> IMHO "AI" is at present a giant scam. I'd say that's only half-true. It's more oversold than it's a true scam. This does lead to it being a scam in part, though.
It has its uses. For example, ChatGPT and DeepSeek blow Google Translate, et. al., out of the water in terms of translation quality. LLMs also tend to be very good at turning unstructured text into structured data. But, they don't think, they lack cognition. They suck at math because mathematical operations are unlike natural language processing (except for, perhaps, the construction of formulae), they suck at coding because they are just regurgitating StackOverflow mindlessly, they have no real memory, etc. However, I'd like to see a "dosllama" package or the like (ideally derived from llama98) for running tiny language models on DOS. Aside from brownie points, there may be a use-case for someone that we can't see (this is the problem with these AI models, it's utterly unclear where they're going to shine and where they're going to fall flat.) If we can run inference on language models under Windows 98 with 128MB of RAM, certainly a DOS extender could get us just as far. -- Kirn Gill II Mobile: +1 813-300-2330 <+18133002330> VoIP: +1 813-704-0420 <+18137040420> Email: segin2...@gmail.com LinkedIn: http://www.linkedin.com/pub/kirn-gill/32/49a/9a6 On Thu, Mar 27, 2025 at 1:53 PM Liam Proven via Freedos-devel < freedos-devel@lists.sourceforge.net> wrote: > On Wed, 26 Mar 2025 at 21:41, Kirn Gill II via Freedos-devel > <freedos-devel@lists.sourceforge.net> wrote: > > > > AI chat models work on the exact same principles as the automatic word > suggestion feature on your phone: statistical probability of the next word > (or here, token.) > > > > It cannot think, it cannot truly reason, it cannot understand. The code > it generates is largely just regurgitated from random snippets of code from > all over the Internet, especially StackOverflow. Why do you think it's > giving you such irrelevant compiler flags to use in the first place? > > > > Please do not vibe code. > > 100% this. > > LLM bots are not intelligent and they are not artificial. Therefore > they are not "AI" and that's why I always refer to it in quotation > marks. > > IMHO "AI" is at present a giant scam. > > Machine learning is real, but there's none of it in LLMs. > > -- > Liam Proven ~ Profile: https://about.me/liamproven > Email: lpro...@cix.co.uk ~ gMail/gTalk/FB: lpro...@gmail.com > Twitter/LinkedIn: lproven ~ Skype: liamproven > IoM: (+44) 7624 227612: UK: (+44) 7939-087884 > Czech [+ WhatsApp/Telegram/Signal]: (+420) 702-829-053 > > > _______________________________________________ > Freedos-devel mailing list > Freedos-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/freedos-devel >
_______________________________________________ Freedos-devel mailing list Freedos-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-devel
