2
$\begingroup$

I wished to compile a (somewhat) comprehensive list of companies and organizations that are developing "Artificial General Intelligence (AGI)" or "strong AI" systems, their products and purported techniques and technologies they are using to create such systems.

The definition of AGI that I'm referring to:

Artificial general intelligence (AGI) is the intelligence of a machine that can understand or learn any intellectual task that a human being can

I wish to limit the discussion to systems that have actually been created and tested (to whatever degrees of success) instead of those that simply advertise they are going to 'do' something about it or have patented theoretical concepts.

First example:

  • AlphaGo
  • Company: DeepMind (acquired by Google)
  • Function: Go-playing bot
  • Concepts: Reinforcement learning to improve already trained networks with new data or Unsupervised learning, e.g. by Generative adversarial network to get improved networks by competition.

Second example:

  • AlphaStar
  • Company: DeepMind (acquired by Google)
  • Function: ‘StarCraft’ playing bot
  • Concepts: not published
$\endgroup$
  • 6
    $\begingroup$ Neither of your examples are AGIs by the definition you post, and I doubt Deep Mind or Google would claim these are AGI candidates. $\endgroup$ – Neil Slater Dec 28 '19 at 11:24
  • 1
    $\begingroup$ Not sure you understand what most people mean when they refer to AGI - basically able to generalise to any situation. AlphaGO and AlphaStar are good but they can not do text inference for example $\endgroup$ – benbyford Dec 28 '19 at 13:53
3
$\begingroup$

As far as I know, no true artificial general intelligent system (AGI) has been implemented or is practically useful. Yes, there is Sophia and similar robots that may look like an AGI, but they aren't really AGI systems, as they lack several capabilities that we humans have and they can't really adapt to new circumstances. AlphaGo and AlphaStar are narrow AI systems. However, there are theoretical frameworks for AGI, such as AIXI, which, in any case, have several flaws (such as incomputability, in the case of AIXI).

The Scholarpedia's article Artificial General Intelligence, curated by Ben Goertzel, one of the leading researchers of the AGI field, provides a good overview of the AGI field, including but not restricted to definitions of AGI and approaches to the development of AGI systems, such as

  • universal (AIXI was created based on this approach),
  • symbolic (Soar was created based on this approach), which is based on the physical symbol system hypothesis,
  • emergentist, which is based on the idea that general intelligence is expected to emerge from sub-symbolic dynamics (where a sub-symbolic system refers e.g. to an artificial neural network),

  • hybrid (e.g. CLARION), which is a combination of the universal, symbolic or sub-symbolic approaches.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.