OpenAi Translation Testing

A question.

Is OpenAi, or similar tools unbiased and linguistically accurate enough to assist in original ancient language to current written language translation?

Testing.

• Create a Python program to connect and query the OpenAi API
• Get a list of the current Artificial intelligence (Ai) models available in the toolset
• Test a few of the models to see which will suit our needs
• Write an initial test for response for a current popular query (English only). Completion tool.
• Test an Active Voice translation (English only). Edit tool.
• Test a first translation (English to Indonesian). Edit tool. Check given translation by converting back to English with established online sentence structure translator.
• Test second translation (Ancient Greek to English). Edit tool. (John 3:16-18)
• Test third translation (Ancient Greek to Indonesian, Indonesian to English). Edit tool. (John 3:16-18). Check given Indonesian translation by converting back to English with established online sentence structure translator.
• Check results. Legibility. Accuracy. As a plain native English reader.
• Extra. Is the Ai model trainable, can we add new written languages? https://analyzingalpha.com/openai-api-python-tutorial#Fine-Tuning_Existing_GPT-3_Models

Outcome and conclusion.

• The public models are somewhat unreliable, almost unethical, in diverting to gibberish to satisfy input and answer; likely bias
• Automation still requires vetting by professional / expert translators in any given language

As I sought counsel from known and trustworthy translators, I was reminded that Scripture is twice inspired, the first time when God-breathed it into His original text via the voice and hands of His authors, and the second time when we read it, write it, recite it. I'm satisfied with that.

2 Timothy 3:16-17