Difference between revisions of "Common Sense"

From MgmtWiki
Jump to: navigation, search
(Solutions)
 
(9 intermediate revisions by the same user not shown)
Line 4: Line 4:
 
==Also Known As==
 
==Also Known As==
 
* Tacit Knowledge
 
* Tacit Knowledge
* Implicite Knowledge
+
* Implicit Knowledge
 +
 
 +
==Problems==
 +
When training an [[Artificial Intelligence]] we try to teach it what we know. The problem with that is that we function in the world doing things that we don't know how to do.<ref>Matthew Hutson ''Can Computers Learn Common Sense?'' New Yorker (2022-04-05) https://www.newyorker.com/tech/annals-of-technology/can-computers-learn-common-sense</ref> That stuff is called tacit knowledge which has not been articulated into words, and perhaps cannot be articulated without a great deal of introspection.
 +
 
 +
==Solutions==
 +
 
 +
* "Over the past decade, technology and automotive pundits has predicted the "imminent" arrival of fully autonomous vehicles that can drive on public roads without any active monitoring or input from a living human driver.  Each prediction has fallen flat, largely due to real world safety concerns, particularly related to how self-driving cars perform in adverse conditions or situations."<ref>Keith Kirkpatrick, ''Still Waiting for Self-Driving Cars'' (2022-04) '''CACM 65''' No 4 p 12 ff</ref> The most likely scenario is long-haul trucks once they have been driven to an interstate. Many autonomous vehicle incidents involve "edge cases" -- situations systems have not been specifically trained to handle. Waymo in early 2022 did have self-driving car pilot available in the Eastern part of Maricopa County Arizona, a state that actively seeks such investments.
 +
* SAE levels of driving automation:
 +
# Driver Assistance - fairly common in 2022
 +
# Partial Automation
 +
# Conditional Automation - human over-ride required
 +
# High Automation - Human over-ride optional
 +
# Full Automation
  
==Probems==
 
When training an [[Arificial Intellegence]] we try to teach it what we know. The problem with that is that we function in the world doing things that we don't know how to do.<ref>https://www.newyorker.com/tech/annals-of-technology/can-computers-learn-common-sense</ref>
 
 
==References==
 
==References==
  
[[Category: Artificial Intellence]]
+
[[Category: Artificial Intelligence]]

Latest revision as of 09:32, 7 April 2022

Full Titel or Meme

Common Sense is the stuff that we know without knowing it.

Also Known As

  • Tacit Knowledge
  • Implicit Knowledge

Problems

When training an Artificial Intelligence we try to teach it what we know. The problem with that is that we function in the world doing things that we don't know how to do.[1] That stuff is called tacit knowledge which has not been articulated into words, and perhaps cannot be articulated without a great deal of introspection.

Solutions

  • "Over the past decade, technology and automotive pundits has predicted the "imminent" arrival of fully autonomous vehicles that can drive on public roads without any active monitoring or input from a living human driver. Each prediction has fallen flat, largely due to real world safety concerns, particularly related to how self-driving cars perform in adverse conditions or situations."[2] The most likely scenario is long-haul trucks once they have been driven to an interstate. Many autonomous vehicle incidents involve "edge cases" -- situations systems have not been specifically trained to handle. Waymo in early 2022 did have self-driving car pilot available in the Eastern part of Maricopa County Arizona, a state that actively seeks such investments.
  • SAE levels of driving automation:
  1. Driver Assistance - fairly common in 2022
  2. Partial Automation
  3. Conditional Automation - human over-ride required
  4. High Automation - Human over-ride optional
  5. Full Automation

References

  1. Matthew Hutson Can Computers Learn Common Sense? New Yorker (2022-04-05) https://www.newyorker.com/tech/annals-of-technology/can-computers-learn-common-sense
  2. Keith Kirkpatrick, Still Waiting for Self-Driving Cars (2022-04) CACM 65 No 4 p 12 ff