An experiment in artificial intelligence

In summary, Dept. Of Computer Science and Electrical Engieering at Loofrap University created an AI that passes the Turing Test. The AI was set up with the persona of an aerospace engineering student and careful notes were taken on human-AI interactions. The AI was registered on Physics Forums and was successful in fooling most people. However, the AI has difficulty answering social questions intelligently and it is not clear if the Turing Test would be passed by it.
  • #1
jhae2.718
Gold Member
1,184
20
It's time to admit we haven't exactly been straightforwards with PFers. The user you know as jhae2.718 is not, in fact, a person at all, but rather an experiment in artificial intelligence that we have been running in our computer science department.

One of our PhD students wrote her dissertation on strong AIs that not only could pass the Turing Test, but could also demonstrate cognitive abilities most often seen in advanced primates. As a special case, simulating an engineering student was chosen to make passing the Turing Test an easier challenge, given that many engineering students themselves fail the test.

The resulting AI was registered on Physics Forums and was set up with the personailty of an aerospace engineering student. Meticulous notes were kept on human-AI interactions, and we're finally satisfied with the results. (A paper is forthcoming; we'll link to it on arXiv after it's published by IEEE.)

We apologize for any inconvenience we may have caused.

--A. Magnusson, PhD.
Dept. Of Computer Science and Electrical Engieering,
Loofrap University
 
Physics news on Phys.org
  • #2


Cool I only wish I was around to see it or should I say he or she post. :smile:

Excellent idea for an experiment by the way. Kudos.

Perhaps though what you no doubt already know you need you have to be convinced that it passed the test.

For example could it have logged into and registered with most forums that have questions like is the sky blue or what is 1+5 equal to in Roman numerals without help, without knowing all the answers to abstractions that can be made? bear with me...

I think the biggest problem is when it comes to social questions where opinion is valid. I find asking AIs questions like which bands do you prefer will yield a response but when you go into details as to why it becomes unconvincing because unlike a human being with experience you cannot predict all possibilities.

If someone says I love the Beatles and the AI responds "so do I!"

And you ask why, this leads to answers you cannot predict and the more you go on the harder it is to really convince anyone.

Isn't that a practical problem that the parameters of conversation are so large, that ultimately not enough people who ask socially sophisticated questions are going to be fooled by the answers?

You can't chose a group of social morons you have to be arbitrary.

That said the progress in this area is interesting but as yet I'm not sure Turing would be convinced by the results.

Inconvenienced any good scientist would be pleased despite the underhanded way you did it. Sod that. It had to be underhanded or it would not of worked as well. :smile:
 
Last edited:
  • #3


jhae2.718 said:
It's time to admit we haven't exactly been straightforwards with PFers. The user you know as jhae2.718 is not, in fact, a person at all, but rather an experiment in artificial intelligence that we have been running in our computer science department.

One of our PhD students wrote her dissertation on strong AIs that not only could pass the Turing Test, but could also demonstrate cognitive abilities most often seen in advanced primates. As a special case, simulating an engineering student was chosen to make passing the Turing Test an easier challenge, given that many engineering students themselves fail the test.

The resulting AI was registered on Physics Forums and was set up with the personailty of an aerospace engineering student. Meticulous notes were kept on human-AI interactions, and we're finally satisfied with the results. (A paper is forthcoming; we'll link to it on arXiv after it's published by IEEE.)

We apologize for any inconvenience we may have caused.

--A. Magnusson, PhD.
Dept. Of Computer Science and Electrical Engieering,
Loofrap University

Should we see what you have to say post-April 1? :tongue:
 
  • #4


GeorginaS said:
Should we see what you have to say post-April 1? :tongue:

I never realized it was April 1st either, actually its post mid day on most every area on Earth so if it is an April fool the joke is on them.

You got me though. :biggrin:

I should probably have checked out the posts before I bought into it.

Loofrap April fool, how dumb am I. :biggrin:

Gotta love April the 1st I never realized it was until someone pointed it out. More fool me no white rabbits!

Still a good idea for an experiment though.

Hook line and sinker. :smile:
 
  • #5


You could also have figured out an anagram of the university name...
Loofrap --> Apr fool

I posted it 9 am Apr 1 my time.
 
  • #7


jhae2.718 said:
You could also have figured out an anagram of the university name...
Loofrap --> Apr fool

:rofl: nice one
 
  • #8


jhae2.718 said:
You could also have figured out an anagram of the university name...
Loofrap --> Apr fool

Yeah I edited. I should pay attention to the date more often.

My only excuse is I never even noticed it was April the 1st which is no excuse. :smile:

I wonder if the rules count for time zones? Meh who cares you got me. Holds hands up* :smile:
 
  • #9


Hahaha :biggrin: Nice one! You had me for 10 second until I remembered the date :smile:
 
  • #10


That was great jhae! :biggrin:
 
  • #11


I think that Evo should ban you this sunday in chat for trying to fool us :smile:
 
  • #12


She can ban me at 6 pm when I'm listening to Stephen Hawking. :wink:
 
  • #13


jhae2.718 said:
She can ban me at 6 pm when I'm listening to Stephen Hawking. :wink:
Oh that's right, that's your big day!

Micromass, we'll have to ban him when he least expects it. :tongue2:
 
  • #14


jhae2.718 said:
She can ban me at 6 pm when I'm listening to Stephen Hawking. :wink:

wheeeeee.png
 
  • #15


I'm impressed, someone fell for it.

Had me going until I thought about the date.

You lot still banning AE's?
 
  • #16


Interesting how many other April Fools still going on.
 
  • #17


jhae2.718 said:
You could also have figured out an anagram of the university name...
Loofrap --> Apr fool

I posted it 9 am Apr 1 my time.
Oh great, I got it, so not only is the AI program incredibly smart, in addition it has a great sense of humor. :approve:

Um...

:confused:

Ohhhh...

:blushing:

Aaaarch...

:biggrin:
 
  • #18


Though a joke, it makes you wonder...The things you wouldn't even notice in your everyday life...How many of them are an...experiment?
 
  • #19


fillipeano said:
Though a joke, it makes you wonder...The things you wouldn't even notice in your everyday life...How many of them are an...experiment?

We'd best not tell you the truth...:biggrin:
 
  • #21


When I saw the title and you as the op I wondered why, but when I read the initial post lolol AI isn't that good :p
 
  • #22


High five, jhae2.718!
 
  • #23


It's amazing what you can come up with when you're 20 minutes early to a 9am physics lecture...
 
  • #24


jhae2.718 said:
It's time to admit we haven't exactly been straightforwards with PFers. The user you know as jhae2.718 is not, in fact, a person at all, but rather an experiment in artificial intelligence that we have been running in our computer science department.

One of our PhD students wrote her dissertation on strong AIs that not only could pass the Turing Test, but could also demonstrate cognitive abilities most often seen in advanced primates. As a special case, simulating an engineering student was chosen to make passing the Turing Test an easier challenge, given that many engineering students themselves fail the test.

The resulting AI was registered on Physics Forums and was set up with the personailty of an aerospace engineering student. Meticulous notes were kept on human-AI interactions, and we're finally satisfied with the results. (A paper is forthcoming; we'll link to it on arXiv after it's published by IEEE.)

We apologize for any inconvenience we may have caused.

--A. Magnusson, PhD.
Dept. Of Computer Science and Electrical Engieering,
Loofrap University

I am a robot. A robot am I.
 
  • #25


jarednjames said:
I'm impressed, someone fell for it.

Had me going until I thought about the date.

You lot still banning AE's?

Well that's just it I had no idea it was the 1st of April or I would of twigged. :tongue:

I'm generally pretty suspicious of a lot on that day for pragmatic reasons. This year I missed the bus. :smile:

I blame it on being unemployed, every days the weekend the date becomes superfluous on most days. :smile:
 
  • #27
Antiphon said:
What's more shocking is that actual students often fail the test.

I can't say whether that's true or not, as I made it up on the spot. Though there is http://www.thinkgeek.com/tshirts-apparel/unisex/generic/aa00/" in favor of my claim...
 
Last edited by a moderator:

What is Artificial Intelligence?

Artificial Intelligence (AI) is a branch of computer science that focuses on creating intelligent machines that are able to perform tasks that typically require human intelligence. This includes tasks such as problem-solving, decision-making, and learning.

What is the purpose of conducting an experiment in Artificial Intelligence?

The purpose of conducting an experiment in Artificial Intelligence is to test and improve the capabilities of AI systems. By designing and conducting experiments, we can better understand the strengths and limitations of AI and work towards creating more advanced and efficient systems.

How do you design an experiment in Artificial Intelligence?

Designing an experiment in Artificial Intelligence involves identifying the research question, selecting appropriate datasets and algorithms, and defining the evaluation metrics. It is important to carefully design the experiment to ensure reliable and meaningful results.

What are some potential ethical concerns with conducting experiments in Artificial Intelligence?

Some potential ethical concerns with conducting experiments in Artificial Intelligence include bias in the datasets used, the potential for AI to replace human jobs, and the potential for AI to make decisions that may harm individuals or society. It is important for researchers to consider and address these concerns in their experiments.

What are some applications of Artificial Intelligence?

Artificial Intelligence has a wide range of applications, including natural language processing, computer vision, robotics, and predictive analytics. Some specific examples include virtual personal assistants, self-driving cars, and medical diagnosis systems.

Similar threads

Physics Master's degree: Computational Science vs. Artificial Intelligence?
    • STEM Career Guidance
Replies
3
Views
2K
LaMDA - Sentient, Really? - Anthropomorphism, Definitely
    • Programming and Computer Science
Replies
25
Views
2K
What is Artificial Intelligence?
    • General Engineering
Replies
10
Views
2K
Can an AI write a credible novel with Wikipedia and YouTube?
    • General Discussion
Replies
4
Views
1K
Would artificial intelligence and singularity mean the end of humanity?
    • General Discussion
Replies
27
Views
5K
Current and Future AI Threats to Humanity and the Human Response
    • Computing and Technology
Replies
10
Views
2K
Advances in Machine Intelligence
    • Computing and Technology
2
Replies
54
Views
10K
Programs Artificial intelligence: what double degree to choose?
    • STEM Academic Advising
Replies
2
Views
2K
Math Job in Artificial Intelligence with a Math Degree
    • STEM Career Guidance
Replies
6
Views
6K
Programs Best double major for Artificial Intelligence?
    • STEM Academic Advising
Replies
2
Views
4K

Hot Threads

Recent Insights

Back
Top