are not in the Greek text.
We maintain traditional chapter and verse numbering consistent with modern critical editions while structuring the text in smaller, more frequent paragraphs to enrich the reading experience. The paragraphing of the AICNT exhibits a balance between the conventional paragraph layout of Bible translations (exhibiting large paragraphs) compared to those translations that provide a verse-by-verse layout (placing each verse on a new line). Smaller groupings of text are also valuable for aiding in readability and comprehension by causing the reader to pause more often and digest a shorter length of text before proceeding.
21. Punctuation
The earliest manuscripts were written in uncial, which is essentially all capital script without punctuation or word separation. These features including punctuation were introduced into minuscule manuscripts (having text with both upper-case and lower-case characters) which began to appear in the 9th-century. The earliest dated minuscule manuscript is the Gospel minuscule 461 from the year 835.d
The Punctuation of this edition is primarily consistent with one or more of the critical editions in the section entitled “Benchmark Critical Editions.” Rendering the non-punctuated text with GPT sometimes informed what punctuation was preferred. The punctuation was also influenced by what was best for documenting multiple variants by using the punctuation that facilitates the critical apparatus featuring bracketed text that is interchangeable for multiple variants.
Online Viewing
The latest updated version of the AICNT can be viewed at https://aicnt.org or with enhanced functionality at https://gpt.bible. If any errors are identified in the translation or footnotes, we encourage you to email us the details at [email protected]. We intend to implement updates before the release of any future versions or editions.
Additional Project Information
Beyond this, we intended to publish translations of the Old Testament and also a complete Bible. Your support is appreciated for this and future publications! Please consider donating at aicnt.org or gpt.bible.