Very interesting, and partially solves the issue of being able to run such an
extensive model. So, in order to send Chat GPT-4 back into the past, we must also send the Azure Supercomputer which it's required to run/train on as well. Next question is this, which is a fundamental one. Does
all the data Chat Gpt-4 is trained on goes back as well? If so, what this means is we essentially have given circa 1990s humanity an oracle, as it has knowledge from its training data up towards 2021. This would have profound effects on the world depending on who gets the technology (luckily A.I. researchers in this case) and what they intend to do with it.
Let's say that it eventually ends up in the hands of Microsoft, in a similar situation to how Open AI did. No doubt, aside from greatly improving (hopefully) their soon to be released Windows platform, reverse engineering and speeding up their tech quotas, what exactly would they do with certain sensitive information Chat GPT has about the future? I think it's for this reason its full nature would remain hidden from the public and even classified. Having information on future leaders, companies, world events, etc, is much too powerful for anyone to truly utilize ethically. Ideally, people would use it to avoid the worst outcomes which may take place, and perhaps, Microsoft decides to inform the U.S. government of certain events which Chat GPT-4 knows about which greatly affect the U.S. and the world.
I imagine in this scenario; MS also supersedes Google by simply knowing of its future potential and utilizing Chat GPT-4's capabilities to make a truly advanced search engine. The A.I. researchers of the time will have their field
greatly expanded, to the point where our current year/world would be completely unrecognizable to us. Having access to advances in the future greatly accelerates the pace of technology and its implementation. The Singularity might just be taking place as of right now, or already might have taken off beforehand. Instead of being impressed by Watson, we'd might literally have C-3PO or Data. At the very least, this would make the "frozen/prison" AGI theory where its kept shackled but granting advancements to humanity a reality.
The most dangerous part about this, again, is the potential of people misusing the stored knowledge or generating absolutely perfect propaganda. Especially when considering 1990s tech which is primitive by comparison. I think such is unavoidable though, so it's a question of whether a more advanced world at the cost of those (elites) utilizing the knowledge of future events being worth it.