From: mk_thisisit
The question of what consciousness is has been a subject of research for decades [00:00:22]. While some theories suggest a link to quantum phenomena, it’s also possible that consciousness is a problem of classical mechanics [00:00:39].
Classical vs. Quantum Mechanics
The possibility that consciousness is rooted in classical mechanics rather than quantum mechanics cannot be ruled out [00:00:39]. While arguments might not strongly push towards one over the other, there is freedom to include quantum mechanics in these considerations [02:11:12].
One perspective, championed by Roger Penrose, posits that human consciousness arises from quantum phenomena within the brain [01:57:32]. This idea is intriguing because it implies moving beyond traditional classical electrical engineering and differential equations [02:19:00]. Achieving immense computational power could involve creating systems that genuinely rely on quantum mechanics, rather than just using it within classical limitations [02:42:00].
However, determining the exact role quantum mechanics plays in consciousness is considered a secondary issue [02:21:21]. It is not possible to definitively state whether consciousness is purely a matter of classical or quantum physics without a deeper understanding of its relationship to other phenomena [02:31:00].
Neurons and Collective Behavior
A key question in understanding consciousness is whether neurons should be considered independent systems or as elements exhibiting collective behavior [00:00:31]. Language, for instance, might be a collective behavior describable at a higher level, not relying on every detail of lower-level phenomena [01:07:00].
The idea of “essence” over “details” is crucial. Modern physics aims to capture the essence of a phenomenon rather than all its intricacies [06:42:00]. This approach has been found more advantageous in creating artificial neural networks, which imitate biology’s essence without slavishly copying every minute detail [05:51:00]. Trying to slavishly imitate every detail consumes computer resources and provides little insight [07:01:00]. When the essence is found, it can be applied universally [07:10:00].
Collective systems, such as phase transitions, have been instrumental in understanding biology and physics [08:31:00]. The concept that “the whole is greater than the sum of its parts” applies here [08:45:00].
Limitations of Current AI and Biological Learning
Current learning systems in AI often do not consider synaptic weights as dynamic variables at the same level as dynamic activity variables [01:30:00]. This separation might not preserve the computational power developed through biological evolution [01:47:00]. There is an expectation that future developments in neural networks will focus on understanding the real learning algorithms in biology and how to incorporate their essence into engineering [01:33:00]. Biological synapses, for instance, exhibit noise and do not change in infinitesimal quantities, unlike many AI models [01:09:00].
Future Outlook
Predicting the future development of neural networks, particularly concerning how neuron activity and machine states will be described, remains challenging [01:53:00]. The choice of dynamic model (e.g., continuous variables or action potentials with delays) will significantly influence future possibilities [01:13:00]. If we don’t know which model will dominate, we have no idea where the field will be in ten years [01:24:00].
Ultimately, the Nobel Committee’s decision to honor work on neural networks is seen not just as a tribute to AI but as an emphasis on science in a broader context, enabling diverse engineering applications [02:51:00]. Physics should look at seemingly unsolvable problems rather than limiting itself to narrow classifications like “AI or not AI” [02:37:00].