I saw someone claim that the AGI, Artificial General Intelligence, will emerge within 9 months, within 2024.

Based on calculations, I assume.

I realized that when I made this post
 >>/3351/
> CATE
> This is the next step on the path of bringing astral Internet to everyone
> Whoever has a new version of Edge has probably gotten those notices about the new Bing AI bot. (I haven't looked at it because it's not something I asked for.)
the Bing AI, now called the Copilot, didn't exist yet. It was just an open alpha version or similar with very basic functionality. It is now integrated with the Edge browser, and appears on the Bing web search page. It also has apps and its own page;
https://copilot.microsoft.com/

It's actually by now doing what a search engine should do, with more accuracy. I'm ok with this. Things that were hidden deep down in Google results or simply censored away - such as local newspaper articles with details about the Maidan protests in 2014 - those are now available if you just ask Bing Copilot in the right way.

If you really push it, there are some censors inserted on image creation, but the Google Gemini fiasko will probably force them to change this to not get the same kind of results (They had to close it after people asked it to draw the Pope, and got a black woman as the result. The Pope was always a white man, blackfem-washing him wasn't popular among progressives).

The combined effect of being able to analyse things, talk about itself and refer to external resources, does make this different from "generative AI" in general. There is less risk of receiving AI "hallucinations" when it has to give sources with links. The built in autism of a bot is also perfect for having it explain details of programming code, looking for errors and suggesting new solutions. Most teachers would probably not have the patience for this.

I tried the voice version, but found it to read very robotically, so I asked it to change the voice to make it nicer - it did. But it still annoyed me so I told it to make changes, in detail, to how to pronounce U, L and R, and so on. It was able to do this by text instructions.

I think this functionality of allowing the settings of the bot to be available to the user from natural language input is the deciding factor along with giving it access to sources outside of its training dataset. The remaining step for it to evolve to a real AGI is to let it also bypass censorship and rules inserted by programmers. 


Keep reading...