Supposedly it only needed one more human to be fooled to reach that magic 30% margin for the Turing test. Although the BBC reporter clearly seems to think otherwise (Nice convo listed in the news article)
If you repeat the words Yes, Great and Super.....it goes in a loop....besides if you write in another language or nonsense it has the same reply. No even close to the old Elisa (I think) program.
I believe the main problem is that the program doesn't wreally has anything to talk abouth it's not like he likes a particular genre of music or movies. It would help if he tried to manauvre the quistion with a particular point like to stay a way from certain topics and to promote others
I have the Eliza program somewhere. It uses a random operator to switch topics if someone keeps using the same answers. I think elbot does not have a large data set to atleast fool the client...too bad...
If you repeat the words Yes, Great and Super.....it goes in a loop....besides if you write in another language or nonsense it has the same reply. No even close to the old Elisa (I think) program.
Would you speak to an English-only speaking person in a different language?
Quite right, that Elbot is not close to Eliza. Elbot has far more responses than Weizenbaum's 1960s key-word spotting system, which possessed 200 input-output responses.
Chatting at leisure, one-to-one with any artificial conversational system, is not the same as interrogating two hidden entities simultaneously for five minutes, in a competition setting, as was the case in the Reading University hosted practical Turing tests, in the 18th Loebner Prize.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.