Skip to main content

Week 4 [06.11-12.11] – Can AI pass a university entrance exam?

Almost every aspect in our lives is processed by computers, and many work places are replaced by machines. In recent years there has increased the  development of AI. Most big tech companies have an Artificial Intelligence project, they are searching for experts, and are ready to pay millions of dollars to help it done. Heads of World Powers are interested in the development of AS as it will ensure victory in a modern battlefield that is Computer Science.

Artificial Intelligence is on a fledgling level, but already achieves better results in university entrance exams than half of the students. This topic is worthy of note and I encourage you to watch this TED talk.



Questions:
1. What is the reason that students get worse results than the machine? Do you think we should blame the education system?
2. Do you think we should be afraid of AI? Will it try to exterminate us like in the Terminator movies? Why? Why not?

Sources:

Comments

Anonymous said…
People get worse results than machines, because there is no way for machine to forget something. Students are required to remember a lot of data, theories and formulas while machine just have some code and just give it in appropriate moment. Machines don't get stressed or distracted - machines just work.
Of course we should be afraid of AI. When it comes to work IT and some creative areas are save for now, but I don't think it will last forever.
Unknown said…
I do think that education system should be changed particularly in my country but I can’t say that we should blame it for computers being better in test than students. The reason me by that people are intelligent, can think in many ways, have emotions like stress etc. and also have their own and only one mind. Computers and machines often combine ideas from many inventors in great way so we can say it is excellent teamwork. We know that teamwork often gives better results. Nowadays AI is dynamically growing and as for now we can be calm but it is about time to get prepared for AI able to replace people.
Unknown said…
Today, artificial intelligence does not understand the meaning of words and sentences, as it was mentioned in a TED talk. That's why it is dangerous to use it in a work that requires making serious decisions(judge, architect, etc.).


But what that's what I'm thinking about AI at all:
Human brain has been evolved over hundreds of thousands of years by many random(!) genetic mutations that helped us to survive. That's why our brain is not perfect: we can easily remember images, but it's hard for us to remember texts or numbers, because it wasn't needed for survival 5000 years ago(just a moment in terms of evolution).
But artificial intelligence is a whole other thing. It has a perfect memory; it can process a huge amount of texts in a seconds.
Evolution as a mechanism cannot improve something on purpose - it is not "interested" in making people smarter. But now it's not evolution who rules the world now - it's humanity. AI was created on purpose, and I believe that if evolution could make us, we can try too.
Sure machines will prove themselves better in tasks that require memorizing, but if students do worse in an essay, it should ring a bell. When the time comes where people would not be needed in IT, then we will have fulfillment of Terminator movie scenario in real life.
Alicja said…
She makes an excellent point about the need to modify our education system to focus on intellectual qualities that can not be substituted by AI methods. Students should be taught innovative and creative approaches to problem solving, not to be trained as "bank clerks".

There are other striking examples of AI winning with our current systems and even human reviewers. For example: MIT’s Computer Science and Artificial Intelligence Lab created a program that generated a gibberish scientific publications, which they submitted to peer-view journals. Tens of these articles where accepted and published by Springer and IEEE.

We should fear AI if we have an employment that involves structured problem solving tasks or routine physical work. AI is not a magic box that can do anything. It is a set of mathematical theories that can solve selected problems. AI methods used broadly today (Artificial Neural Networks, SVM, Markov Decision Process, naive Bayes classifier) were developed in 40s, 50s and 60s and have not changed since then. The math behind them is still the same and has the same limitations.
Unknown said…
This speech brought an attention to me, personally in my studies. Throughout my diploma learning I've made used of countless memory techniques to ace my exams, combined with true understanding I can bring out the answer that no AI can, however memory was one of the lacking thing, memorisation. If AI could produce meaning and understanding, the end of humanity will be close.
Filip Sawicki said…
Students get worse results than machines when dealing with questions based mainly on logic and memory, however they are outperforming most advanced A.I. solving abstract and complex subjects. The problem lies in the fundamental difference of natural learning process of us and machines. We do not fully understand capabilities and functions of our brain, there are many mysteries to be revealed in the future. A.I tries to mimic human brain, create artificial neurons -> perceptrons and then apply complex mathematical functions to train them, which is very basic generalization of our vast and complex neurological structure and learning process. Also, I don’t think that we should be scared of A.I. rather try to live with it in symbiosis as we can both benefit much more than exist alone.
Jakub Lisicki said…
Some of the students get worse results than an AI, because they don't have access to as many resources as the mechine. They simply can't memorize anything. Some other problem is that students who got used to really hard and abstract tasks, can really solve the easiest questions which are based on pure reading anymore. I don't know if we should blame the current education system for what it currently is, but I also think that we should blame it for it's reluctancy to evolve. We can't know if it's the best system that we can achieve if we don't try and research other ways of learning.
As far as I know, AI is not and will not be a threat to human race anytime soon, unless it would be specifically trained and developed this way. Would it be still an AI fault, though? Or just another human mistake?
While it might be possible to create an intelligent, silicon based life form, AI as we know it won't be able to wreak havoc like the system created by Skynet. The possibility of recreating emotions and feelings in a foreseeable future is so low that it's just not worth thinking about it too much for now. Without these capabilities AI is just a tool incapable of defining its own actions.

When it comes to students getting lower results than a machine it's just the way things are. Apes are known for a very specific form of intelligence, that is pattern recognition. That's what allowed us to come up with all the technological advancements and no wonder why we also base a lot of AI functionalities on the same skill we posses. The reason why machines can outperform us is because we made them for this reason. Personally I think it's a really bad trend, because we as animals are inherently short term planning form of life. Can't wait to see when the power shuts down and the hordes of skill-less people roam the earth unable to even wipe off their buttocks.

Foodocado said…
There are many factors which are responsible for AI better results. One of the most important are emotions like stress. Another factor is our educational system. Students are being learned lots of useless information. I believe they should be learned creativity, logical thinking etc.

Nowadays AI is growing rapidly. For sure it will replace people at work, but when it comes to humans life - I don't know. We should be aware the danger AI brings.
Vladlen Kyselov said…
Artificial intelligence is simply able to remember more information and process the available data faster than a person, but it does not have its own type of thinking, emotions and other qualities inherent in a person.
It seems to me that there is no need to be afraid of artificial intelligence in the next 10 years, or even until the moment when the artificial intellect will be able to independently make its own creative decisions not based on information stored on it`s harddrive.
Unknown said…
The human brain and computer's procesor is two differently working systems and the comparison of them is always be rather innacurate. Nowadays, AI could be better in collecting and processing data, but it can't come up with some creative solutions for some problems. But still, it is evolving and it is doing it very fast.

Call me luddite, but I am frightened about possibilities of what AI can become. Imagine that one day human will create a machine that can feel, think and make moral decisions. How long will it take for this machine to understand that he is better than any human on Earth? What it is going to do with this knoweledge?
Humanity should never give oportunity for AI to replace us. Like many wise people say, one of them is Elon Musk, AI is a threat that could lead to third world war nowadays.
I could not agree that evolution is a random factor. Evolution is a process of adapting biological organisms to the conditions they are living in. Now humankind is in controll of the world, but one day we will lose control of the AI and it can take over.
As I mentioned, AI is on a fledgling level for now. Also I heard about a group of mathematicians that claim that there is not enough atoms for quantum machines to simulate a real brane etc. But if we take a look how young this field of science is, and how far it is starting to develop, we could not be sure that, humanity will find something to bypass this obstacles.
Unknown said…
That was a good speech. It was very interesting. I don't think that we should blame our education system. I think the reason why students get worse results than machine is a memory. Students can't remember 100% of data which they learn,because it's impossible. The machine memory allows to do it.
The machine never has a worse day - stress has no effect for it.
A robot once learned something will always do it perfectly with no mistake. In my opinion this is not fault of a education system.

I don't think so we should be afraid of AI. AI is developing very fast, that is true, but i think in the near future there are no chance to replace people by machines in every field.
Robots can improve our lives, make life easier and better for us. We should develop them and remember that people are creating robots not the other way around.
At my point of view, the study is flawed as the computer has access to infinite information & solutions available on the internet. A smart person equipped with a smartphone and a good understanding of search engine capabilities can do the same job in general knowledge and even essay writing. Math might be harder but still doable using online tools for assistance. It would be interesting to see the AI that can compartmentalize and store information as humans do without having access to the worldwide web.
Maciej Główka said…
In my opinion students get worse results than an AI, because AI has got easy access to incredible amount of resources in the internet. Books, scientific papers, that's all in it. I think people getting worse results is not education system failure.
Should we be afraid of AI? At this moment certainly not. Current AI might be good in one field, for example playing games, helping discover cancer based on photos etc., but it is not so good in resolving all those tasks together. However, this might change very quickly. That's why it is good to create some organisation to watch our AI development and create some safety regulations. As far as I know, there is already such organisation containing people from Facebook, Amazon, Google, etc. (companies, that has greatest AI at this moment)
Unknown said…
Asking why a machine can calculate faster or not forget things? Hmmm it's a machine. It has storage and compute power unmatched compared to humans but in the end we would still win. Our human ability to adjust to any situation is unmatched. So if circumstances are unpredictable, volatile, incalculable a machine gets confused as in "instructions unclear, the cat sleeps on ceiling".
We shouldn't be afraid of AI, we should use it to our benefit as much as possible. Those repetitive task jobs which don't add any value to person doing it? Give it to AI/robots and let pepople do something which will develop, exapnd them (like building more robots and creating smarter AI so people don't have to work but live in utopia ;)).
Unknown said…
When it comes to tests, the memory is the most important thing. AI can remember everything without forgetting, and this is a huge advantage. But when it comes to the open questions, or "describe with your own words" questions, human factor wins.
It's reasonable to assume that machines would outperform the human brains when it comes to data capacity and processing. They are tools. And as with all the other tools with created over the years, we designed them to complement our own bodies, to help us with tasks we cannot perform on our own.

That said, there's a fundamental difference between memorizing and comprehending information. We can create a machine capable of recognizing cat pictures, be we cannot really convene to it the relevance those particular pictures have to us. Not in a way that would make the machine understand.

Since the dawn of computers science-fiction authors have been dwelling on the question: Can machines think? It's important to understand however that machine brains don't work the way human brains do, at least for now. If so, maybe the error of such thinking lies in the basic assumption that we can quantify the machine intelligence using the same measuring devices we use to quantify it in humans.

Another interesting by-product of the sci-fi genre is the fear of sentient machines. I realize it's the most popular outcome, as it makes for a good action movie plot. It's not however the only one.

We might one day find ourselves in a future where all the basic jobs are performed by machines. But does that mean that the humans won't be needed any more? And if so, what would we do with all that spare time?

I found the following video to contain some interesting point on this topic. (made in 2014)

https://www.youtube.com/watch?v=7Pq-S557XQU
sasha | s14611 said…
I'm afraid of AI. Sometimes you don't know what expect from human in certain situation, but I think you'll never know what expect from AI. From my perspective we should create only things that couldn't understand what they do and why.
Unknown said…
This comment has been removed by the author.
Unknown said…
I think the reason behind AI outperforming students is because AI is still "a computer" - something works or doesn't for it, there is only black and white, no gray. If there is no clear answer, the statistics comes in, solving any doubts (like "Todai" did in 5:20).
Besides, machine can memorize everything and store it as "pure information" (e.g. the answer to 1. is "We did something, it was good [500 word essay]"), but for now it can't use it abstractly- it doesn't understand the sense of the essay, it just "knows" that it is the answer for 1.
As for the education system - everyone knows its bad. Even though we are in private university, we had our timetables filled with table-fillers, just so the abstract norms are met, and "we had xx hours of yyy, so in theory we know zzz, even though we will never use it again".

>Do you think we should be afraid of AI? Will it try to exterminate us like in the Terminator movies? Why? Why not?
The AI is as dangerous, as the hardware its connected to - T1000 wouldn't kill anyone, if its AI was written into... a toaster?
If it was a highly advanced, military-grade toaster that shoots .50 cal and nukes any enemies within 500km range, then it would be dangerous, otherwise, the hardware defines the dangerousness of the software...
Unknown said…
I think the reason that AI is passing tests better than the University students is exactly our educational system. It is expected from students to do all the tasks according to a script given to them by a teacher. There is an often repeated mantra " There are no stupid questions", but it's rarely upholded. Those so called "stupid questions" are mocked and disregarded, when they could have been used as an opportunity to draw some conclusions. We are given a picture witch we are suppose to colour and any dot outside of the lines is punished. The learning from mistakes and questions method would look more like carving your individual picture with colours already in. I know it's easier said than done but the knowledge acquired in such a way would be almost permanent. The machine learning, which is the closest we currently are to AI, as far as I know, works on a same principle. It is run many times and each time it doesn't achieve a designated result it changes an element of the equation and checks if it works better. That is why it does better on tests. That is also why a lot of people of science are afraid of AI. If it ever reaches the point of self averness, it would probably draw the conclusion that human race is obsolete. Our specie is not very efficient in almost any way. We are a byproduct of the evolution and we have a lot of junk DNA etc.
Unknown said…
1. In my opinion, the reason is that not every one study what they like. And when you don't like something, it will take you a lot of efforts to be good at something. Not everyone has time and desire for this. But machines don't have any desire, they have data, processes.
2. No. At least, not now. Machines could have started to exterminate us or to do something people don't like if they had been able to create and think/want something. But they can't desire anything, they are only processes, programs. And as the speaker said, they don't read and don't understand.
I agree with the speaker from the video, many people will lose their jobs in the future. Machines could do many tasks with fewer efforts so that companies will economize.
Can I have a robot which will do my homework, please? :)
I couldn't agree more :)
Patryk Pohnke said…
It is not fault of our educational system. We are just humans. Our ability to think and focus depends on many different factors. Amount of sleep, atmospheric pressure, stress level - all these factors may influence on our performance. Obviously, the machine will always be faster in arithmetic, but AI is not as flexible as human being. We understand the context of word and sentences, because there are often situations where some statement has different meanings. Machines will never be capable of doing so. We should let robots do all simple and repetitive tasks while we do something more complex requiring "out of the box" thinking.
I often struggle with such topics they deal with wide implications. As to whether A.I. could pose a threat to humanity not in terms of sheer power to overtake and most likely impose a communist system on to society as to demonstrate their pure evil; but also cause such dramatic shift into life as we know it, rendering the underprepared citizens of the forever to be changed world paralysed with the implications of changes that not many will be able to adapt to. In reality it comes down to people with at least slight authority on the subject such as Stephen Hawking or Ellon Musk are saying that we should be afraid, so we kind of should be because they like really know their stuff. Although in the case of the latter one it may just be a part of some bigger marketing scheme. In response to the second part of the question; I highly doubt that James Cameron was able to predict the future in his movies and I’m willing to say that cow farts helping to trap the heart in the earth’s atmosphere pose a much higher threat to humanity than the scenario which he depicted.
Magdalena Popek said…
When it comes to remembering things machines will always be better than humans. For them there is no way to forget anything. They can use all their knowledge without the possibility of forgetting or making an error. I wouldn't blame the system for the results.
I can't say whether or not should we be afraid of AI. It's too early to state. Such movies are directed just to shock or thrill and most of the time the story has nothing to do with reality.
This is one of the things that a machine is simply put EXPECTED to do better than human, because it's the most basic capability a machine has that humans don't - permanent storage. Brain, while being huge for its storage capabilities, is very slow for accessing information, it's like we can store (remember) almost everything if we try hard enough, but if we don't access that information often enough then it's lost somewhere deep and we're not capable of just grabbing it like that.

AIs don't have this problem. They can storage anything and access anything in given point of time. This means that a human that scored the highest possible degree in this test would perform significantly worse after just several months if not weeks, assuming that he wouldn't do similar tests every day. Machine, once it scores X, will only score X or higher, assuming the actual material for the test is fixed.

Popular posts from this blog

Week 11 [03-09.06.2019] The problem with ecological cars emission in UK

The problem with ecological cars emission in UK Since the adoption of the European Emission Allowance Directive in the European Parliament, all car makers have tried to submit. Since 1992, the Euro I standard has been in force, which limited the emission of carbon monoxide to the atmosphere. The Euro VI standard currently applies, which limits the series of exhaust gases. These include: hydrocarbons, nitrogen and carbon oxides, and dust.   The most significant change was brought by the Euro IV standard. For the first time it introduced the limitation of nitrogen oxides, which are responsible for the harmful compounds of smog.   What is smog?   Smog consists of sulfur oxides, nitrogen and carbon. In addition, solid substances such as suspended dust (PM). Dust suspend in atmospheric aerosols may be in liquid and solid form. These can be particles of sea salt, clouds from the Sahara and artificial compounds made by people. These compounds often come fr

Week 12 (12.01-18.01.15) Are you an early bird or a night owl ?

Owls are nocturnal creatures. They’re wide awake at night and they sleep during the day. If this sounds like bliss to you, then, like about 20 percent of the population who find themselves most active at around 9 pm, you may fall into the same category as our feathered friend. Night owls often have difficulty waking up in the morning, and like to be up late at night.  Studies of animal behaviour indicate that being a night owl may actually be built into some people’s genes. This would explain why those late-to-bed, late-to-rise people find it so difficult to change their behaviour. The trouble for night owls is that they just have to be at places such as work and school far too early. This is when the alarm clock becomes the night owl’s most important survival tool. Experts say that one way for a night owl to beat their dependence on their alarm clocks is to sleep with the curtains open. The Theory is that if they do so, the morning sunlight will awaken them gently and naturally.

Week 4 [06-12.11.2017] This is what happens when you reply to spam email.

James Veitch is a British comedian. In today’s Ted Talk James with characteristic for himself a sense of humor shows how he deals with spam emails and why responding to junk messages may be sometimes dangerous. Questions: What do you think about James’s  way of dealing with spam? Why are junk messages legal, even though it sometimes may be a fraud? Dou you have a problem with spam? How do you deal with with it?