On Wed, 2026-09-02 at 17:59 +0200, Quan Tesla wrote: > LLMs are trained on 20 TB of text, which is 20,000 times as much > conscious experience as a human. If that's the definition you want to > use. You tell me. > > With regards consciousness experience, we may consider qualia as > related to states of consciousness, including for the possibility of > a state where consciousness cannot be measured.
My opinion is that the huge amount of data required to train LLM is disqualifying them and perhaps the entire large language model approach (at least when applied to textual data which is never that large: even all of Shakespeare's writing should fit in a few gigabytes, not terabytes). "expert system" inference engines are in that aspect much better, they can inspect at runtime their call stacks (using https://github.com/ianlancetaylor/libbacktrace ...) if you do follow consistent calling conventions (which also is needed for efficient garbage collection implementations). I am French and trying to develop such an engine (see amateur website refpersys.org). See also https://arxiv.org/abs/1109.0779 regards -- Basile STARYNKEVITCH <[email protected]> 8 rue de la Faïencerie http://starynkevitch.net/Basile/ 92340 Bourg-la-Reine https://github.com/bstarynk France https://github.com/RefPerSys/RefPerSys https://orcid.org/0000-0003-0908-5250 ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T906c878e4ba411be-Mf8a90e54489aae77f394a686 Delivery options: https://agi.topicbox.com/groups/agi/subscription
