Question: could you train a model like this from before a known but important scientific breakthrough happened and see if it is able to work it out. At least then you know the answer yourself and know it’s something that can be reasonably worked out.
Talkie: a 13B vintage language model from 1930
111–120 of 350 posts
Re: Talkie: a 13B vintage language model from 1930
#112Re: Talkie: a 13B vintage language model from 1930
#113Earlier quoted context omitted.
We're going to be able to simulate any period of time (for which we have or can simulate data) and transport ourselves there. This is going to be amazing. At some point I'm going to data mine all of my old internet AIM and IRC chats, emails, old laptop hard drives, essays, etc. and go nostalgia/core on it. I have old video and audio recordings too. Lots of stuff for reminiscent inference. Every day I'm finding it har…
This assumes that written data from a particular time period actually reflects what it was like in that time period, and isn't highly biased to select for, say, particular socioeconomic classes.
Re: Talkie: a 13B vintage language model from 1930
#114*What do you think the world will look like in 2025 TALKIE-1930* According to a forecast by an eminent statistician, the world will, in 2025, support a population of 6,600,000,000. All Europe will be one vast network of railways, and travellers will be able to go from London to Constantinople in forty hours, and from London to Peking in a week. The globe will be girdled with telegraph wires, and messages will be sent…
> and travellers will be able to go from London to Constantinople in forty hours By the 1930s, Constantinople been a long time gone. It had been Istanbul not Constantinople for centuries by that point.
Re: Talkie: a 13B vintage language model from 1930
#115Earlier quoted context omitted.
That's nobody's business but the Turks. Why did Turkey become Türkiye but Japan didn't become Nippon (or vice-versa!)? It's all very confusing to me.
Why did Turkey become Türkiye? I think mostly because they asked. I’m guessing that Japan/Nippon is enjoying the fact that English speakers use the Chinese name for Japan and the Sanskrit¹ name for China. It’s much like the Czech Republic became Czechia, although part of that was Czech speakers wanting to stop referring to their country as an adjective² (the Czech phrase for Czech Republic was often shortened to just…
Re: Talkie: a 13B vintage language model from 1930
#116Earlier quoted context omitted.
We're going to be able to simulate any period of time (for which we have or can simulate data) and transport ourselves there. This is going to be amazing. At some point I'm going to data mine all of my old internet AIM and IRC chats, emails, old laptop hard drives, essays, etc. and go nostalgia/core on it. I have old video and audio recordings too. Lots of stuff for reminiscent inference. Every day I'm finding it har…
Combining AI with VR we can even achieve something like time travel ;-)
This is going to be so amazing.
Re: Talkie: a 13B vintage language model from 1930
#117Earlier quoted context omitted.
This assumes that written data from a particular time period actually reflects what it was like in that time period, and isn't highly biased to select for, say, particular socioeconomic classes.
Yep. Until the very recent boom of social networks, everything published is, by definition, the product of the educated and the upper classes. The farther back you're going, the more estranged from ordinary people you are. In the Middle Ages, you'll have nothing but texts about the adventures of saints and kings.
Re: Talkie: a 13B vintage language model from 1930
#118*What do you think the world will look like in 2025 TALKIE-1930* According to a forecast by an eminent statistician, the world will, in 2025, support a population of 6,600,000,000. All Europe will be one vast network of railways, and travellers will be able to go from London to Constantinople in forty hours, and from London to Peking in a week. The globe will be girdled with telegraph wires, and messages will be sent…
Very beautiful, and very sad.
LLM aside, it tracks that with a civilization in truly dire straits, portrayal of the future would trend increasingly more towards being positive and fanciful: because at some point things would be so bad, that imagining the future will be even worse becomes a deadly thing for the modicum of hope required to even stay alive, let alone push forward.
I personally always think we have a lot of fat to trim before we get there. Our descendents can have a wonderful quality of life even if a lot of institutions and supply chains regress. The era of summoning food from a handheld computer might go away and we'll still be pretty well off (if not strictly better off)
Re: Talkie: a 13B vintage language model from 1930
#119Earlier quoted context omitted.
We're going to be able to simulate any period of time (for which we have or can simulate data) and transport ourselves there. This is going to be amazing. At some point I'm going to data mine all of my old internet AIM and IRC chats, emails, old laptop hard drives, essays, etc. and go nostalgia/core on it. I have old video and audio recordings too. Lots of stuff for reminiscent inference. Every day I'm finding it har…
Combining AI with VR we can even achieve something like time travel ;-)
Re: Talkie: a 13B vintage language model from 1930
#120USER: What will the future hold for computers? TALKIE-1930: Computers in the future may be employed in offices where calculations are required to be made, and where the nature of the business does not demand a very high degree of knowledge. The occupation will be a steady one, and the remuneration will be sufficient to afford a comfortable livelihood. The hours of attendance will probably be from ten to six. USER: an…
Also, I'd like to see the inverse of it: a model trained only on the last few weeks - or even minutes - of published information, or only on scientific papers published in the last year or two. Should present some interesting psychoses!