Good enough A.I.

Typical assembly line robots need not be mobile. The work can be brought to them. Clean rooms are only required for certain operations. Robotic surgery needs to be precise but it need not lift heavy loads. My A.I. specialists only need to be good enough for their specialized tasks. The agent's required memory, speed, processing power (I.Q.), architecture*, and cost depends on the specialty.*

* There's way more to A.I. than neural networks. See my blogs of 9 and 13 September 2010. (The majority of these subfields were needed in order to create A.s.a. H.) The big AI companies today have much too narrow a focus.

The limited ability of LLMs to learn temporal sequences*

 Huang et al** argue that "transformers incur a quadratic attention cost, limiting their ability to model long spatial and temporal sequences..."

* see my blog of 1 Feb 2025

** Jihao Huang et al, LADY: Linear Attention for Autonomous Driving Efficiency without Transformers, arXiv:2512.15038v1, 17 Dec 2025.

AI progress

It remains to be seen if transformer neural networks (LLMs) can learn (approximate) the human cognitive architecture.  Even if they could, humans are not, themselves, rational* nor do they possess a sound value system**.  Noorbakhsh Amiri Golilarz et al have a good discussion*** of some of the things that are missing with LLMs. With my A.s.a. H. architecture I have tried to address the shortcomings Golilarz et al outline. Self-monitoring, allocation of resources, and meta-cognitive awareness are discussed in my "Experiments with Asa H" in my book Twelve Papers.**** The learned case-base (knowledge representation/concept structure/memory) is corrected/repaired, grows, updates, and evolves over time. Deployed on robots Asa is fully embodied (sensorimotor grounding). The hierarchical case-base is distributed across multiple timescales.

* see Predictably Irrational, Dan Ariely, Harper, 2008.

** We are, after all, apes. (Gentle, very modern apes as Erika would say.)

*** Bridging the Gap: Toward Cognitive Autonomy in Artificial Intelligence, arXiv:2512.02280v1, 1 Dec. 2025.

**** Available under "book" at www.robert-w-jones.com

ESP32

Science is, inherently, an open-source activity. So, with the threat to the open-source character of the Arduino ecosystem I am now buying some more ESP32 development boards. 

What is real?

For some of us something is not real if it can not be measured (sensed/observed). For others something is real if our best models of the world require its existence. The quantum wave function may then be real. (Such is certainly the case in both the Bohm interpretation of quantum mechanics and in the Everett interpretation and perhaps more generally.) "Real" might best be considered to be a vector (multi dimensional) quantity.

There are even some things like the magnetic vector potential which are perhaps more concrete than the quantum wave function but which (being gauge-dependent) can not be easily measured like the magnetic field itself.