Chatbot Interview

H

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
Yesterday we told you about the University of Reading conducting those AI tests (here), well it seems the folks at New Scientist Tech got the chance to “interview” the winning chatbot and the results are rather unimpressive.

New Scientist: How do you feel about winning the AI competition yesterday? Elbot: I have 100,000 separate sensations per second. NS: Wow, that sounds pretty intense? Elbot: This fits the description of magnets. I can hardly tear myself away from them.
Click to expand...
 
unimpressive to say the least.
40+ years of AI/chat bot programming and there still is no detectable conversation going on, just random blithering.
 
Monkey God said:
unimpressive to say the least.
40+ years of AI/chat bot programming and there still is no detectable conversation going on, just random blithering.
Click to expand...

I agree. However, if they would have named the bot McCain or Obama and had it talk about the war or the economy instead of technology, it would be VERY realistic. :p
 
AlienKing said:
I agree. However, if they would have named the bot McCain or Obama and had it talk about the war or the economy instead of technology, it would be VERY realistic. :p
Click to expand...

Yeah it would. Ask a Yes or No question, get a rambling about another unrelated subject.

So I guess our politicians have finally evolved to the intelligence of a crappy AI-simulator program. Wait...maybe they ARE robots?!?!

ZOMG I for one welcome our robot overlords clothed in human bodies pretending to be politicians.
 
Wow. My home town's university is on [H]. Yeah, after speaking to some of the students, the stuff they do there isn't that impressive tbh. Okay, so that's all second hand knowledge, but still.
 
That was very unimpressive. Even though computers get more powerful all the time, with more and more cores, more advanced architectures and higher clockspeeds, AI "bots" don't get any more intelligent. This bot doesn't seem to be much better than some "chatbots" for DOS that I tried out like 10 years ago.
 
So what this particular test proved is that several of the judges are fooled by complete subject changes instead of real answers to questions.

A chat-bot is at it's most obvious when it takes a single keyword out of the question or statement and comments only on that.

Maybe the human counterparts also tried to sound like a chat-bot?
I'd very much like to see the side-by-side conversations that fooled the judges.
 
Actually this is pretty impressive from the coders P.O.V.

He seems to have gone out to win (which he did) AND he did that by making it really obvious it was a robot :D
the judges were convinced a robot wouldn't do that and it must be a human pretending to throw them off

Very clever, very clever indeed (from a conversation P.O.V. yer complete fail, but from everything else, fantastic!)
 
How the hell was anyone fooled into thinking this ambiguous garbage came from a human? And THREE of these "judges" bought it! Unbelievable. I think there was maybe one direct answer in the entire "conversation". Most of the responses had nothing whatsoever to do with the question that was asked and instead seemed like an incredibly random response. What is this, the tech version of the Magic 8 Ball? :rolleyes:

Q: "How do you feel today?"

A: "Ask again later."

I don't see this thing developing into Hal any time soon.
 
You must log in or register to reply here.
Back
Top