Artificial Technology

"Intelligence is whatever machines haven't done yet."

-- Larry Tesler 

In the 90s I took a course on AI.  It was about expert systems, decision engines, and neural networks.  In the 2000s ML methodologies like PCA and Logistic Regression were widely regarded as AI.  In the 2010s it was CNNs.  Now it's Agentic LLMs. What constitutes "AI," seems to change every few years. 
If Intelligence is what machines haven't done yet and Artificial Intelligence is them crossing that boundary, we can consider much of business Operations the opposite, Artificial Technology. Humans doing what machines can/should.  A human acting as a mechanical component in a process that shouldn't require one. 
Consider this reframing:
  • Data entry — humans as OCR
  • Call center scripts — humans as decision trees
  • Rote code review checklists — humans as linters
  • Approval chains for trivial decisions — humans as if-then statements
  • Manual report generation from dashboards — humans as SQL queries
With the introduction of Agentic technologies and workflows, these aren't just inefficiencies, they're a kind of category error, using the wrong substrate for the computation.
While the AI discourse is obsessed with machines becoming more human, the more interesting problem may be humans who have already become machines, performing work so procedural and rule-bound that it doesn't use what's distinctly human at all.
 Our economy is undergoing a swift upheaval due to these technologies.  To stay competitive, this framing may be useful.  Look at your business, every human's role as "Artificial Technology."  Ask yourself, how could we enhance this with a system?  How could we liberate this human?

Comments

Popular posts from this blog

Rules and Conventions

Engineering Truisms

The Contemporary Yamanba