[Mordisquitos]

I very much doubt it, at least with regards to GPT-n style models. In this particular example, it is not actually being fed any knowledge about the physics of the world. Rather, it is being fed the texts of a very specific subculture (that of scientific research and publishing) which is based not only on the prior sensory experiences of human beings, but also following the arbitrary agreements and expectations of the members of the scientific community that have developed over time. Even the most intelligent human minds would be unable to learn anything meaningful from scientific papers if they were expected to read them from scratch having been brought up completely isolated and without any prior knowledge.

On the other hand, an interesting possibility with well-designed text-mining and AI models would be for them to generate valid hypotheses that hadn't been contemplated earlier, based on the massive corpus of scientific publications. The model may be able to find possible correlations or interesting ideas by combining sources from different fields that would normally be ignored by the over-specialised research community. However, in that case the model wouldn't be valuable for providing answers—rather, it's value would be in providing questions.