Artificial intelligence

  

Artificial intelligence is an attempt to automate thinking like robots attempt to automate manual labor. — Fletcher

I found the above video on The Feral Irishman blog … and its one phrase, “outsourcing thinking” got me musing again about artificial intelligence (that ubiquitous AI that everyone anticipates). Anyhow, watch this video … maybe a few times like I have … and then come back for some of my own pontificating on the subject.

What you ask are my credentials to write on this subject? Well, in the early 1990s I founded a company, Occam Research,  to develop and market a product, MUSE (Machine-User Symbiotic Environment),  an attempt at a natural language//database/spreadsheet product that tried to ease the burden of data analysis for the unsophisticated computer user … dancing around the edges of AI. 

The company failed … not because the product wasn’t impressive … but because we had a poor business model … and because the product offered more analytic power than most users really wanted. In other words, they were not hot on outsourcing their thinking … even if it’s wrong.

Pilgrim, to me the expectations for AI  are far too optimistic and uncertainm. The human brain can short circuit very long and nebulous inductive decision trees deductively and generally with better results.

This may be because hammer-and-tongs AI programmers run out of patience coding for never-ending conditionals. And deductive logic is difficult to replicate in AI except through impossibly huge data bases of actual human experience which are difficult to collect and organize. And I suspect this fact is motivating Elon Musk’s goal of connecting a computer to the human mind.

As testament to this conclusion reference the continued failures of driverless cars and online searches.

Even IBM’s successful Big Blue computer chess expert derived much of its success from a huge database of past human chess games … which provided the deductive piece of this gaming puzzle.

Bottom line: If AI really ever occurs, this blogger thinks it may be many many years hence.

Afterward: another thought … the natural language component of AI depends a lot on Chomsky-like transformational rules … which, as we have seen of late with our playing around with pronoun agreements … can change in the blink of a woke eye.

STAND UP FOR THE HUMAN MIND!