From Turing Tests to Tool Libraries
Morning routine. Plan Today completed. Cliffe call on KAI matters. AI Infrastructure wrestling. Book Club on Melanie Mitchell's AI book. Ashley catch-up and converter tool triumph. Skipped yoga. Suds & Suds walk. Lyft home. Formal Logic.
Book Club tackles AI consciousness while a quick hack delights a colleague
Events and activities that occurred on Thursday, February 12, 2026
The morning call with Cliffe covered the usual ground of keeping Kitbag Africa International moving forward through a crazy month. Leadership meetings need scheduling with Ameerah and Janet before my New Zealand departure on February 22nd. The USA financial proxy work requires a PowerPoint presentation for outreach to potential fiscal sponsors. Cliffe's father faces a hip replacement that government insurance may not fully cover, and the financial strain of supporting both his father's care and his daughter Jewel's school fees weighs on him. Meshak continues working on the website, and I need to review his updates and provide feedback.
 |
| Random coffee picture |
The JRM Book Club convened to discuss Melanie Mitchell's Artificial Intelligence: A Guide for Thinking Humans. Rajeev introduced Mitchell's background—PhD under Douglas Hofstadter and John Holland at Portland State, expertise in complexity and emergent behavior. He recounted how Mitchell attended a Google meeting where Hofstadter expressed being "terrified" about AI developments. The discussion ranged across the nature of intelligence and whether the Turing Test serves as an adequate measure. Jon suggested that attributes of superintelligence might include multitasking ability, larger data access, and faster processing. Jim raised concerns about anthropomorphism in our definitions, while Jon mentioned octopuses as alternative intelligences that demonstrate problem-solving without passing any human-centric test. Rajeev traced AI history from the 1956 Dartmouth Workshop through the cycles of AI winters and summers to present breakthroughs. The conversation turned to consciousness, emergent behavior, and the question of whether AI systems might become goal-generating rather than merely goal-achieving. The guardrail discussion explored how effective constraints might be implemented when the emergent and probabilistic nature of modern AI systems makes them inherently difficult to control. Peter emphasized that humans have internal moral guardrails developed through social evolution, but Jim questioned whether any guardrails could be universal across cultures. The ikigai risk emerged as a concept—the possibility that AI could undermine human purpose even without causing extinction.
Ashley and I caught up in our Thursday evening session. He faced a technical demo the next morning and was building custom libraries with only hours to spare. The bigger conversation centered on his frustration with limitations in PTC's CAM optimization tools compared to competitors. He needs a parser that converts ISO format files to XML with a simple UI—functionality surprisingly absent from their product. I recommended Claude for the development work and suggested the $100 Max subscription to overcome usage limits. Later in the evening, I sent him a converter tool I'd thrown together in about ten minutes based on an overheard conversation. His response escalated from "This is awesome!!" to testing it with Kennemetal tool library files to the triumphant "IT WORKSSSSSSSS" and "you have ruined me!! I won't get any work done now." We both had beers—he brought one back to his desk, I walked to Suds & Suds.
The decision to skip yoga simplified the evening. The class supported a benefit for a worthy cause, but my capacity to tolerate requests for money has reached its limit for the moment. The studio got prepped for tomorrow's photo shoot with Joey instead.
The walk to Suds & Suds found Aaron bartending with basketball on TV driving the conversation. Lyft provided the ride home—a notably better experience and significantly less expensive than Uber. The evening wound down with most of an episode of Formal Logic from the Great Courses, which continues to be both interesting and appropriately arcane for late-night viewing.
The day's through-line connected high abstraction to immediate utility. Book Club wrestled with whether machines could ever truly think, whether consciousness might emerge from sufficient complexity, whether guardrails could contain systems we don't fully understand. Meanwhile, a ten-minute coding exercise produced genuine delight for a colleague facing real constraints. The philosophical questions remain unresolved, but the practical value of putting tools in capable hands revealed itself immediately.
Comments
Post a Comment