News: [Artificial Intelligence Group] Re: Why AI Fails

--- In artificialintelligencegroup@yahoogroups.com, fenris_23
Note to fenris_23.
You missed the point of my blog.
It is not an argument against strong AI.
It is an argument that AI R&D has been looking in the wrong direction.
This is consistent with the views of Marvin Minsky.

My view is this:

No general-purpose AI exists.

Why?
Not because general-purpose AI is impossible,
but because general-purpose AI requires Basic R&D,
and almost all AI developers are driven by Technology R&D.

Marvin Minsky suggests in his 2008 pod-cast that less than 10 scientists in the United States are working on general-purpose AI.

Computer science has failed to produce general-purpose AI because nearly all AI developers seek special-purpose application goals and models and not general-purpose AI models.

This is like trying to invent the gun before inventing gun powder.

This is why after 50+ years of AI R&D we have yet to produce a practical general-purpose AI.

This is why AI R&D fails.

<no_reply@...> wrote:
>
> Just my two cents. But this argument against strong AI always seems,
> to me at least, to fall into the logical fallacy of argument from
> ignorance. You are basically stating that since (1) we have been
> unable to develop strong AI and (2) it seems so impossibly difficult,
> then therefore it must be impossible. However, there is no physical
> law or impediment to emergent phenomenon to prohibit such an
> intelligence to exist. The fact that we have not ourselves built
> strong AI does not imply that strong AI is impossible or "fails".
> Indeed the pursuit of strong AI has driven massive advanced in various
> fields from neuroscience, cognitive science, and even geopolitical
> strategy.
>
> It is kind of pointless to throw around opinions regarding the
> ultimate feasibility of strong AI, or the relative merits between weak
> and strong AI. Everything has its place.
>
> But you can't prove something can't exist because you have not ever
> seen it. That is the black swan fallacy.
>
>
> --- In artificialintelligencegroup@yahoogroups.com, "bnewbnewb"
> <bnewbnewb@> wrote:
> >
> > Why AI R&D Fails
> >
> > All software developers are Research and Development (R&D)
> professionals.
> > There are 2 kinds of R&D, Basic R&D and Technology R&D.
> > Technology R&D takes 6 months to 6 years. An example of Technology R&D
> > is inventing the next generation gun.
> > Basic R&D takes 20 to 30 years. An example of Basic R&D is the
> > invention of gunpowder.
> > Nearly all computer scientists fall into the Technology R&D category
> > because nearly all software is based on previous software generations
> > and functionality.
> >
> > Specific-purpose AI falls into the Technology R&D category because it
> > is based on standard functionality such as neural networks, decision
> > trees, case-based reasoning, fuzzy-logic, expert system rules, and
> > other known AI related data structures. An example of
> > specific-purpose AI is offering a menu of cable TV packages that best
> > matches all the TV show categories that the customer checkbox-selected
> > on a webpage.
> >
> > General-purpose AI falls into the Basic R&D category because no
> > computer model has been created that accurately duplicates all the
> > functionality of the human brain. An example of general-purpose AI is
> > natural language input and the computer understands what you mean from
> > the many ways of describing it, and then knows what action to perform
> > and can explain why.
> >
> > General-purpose AI has been fantasized in books and movies such as
> > "Metropolis" (1925), "2001 - A Space Odyssey" (1968), and "Star Wars"
> > (1976), but to date no one has achieved these fictional goals for real.
> >
> > Marvin Minsky, the head of the AI department at MIT, is interested in
> > general-purpose AI and discusses the failure of achieving
> > general-purpose AI in a March-2007 podcast. The link can be found at
> > www.slashdot.org.
> >
> >
> >
> http://www.informationweek.com/news/showArticle.jhtml?articleID=197700609
> >
>