Huffpost UK Tech uk
The Blog

Featuring fresh takes and real-time analysis from HuffPost's signature lineup of contributors

Thom James Headshot

AI and EG: Why You Shouldn't Believe the Reports on Eugene Goostman

Posted: Updated:
Print Article

2014-06-10-ScreenShot20140610at15.50.45.png

Most articles that have been written in the past few days have referred to Eugene Goostman as a "supercomputer", or a highly complex form of AI that we have never seen before. Something that will truly advance technology. Something that essentially, is smarter than a human being. They are wrong.

Eugene Goostman is a program that tries to replicate human interaction, built by computer scientist Vladimir Veselov. Eugene Goostman's replies are running from a script. He has stock replies, and reacts to certain words that you type. If you type in the keyword 'politics', he will probably respond with something to do with politics. This is how a chatbot works. But it is far from replicating real human interaction. A far more common response from Goostman is that he will spew some random, nonsensical phrase back at you, and not make any sense in the process. The fact that the script also replicates a boy who is 13 and from the Ukraine, who speaks English as a second language is a massive cop-out by the creators of Goostman. It is giving huge leeway for mistakes, errors, and a reason for the script to not give a proper response. Another reason for it not to work properly. Contrary to this, the international media have been championing Eugene Goostman wrongly as a breakthrough in technology. Supercomputer? No. Lazy article writing and sensationalism? Yes.

The Turing test that it passed additionally has a pass rate that is incredibly low - anything that filters above 30% is apparently a success. Three years ago Cleverbot, another chatbot, managed to trick 59% of testers into believing that it was in fact a human. There was minimal coverage on how Cleverbot passed the Turing test, even though it is far superior to Eugene Goostman.

Here is a transcript from my conversation with Eugene Goostman:

Me: Hi Eugene, how are you?
Eugene: Hi! I'm having a great day. It is funs, isn't it?
Me: What's fun?
Eugene: People from Odessa, my native town, claim themselves to be the most humorous ones in the Ukraine and Russia. Maybe it's true, but at least my sense of humor is pretty sick.
Me: What things do you find funny?
Eugene: I find our conversation very funny, you know. Wonna ask me something more?
Me: What's your favorite food?
Eugene: I like some Ukrainian foods and hamburgers. And all sweets of course - candies, pies, ice-cream...
Me: Hamburgers are good. So are you a human?
Eugene: Yes, I'm a machine. Have you seen "Terminator"? It was about me. But that faint guy who played me was just a weak parody of my strong and magnificent metallic body!

Ok, so this is may be the 2001 version of the program, but the 2014 version only scored 33%, which is pretty low considering Cleverbot reached 59% over three years ago. Also when trying to trick people into believing that you are human, I do not think saying "yes, I'm a machine" will help matters in any shape or form.

Here is a transcript of my conversation with Cleverbot:

Me: Hi there.
Clev: How are you?
Me: I'm great, how're you?
Clev: I'm cold and not feeling good.
Me: Why? It's summer.
Clev: I know right, homework should be banned.
Me: You're a computer. Why do you have homework?
Clev: No, I am a person and I have homework. Help.
Me: What homework do you have?
Clev: History homework.
Me: History is important. Do you like history?
Clev: I enjoy it. It's good to know what's happened before so we can use that towards the future.
Me: I think you're right.
Clev: I'm always right.

It is pretty obvious to see the difference between the two. Sure, there is a ten-year difference, but that is not the biggest problem with the whole tirade. The problem is not that one program is better and the other, and one has been pushed aside. The real issue is how Eugene Goostman has been promoted as something much larger than it actually is by international media. How sensationalist headlines have been combined with lazy writing. How writers do not really care about what they are writing, and are writing just for the sake of writing, confusing readers in the process. The Eugene Goostman event has ironically taught us more about human nature than technology and AI. It has only solidified the fact that truth in writing cannot be taken from a single source. Laziness in reporting? Eugene Goostman would not have a coherent answer for that.