The curious AGI-esque results in thinking and in conversation are very new and were achieved not yet in the JavaScript FirstWorkingAGI.html but rather _as_described_in_the_AGI_post in the ghost.292.pl AI in Perl and then more elaborately _as_described_in_the_AGI_post in the ghost294.pl AI in Perl. Achievements in one of the three AI Minds eventually get ported into the other two AI Minds, but the very first announcements of a result usually refere to only one of the AI Minds -- the one in which the result was first achieved.
So if any any AGI-list memeber downloads http://ai.neocities.org/perlmind.txt and uses Strawberry Perl Five or some similar Perl to run the Perlmind code, the claimed results should appear as posted to the AGI mail-list. However, on an important other matter called "The 'Pizza with...' demo" as described by Stefan Reich in the http://agi.topicbox.com/groups/agi/T9ccd0aac7d42f57b post of 2019-02-06 to this AGI mail-list and as shown further at http://tinybrain.de/1021251 by Stefan Reich, Mentifex here has come up with a different approach but has not yet coded the idea now on-line in the "Future Development" section of the http://ai.neocities.org/EnPrep.html#future webpage on logical inference: 7. Future Development The AI Minds with the concept-based, first working artificial intelligence will evolve beyond simple subject-verb-object (SVO) thought-structures into mentalities which may include a prepositional phrase in association with any noun or verb or adjective or adverb. The seq flag may be used to attach a prepositional phrase to a previous word. In the future, an AI Mind may be able to disambiguate sentences with prepositions in a way which reasons from the following inputs. Human: john eats with a fork Ghost: Human: john is with bob Ghost: Human: john eats pizza with a fork Ghost: Human: john eats pizza with bob Ghost: Human: john eats pizza with anchovies Ghost: First the SpreadAct module must evolve sufficiently to spread action from one concept in a statement to all associated concepts -- subject, verb, object or preposition. At that point, the inputs presented above may yield an associative knowledge base that links each instance of the preposition "with" back to its truly corresponding concept, from "with Bob" back to "John"; from "with a fork" back to "eats"; and from "with anchovies" back to "pizza". The AI Mind may then be able to respond properly to queries in the following manner. Human: Who eats pizza? Ghost: JOHN EATS PIZZA AND BOB EATS PIZZA Human: How does John eat pizza? Ghost: JOHN EATS PIZZA WITH A FORK Human: What does John eat? Ghost: JOHN EATS PIZZA WITH ANCHOVIES The AI will think and output each response because the concepts in the response have been activated by the SpreadAct module, and because the preliminary ideas in the knowledge base have established retroactive associations from the concept "with" to the proper antecedent. On Sat, Feb 9, 2019 at 4:37 AM Stefan Reich via AGI <[email protected]> wrote: > I mean, we're all tired of this now, but, one last time... I'm taking you > at your exact words. > > I entered *exactly your example* on *your page* ( > https://ai.neocities.org/FirstWorkingAGI.html). First I typed "John has a > house", then "What does John have". The system said nothing, then just > this: "I HELP KIDS" > > So it does not even do what you claim it does. > > Do you accept this as a bug report? Or do you not even care if anything > you say is true? Did I enter the lines in the wrong place? (Then that's > also a valid bug report.) > > Come on dude. I'm sure you're capable of sanity. You have working > vocabulary, just the semantics have a major bug (diagnosing you like one > would an AI). > > On Sat, 9 Feb 2019 at 13:26, A.T. Murray <[email protected]> wrote: > >> Recently in alt.folklore.computers on Usenet our name was taken in vain >> and counter-compensated as follows: >> >> >> I wanna see Gareth write Alexa and have it >> >> work as well as Amazon's version. >> > Hasn't Mentifex already done that ? >> >> No. Mentifex has not written software that looks up and >> tells you the score of the football game in progress, or ships pizza to >> your doorstep. >> >> Mentifex has written three AI Mind programs in Perl and Forth and >> JavaScript >> which converse with you and learn new words from you in English or >> Russian. >> >> http://ai.neocities.org/EnVerbPhrase.html -- shows the latest AI >> progress: >> >> 3.D. Leaving out the repetitious verb from any additional phrase being >> joined together with a first phrase for a resulting compound sentence that >> may be replaced with a single one-phrase sentence containing one or more >> conjunctions. >> >> The user may tell the AI Mind two things: >> >> John has a house >> John has a car >> >> The user may then ask the AI: >> >> What does John have >> >> The ghost292.pl AI was able to respond with the following output: >> >> THE JOHN HAS THE HOUSE AND THE JOHN HAS THE CAR I UNDERSTAND YOU >> >> Suppose that we want the AI Mind to join two sentences together: >> >> JOHN HAS A HOUSE >> and >> JOHN HAS A CAR >> >> >> By not repeating the subject, we get: >> JOHN HAS A HOUSE >> and >> JOHN HAS A CAR >> >> >> By not repeating the subject and the verb, we get: >> JOHN HAS A HOUSE >> and >> JOHN HAS A CAR >> >> The AI coding task for ghost294.pl in Perl was how to get the >> sentence-generation modules to skip repeating the subject and skip >> repeating the verb. Note that even a human being will sometimes dawdle or >> linger while uttering the compound-setences, as in: >> "John has a house...and a car...and a lawnmower...and a canoe, etc." >> >> It was necessary to use one of the "chain" or "conCATenation" variables: >> >> >> our $catiobj = 0; # 2019-01-28: concat-indirect-object for >> ConJoin() "AND". >> our $catdobj = 0; # 2019-01-28: concat-direct-object for ConJoin() >> "AND". >> our $catsubj = 0; # 2019-01-28: concat-subject for ConJoin() "AND". >> our $catverb = 0; # 2019-01-28: concat-verb for ConJoin() "AND". >> >> We load the $catsubj <http://ai.neocities.org/var.html#catsubj> variable >> inside the query-response code in the SpreadAct >> <http://ai.neocities.org/SpreadAct.html> module. After the first phrase >> in response to a query, such as "JOHN HAS A HOUSE", the AI may check each >> activated idea for starting with $*catsubj* >> <http://ai.neocities.org/var.html#catsubj> and for having $*catverb* >> <http://ai.neocities.org/var.html#catverb> in the second place. In that >> way, the various direct-objects may be isolated, introduced with "AND" from >> ConJoin <http://ai.neocities.org/ConJoin.html>, and sent into the Speech >> <http://ai.neocities.org/Speech.html> module for the output of a >> meandering, run-on response such as "JOHN HAS A HOUSE AND A CAR AND A BOAT >> AND A SLED AND A MOTORCYCLE". Such an AI output is like human speech, where >> the person states the basic premise before even thinking of all the nouns >> about to be concatenated with "AND" into the meandering response. >> >> Once the chain-variables are loaded with the concept-numbers of the items >> to be omitted for the syncopation or shortening of a compound-sentence, >> each item is omitted simply by letting Indicative >> <http://ai.neocities.org/Indicative.html> call EnNounPhrase >> <http://ai.neocities.org/EnNounPhrase.html> or EnVerbPhrase >> <http://ai.neocities.org/EnVerbPhrase.html> but preemptively using >> conditionals to make each module "return" to Indicative >> <http://ai.neocities.org/Indicative.html> before sending a noun or a >> verb into the Speech <http://ai.neocities.org/Speech.html> module. >> Special code in the EnVerbPhrase >> <http://ai.neocities.org/EnVerbPhrase.html> module must still call >> EnNounPhrase <http://ai.neocities.org/EnNounPhrase.html> to provide the >> direct object of the omitted verb. >> >> >> >> > > -- > Stefan Reich > BotCompany.de // Java-based operating systems > *Artificial General Intelligence List <https://agi.topicbox.com/latest>* > / AGI / see discussions <https://agi.topicbox.com/groups/agi> + > participants <https://agi.topicbox.com/groups/agi/members> + delivery > options <https://agi.topicbox.com/groups/agi/subscription> Permalink > <https://agi.topicbox.com/groups/agi/T455bc32c8a6722e4-Mf2bf6175543b01f6b9acc3de> > ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T455bc32c8a6722e4-M60ab31effb52fdd9fad15d6a Delivery options: https://agi.topicbox.com/groups/agi/subscription
