On Wed, Feb 26, 2020, 11:14 AM stefan.reich.maker.of.eye via AGI <
[email protected]> wrote:

>
> Hey Matt, I already did that example in early 2019.
> https://slides.com/stefanreich/how-about-thinking-machines/#/
>

Yep. And the pizza example was mine, not Doug Lenat. But he has another
example you can test. Who does "they" refer to?

The police arrested the demonstrators because they feared violence.

The police arrested the demonstrators because they advocated violence.

And why don't we use rule based language models or common sense knowledge
bases like Cyc? I guess because you need 2 more rules to handle this case.

Words have a Zipf distribution. In any text corpus there is a 50% that you
are seeing it for the first time, so you need a grammar rule to handle it.
Over 20 years you are exposed to 1 GB of text, which is 200M words. That is
100M rules. Lenat gave up after 1M rules. The problem is he had no way to
estimate how many rules you need. But now you do.


------------------------------------------
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/T21bdc2c440c86db7-M9b081fc6c1cfe756f42250a6
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Reply via email to