Robots can flip burgers, too!

In recent posts, I haphazardly introduced some taxonomical terms when speaking about labor automation. Here are some classification dimensions we may use.

  • low/high intelligence-intensive task: A task may demand a low degree of intelligent thinking to solve a problem, or it may use a high degree. A problem that requires one to imagine a new model or method to solve a problem would be highly intelligence-intensive, while a job that requires one to apply a method continuously or repeatedly would require low intelligence-intensive processing.
  • low/high data-intensive task: Does the task require “big data” to perform adequately, or is a small chunk of data sufficient?
  • low/high knowledge-intensive task: Writing a philosophy essay requires a lot of knowledge, while cleaning the dishes requires little. Solving an engineering problem requires intimacy with a lot of theoretical and technical knowledge. Replacing an expert psycho-therapist might require one to know a great deal about the field of psychology. Knowledge-intensive tasks tend to be more expensive to perfect for an AI system, since more training/testing is required to measure the acquisition of knowledge.
  • low/high creativity: How creative does an agent have to be to adequately address a task? You would have to be very creative when designing an original art piece, while no degree of creativity is required when flipping burgers.
  • low/high skill: How much real-world skill in a particular area required to solve the problem? Piloting a spacecraft might require a lot of skill, while it takes little skill to carry a plate.
  • low/high interaction: How much does the task require the agent to interact with other people?
  • low/high language-intensive: How much linguistic competence does the agent need to perform to successfully complete a task? A medium amount of linguistic competency may be enough for emulating a secretary task, but a very high degree of linguistic competence would be expected from an artificial legal expert.
  • low/high repetitiveness: How repetitive is the task? Is it like milling the same block of metal many times? Is a task being repeated often or rarely?

Using such dimensions, we can make a quite informative taxonomy of existing tasks/jobs for automation targets. I used these terms in the course of explaining why I am trying to automate scientific tasks.


Labor Automation Concepts


Eray Özkural has obtained his PhD in computer engineering from Bilkent University, Ankara. He has a deep and long-running interest in human-level AI. His name appears in the acknowledgements of Marvin Minsky's The Emotion Machine. He has collaborated briefly with the founder of algorithmic information theory Ray Solomonoff, and in response to a challenge he posed, invented Heuristic Algorithmic Memory, which is a long-term memory design for general-purpose machine learning. Some other researchers have been inspired by HAM and call the approach "Bayesian Program Learning". He has designed a next-generation general-purpose machine learning architecture. He is the recipient of 2015 Kurzweil Best AGI Idea Award for his theoretical contributions to universal induction. He has previously invented an FPGA virtualization scheme for Global Supercomputing, Inc. which was internationally patented. He has also proposed a cryptocurrency called Cypher, and an energy based currency which can drive green energy proliferation. You may find his blog at and some of his free software projects at

Leave a Reply

Your email address will not be published. Required fields are marked *

Translate »