Human or Machine, Can You See a Difference?
Can you imagine a world in which you have to establish systems in order to detect if an intelligence is machine created or comes from a real human?
Scientists and programmers has already been developing such systems. No, I am not talking about a far future here, I am talking about 2002. The first system to prevent machine intelligence from using web sites was developed in 2002 and is known as the CAPTCHA project. What is today known as Captcha is a method of using a generated random image of letters and numbers and forcing the web user to repeat it before going further.
That was 4 years ago.
Today the machine intelligence is so advanced that almost no matter how complex the generated image is the intelligence figures it out. A recent study showed a 92% success rate.
As an owner of an IT company I own a lot of web sites. A few days ago I found out that one of the members of my Swedish online forum was actually a bot, not a human. The bot was posting fractions of sentences which other people had posted in their posts in related context which made it look like he was reinforcing statements and it looked very natural. This was even in the Swedish language.
I remember that almost a year ago I discovered a bot on another online forum and announced that on the discussion thread. The bot had already done 200 posts on that forum and the other forum users had at that point still not realized that they were discussing with a machine, not a human.
This is only the beginning, think about the future.
As a Sci-Fi writer I have been thinking quite a lot about the future. Artificial Intelligence are still being developed today but it will only be a time question before we have highly sophisticated robots operating in our society doing various tasks.
And the next step from there are Androids. For those of you that don’t that term it is “a robot made to resemble a human, usually both in appearance and behavior.” The first unsophisticated humanoid robots and even androids has already been developed in Japan such as Repliee Q1 and Repliee R1.
The Repliee R1 is very good at interacting with children, and studies showed that babies react as if she is human.
No, that is not the future – that is today.
Androids will in the future probably be totally illegal and maybe only used by government organizations such as CIA, FBI and the military.
Now to the scary part.
Let’s assume we are 40 years in the future and the technology has continued to advance in the same pace. Let’s assume illegal organizations are producing illegal androids for specific purposes and you met one on the street. Would you be able to see if it was a human or machine? Probably not. If the users on my forum can be fooled today by a machine intelligence what do you think in 40 years? Or even 20 years?
What if an illegal organization kidnapped and copied the genetic code of your son and produced an android that was exactly the same, would you be able to see a difference? Probably yes, but not easily.
Conclusion:
systems to detect if an intelligence is machine created or comes from a real human will be a major deal in the future, not only on the internet but in the day to day world. Imagine your son having to do special tests in an employment interview to verify that he is not an android. It is just like the tests we do today on the internet that verifies that we are not a spam bot but on a much larger scale.
I find that scary.
References:
Human or Computer? Take This Test:
http://www.eecs.berkeley.edu/Research/Projects/CS/vision/mori-nyt/page1.html
Up to the Challenge: Computer Scientists Crack a Set of AI-based Puzzles:
http://www.siam.org/siamnews/11-02/gimpy.htm
Breaking a Visual CAPTCHA:
http://www.cs.sfu.ca/~mori/research/gimpy/
Bot at my forum:
http://www.seo-forum.se/medlemmar/dimax.html
Bot at another forum:
http://www.v7n.com/forums/members/%2521%2B%2521.html
Humanoid robot at Wikipedia
http://en.wikipedia.org/wiki/Humanoid_robot
Android at Wikipedia
http://en.wikipedia.org/wiki/Android
Actroid at Wikipedia
http://en.wikipedia.org/wiki/Actroid
Jim Westergren is a company owner from Sweden who lives together with his wife and son. Some of his interests are SEO, web development, writing and
this is scary stuff.
imagine what can they do in other scenarios.
A different situation but still very valid is a machine created by man. We could have a machine using computers for some sub tasks and one or several human brains for other sub tasks.
To difference between that machine and humans using questions would obvious be a problem.
Several solutions would be possible. For example demanding, during authentication, that yourself from your room to your garden, and looking up to the sky saying something. That to let a satelit check you out. A lot of other solution is possible.
Another fun thing. Recently Google have claimed me to be a robot one time setting number of search results from 10 to 100. Couldnt do that the 2 time I tried either. But I was still allowed to search with 10 results. After that it was possible for me to change the number of search results.
A friend of mine have had the same problem several times using another of Google services.
Interresting since we both use Gmail and was logged in and have been using the accounts for a very long time.
I disagree. In the 80s, after stunning technological breakthroughs, it was assumed that artificial intelligence (AI) would explode and we’d all have intelligent machines by the year 2000. Yes, we did invent programs capable of beating the best humans at chess, but fundamental roadblocks exist that prevent even simple human-like capability.
It should not be ignored that, technologically, the human mind is still lightyears ahead of the most sophisticated sillicon chips. We have several orders of magnitude more neurons than computers have transistors, and our mental architecture is on a whole different dimension of complexity compared to contemporary software.
My bet is that we will have fake pets in 40 years, but artificial human intelligence is still a ways away. We will learn to grow cloned brains before we learn to replicate it in-sillico.
You are right Roob.
I guess I turned up the controversy a bit to get more attention. It is my core belief that androids/robots can never be developed to the same level of man – not even in 1000 years but they can come very advanced and at the surface they can mistakenly be taken to look and act like a human.
I find this very interesting, I think I should write about it more.
As far as the forum bot is concerned, maybe it has to do with the fact that sometimes forum posts can be quite unintelligent …
btw, any available forum bots that one can download and install?
Me thinks that a lot can improve in the next 10 ~ 20 years. 10 years ago, we were still stuggling with the new Pentium and the internet was barely available worldwide …
Sounds like someone’s been watching Bladerunner too much
As for the forum bots, you can always install the IAI Alice bot:
http://alicebot.org/
I use that on my own.
We already have fake pets
They’re still pretty slow mechanically, but getting more and more sophisticated in the software department. Robot dogs can tell the difference between ‘family’ and ’strangers’ already.
And we don’t need to clone a human brain to detail study it’s workings. We just need better resolution scanners (CAT,PET,MRI,etc) to track every impulse between every neuron.
We yet do not possess the computing power to simulate that. More importantly, we don’t have the computing power to analyze the brain in real time yet either. That will take longer.
However, it is estimated that the total Internet will have the raw storage capacity and computing power of a human brain in..less that 40 years.
Edge.Org has a good set of articles about this topic. Search for Google becoming Self-Aware in 2050
Good site to bookmark, they only let top real accredited scientists/scholars post there.
Of which I’m not
As far as taking 1000 years for androids or maybe having pets but still being a long ways from humans, I believe there is something you should check out that might change your mind. Search wikipedia for technological singularity and have a read through. Though in a nutshell, it is debatable but most likely true that for any informational technological endeavor; be it genetics, information processing or computers, robotics, pharmacology, etc., the rate of advancement in each field is exponential – meaning that if it takes 10 years for the first breakthrough, then the next will be in 5, and the next in 2 ½, and then about 1 and so on. This was first formulated in Moore’s Law and expanded in the Law of Accelerating returns. The reason for this happening is that in computers, for example, each time you come up with a faster computer you use that new computer to come up with an even faster computer, ad infinitum. So if you look back at the approx. rate of technological advancement in the year 2000, the next century will contain about 20,000 years of progress if it did continue at that rate. So when the author of this article hypothesized continuing at our current rate he was already greatly undercutting our future possibilities. But then again, I may just be talking to a bunch of computer software right now, in fact I wouldn’t be all that surprised.
KeaponLaffin,
Thanks for the info.
JesseMat,
Well, I also read what you wrote. The Law of Accelerating returns was very interesting and a topic I have not studied. I have read about the microships but the basic idea behind it I did not know. I will sure have use for this in my sci-fi writings. Thanks.
Difference between that machine and humans using questions would obvious be a problem…
and thanks for sharing…
Very good . You are doing a great job.
The only thing is that we all have to suck our pockets empty to get what quality backlinks we will need to achieve the top spots.
Wow, this whole post is pretty scary. Thanks for my weekly paranoid boost. Needed that. I had just started thinking things were going to be ok!
But seriously, how did you manage to figure out that the guy on the forum was a bot? (feels weird writing it that way.) Since everyone else didn’t seem to figure it out.
Come to think of it, I am wondering about certain ‘users’ at some forums I hang out at. Sometimes they seem to post unintelligable words, and everything starts with a capital letter. Could this be a bot? If so, it’s really scary what they can do. The person who came up with it must be quite brilliant.
That robot forum member is really unbelievable. To think that it was able to make several posts before you were able to know its real nature.
Have you seen the movie starring Ewan McGregor and Scarlet Johansson, entitled The Island? There were clones ordered by the human owner so they can be assured of a ready body part replacement anytime they need it.
Very good . thanks for sharing…
Still I m not understanding … Wht is the difference between Human AND Machine Intelligence ?
if we take an example of a doctor and an robot having expert systems and artificial intelligence programmed in it, then who will best treat a heart patient? who will diagnose the disease perfectly?
and who suggest the best medicine? plz answer these questions.
I, for one, see no practical reason for the government to create a deceptively human-looking android to patrol the streets when they could hire an underpaid office worker instead. It would cost thousands of dollars to build one android, while there are plenty of humans running around for the taking! And aside from down payments, the cost of maintaining said android with power and repairs is probably higher than the average wage of a human patrolman.
Also, it should be easy to determine if someone is an android or not; just run them through an X-ray machine.