Jean Louis <b...@static.rcdrun.com> writes: > * Van Ly via Emacs news and miscellaneous discussions outside the scope of > other Emacs mailing lists <emacs-tangents@gnu.org> [2025-03-22 19:07]: >> >> What may be the design constraints on computing excellence in the Starship >> ITS AI hardware going to Mars? >> >> I asked Grok3 the following: >> >> https://x.com/i/grok/share/hAaMjRLuFkrEvz3V5m87yPest >> >> and because Starship AI has not been thought of that far then made and >> tested I assume it would be a hybrid general purpose Cray-1 modernized type >> and Psychological AI specialized Dojo-1 and ask here before asking Grok3 >> further. >> >> It has to do emacs, celestial mechanics, fluid dynamics and cryptography. >> >> It has to be real by 2030. > > Maybe in some future, we get some true creation by robots. >
Inside 10 years when there are enough Tesla Optimus fully self driving Cybertrucks, robots may reconstruct a model Pyramid in Egypt, or lift a plot line from Asimov's SciFi and perform a detective story thriller on the case of the OpenAI whistleblower whose mother appeared on Tucker Carlson's VPodcast. Automatapragmata would have more free will than robots. > Today, I have genuinely tried to make LLM innovate for me, but all it gave me > is just generated bullshit. > > https://pageassist.xyz/share/cm8le9br0008u8c5q5irqd980 > > Even though such innovation exists, it seems totally out of any reality. > > Then I tried again with some better model with thinking and Internet search: > > https://chat.qwen.ai/s/fc1e8256-6bd6-4375-a555-f51260afa6a3 > > It basically just gives guidance on what I am going to do, based on > LLM knowledge and Internet search. It is referencing existing > innovations. I got no documents, no drawings, nothing. > > It told me nothing new, apart from URL references to other > innovations. > > It did not create nothing so far. > I asked Grok3 the following: https://x.com/i/grok/share/OgPKGljuCRXBQ7eNEwd0ivYDy and learned. Those two example links you provide I can imagine a monthly $20 thousand or $200 paid subscription to Google or Microsoft would enhance the decision making process among professionals of the grade that stood with Steve Jobs on what to do to achieve a goal like iPad and iPhone. Currently, headlines say Apple loses $1+ billion a year on imitating Netflix and the Apple project to build a bevcar spending $1 billion a year didn't even build one after many many fruitless years. > Let's see in next years if something change. > > If you try with some better model, let me know. https://oecs.mit.edu The above has an entry for theory of mind but not consciousness. -- vl --- via emacs-tangents mailing list (https://lists.gnu.org/mailman/listinfo/emacs-tangents)
