The overwhelming majority of Sumerian cuneiform tablets have yet to be completely deciphered. Given the sheer number of tablets - running into tens of thousands - coupled with the complex nature of the language, the task is daunting. With a limited number of experts who can understand this ancient script, and the content of the writing predominantly being mundane accounts and bureaucratic records, the process is slow.
Advanced language models like GPT, however, have the potential to not just translate these voluminous documents, but also to extract concealed contexts and relationships from these artefacts. This would allow for the recreation of an ancient Sumerian mindset that could provide us with insights into their civilization.
Provided with sufficient descriptive data, similar feats could be achieved for extinct languages, the communication patterns of a pod of whales, the social behaviours of a tribe of chimpanzees, or a transcription of biochemical reactions. This would enable the creation of a translation interface capable of interpreting datasets far beyond the comprehension of the human mind. Such advancements are set to unlock expansive new realms of knowledge in the coming decade.