The Wartburg Project

March 25th, 2024

106. Artificial Intelligence and the Evangelical Heritage Version

Since AI is a relatively new trend, I assume AI was not used to produce the EHV, but do you foresee that AI will be used to produce improved editions of the EHV in the future?

That is a hard question, which can only be answered with a touch of prophetic insight, but the short answer is “No” for reasons discussed below.

The first reason for a “no” answer is that AI of the type needed to do Bible translation does not exist and, in the view of many experts, never will. As of now, no true Artificial General Intelligence (AGI) exists. It remains in the realm of science fiction. In the wide sense, AI is the use of computers and machines to mimic the problem-solving and decision-making ability of the human mind. The difference between existing low-level AI and advanced AGI is the difference between Machine Intelligence (data intelligence) and Human Intelligence (emotional and spiritual intelligence). It is similar to the difference between knowledge and wisdom.

AI has been around for a long time (the term was coined in 1956). The 1968 movie 2001: A Space Odyssey was already warning against the alleged dangers of AI more than 50 years ago. You are interacting with AI in your attempts to converse with Siri and Alexa and to deal with the AI robots that answer your calls to the customer service departments at various businesses you deal with. AI is now being used to produce self-driving cars, and it is used with GPS systems to provide directions. It is clear that present AI often comes up short of common sense in these systems.

If your experience with AI phone-answering robots has left you with the conviction that you get better answers from an AI robot than from a reasonably competent human, whose first language is not English, you are ready to plunge full speed ahead into an AI world. If you still prefer even an average human at the other end of the line, you are not ready for an AI Bible translation.

Bible translation requires a very advanced AI system called Natural Language Understanding (NLU) or Natural Language Processing (NLP). Advanced algorithms and improvements in computing power and in storage capacity have increased the ability of AI to assist humans in Bible translation. A recent article in Biblical Archaeology Review provides a nice introduction to the topic.1 We will quote a summary of its main conclusions.

NLP software may improve quality and efficiency, but Bible translation is best done by human beings, using software as a tool. Producing a first draft is only about 10% of the work involved in creating a translation, whether that draft is produced by humans or NLP. Even if NLP could produce a perfect translation—and it cannot—the process of translating the Bible into a local language creates a sense of ownership and community, bringing together the translation team and the group they serve, while also preparing that community to study and teach using the translation it produced. If a group for whom a translation is intended does not feel connected to it, they may never use it. Software and resources are best used in ways that help translation teams work together.  …

While NLP can rapidly produce results that are often good, these results still need to be verified and edited by human beings before they can be considered trustworthy.2

AI can assist translators with functions such as spell check and grammar check, checking consistency of translation, producing a rough draft, and other forms of data processing. Computer programs assisted with all of these function in the production of the EHV. AI is very good at the machine aspects of translation. What it is not good at is the human aspects of translation.

The BAR article pointed out AI’s inability to provide the social and fellowship aspects of translation. Even more crucial is AI’s inability to provide the emotional and spiritual aspects of translation, aspects like faith and a knowledge of God, conscience and moral understanding. While any competent linguist can produce an acceptable literal translation, the most important attributes for a Bible translator are faith which submits to God’s law and gospel and an experience of the grace of God. AI may someday possess something that can be called a mind. It will not possess a soul.

[1] “Artificial Intelligence and Bible Translation” by Jonathan Robie. BAR, Winter 2024, p 62-64. The article is worth reading.

[2] This sentence hints at another severe problem with AI: AI does not think; it produces the results that it is programmed to produce, but it produces those results more quickly than a human could. 

If AI were really a form of intelligence superior to human intelligence, users would trust its results. They do not.

While I was thinking about this article, I heard a discussion of AI on a radio talk show. The host warned listeners against trusting AI by citing a recent example of AI. A company had produced an AI to help in its hiring processes. The problem that developed was that the AI was hiring too many men. Their conclusion: obviously the AI was bad and had to be reprogrammed. Users were warned to always check AI to be sure that it is achieving the intended results. If the AI was really a superior form of intelligence, its decision to hire too many men would be evidence of its belief that those applicants were the best people to fill the jobs. But such an outcome cannot possibly be correct; since it clashes with the intended results,  the AI’s way of thinking had to be rejected. It was back to the drawing board for the AI system. What good is an AI program if it is not producing the intended, pre-programed results?