Artificial Intelligence 

Artificial intelligence (AI) has entered the vision of the overall public, and that we can see many AI-related products in our lives. like Siri, AI beauty, AI face change...

Although everyone listens tons , most of the people don't understand AI, and there are even some misunderstandings. this text won't involve any technical details to assist everyone understand the character of AI.

What is artificial intelligence?

Many people have some misunderstandings about artificial intelligence:

  1. The robot within the movie may be a typical representative of AI 
  2. Artificial intelligence seems to be omnipotent
  3. Artificial intelligence will threaten the survival of mankind within the future
  4. ……

The reason why you've got many misunderstandings about AI is especially because you simply see some people’s speech, but don’t understand the essential principles of AI. this text will assist you understand the essential principles of AI. The essence of things is usually not what everyone said. So complicated.

We use traditional software to match with AI . it's easier to know with a frame of reference.

Traditional software VS artificial intelligence

Traditional software

Traditional software is that the basic logic of "if-then". citizens sum up some effective rules through their own experience, then let the pc run these rules automatically. Traditional software can never exceed the boundaries of human knowledge, because all rules are made by humans.

To put it simply: traditional software is "rule-based" and requires artificial setting of conditions and telling the pc what to try to to after meeting this condition.

This kind of logic is extremely useful when handling some simple problems, because the principles are clear, the results are predictable, and therefore the programmer is that the god of software.

However, real world is filled with various complicated problems, which are almost impossible to unravel by formulating rules. for instance , the effect of face recognition through rules are going to be very poor.



Poor Artificial Intelligence

Artificial intelligence has now developed many various branches, and therefore the technical principles also are diverse. Only the most well liked deep learning at the instant is introduced here.

The technical principles of deep learning are completely different from the logic of traditional software:

The machine sums up rules from an outsized amount of "specific" data, sums up some "specific knowledge", then applies this "knowledge" to real scenarios to unravel practical problems.

This is the essential logic of the event of AI to the present stage. The knowledge summarized by AI isn't like traditional software, which may be expressed intuitively and accurately. it's more just like the knowledge learned by citizenry , which is more abstract and difficult to precise .


The above statement remains quite abstract, the subsequent will assist you understand thoroughly through several aspects:

Artificial intelligence may be a tool

AI is that the same because the hammers, cars, and computers we use. it's essentially a tool.

Tools must be employed by someone to be valuable. If they exist independently, they're worthless, a bit like a drill in a toolbox. nobody wields it with none value.

The reason why the entire society is talking about AI is that it greatly expands the capabilities of traditional software. there have been tons of things that computers couldn't do before, but now AI can do them.

Thanks to Moore's Law, the facility of computers has risen exponentially. As long because the computer can solve and participate within the links, productivity has been greatly improved, and AI allows more links to catch the express train of Moore's Law, so this alteration it's of extraordinary significance.

But regardless of how it changes, traditional software and AI are tools that exist to unravel practical problems. now has not changed.


Artificial intelligence only solves specific problems

"Terminator" and "The Matrix"...Many movies have appeared against sky-defying robots. this type of movie makes everyone feel that AI seems to be omnipotent.


The actual situation is: current AI remains at the stage of one task.


Single task mode.


Landline for phone calls, game consoles for games, MP3 for taking note of music, navigation for driving...


Multitasking mode


This stage is analogous to a sensible phone. you'll install tons of apps on one phone and do tons of things.


But these abilities are still independent of every other. After booking a ticket on the travel app, you would like to line the timepiece with the timepiece app, and eventually you would like to use the taxi app to call a taxi. The multitasking mode is simply a superposition of one task mode, which is way from human intelligence.


Integrate


You are playing accompany your friend, and you discover that your friend is during a very bad mood. you'll have won easily, but you deliberately lost to the opposite party and kept praising the opposite party because you don’t want to form this friend more depressed or more depressed. Irritable.


In this trivial matter, you've got used a spread of various skills: emotion recognition, Go skills, communication, psychology...


But the famous AlphaGo will never do that . regardless of what the opponent's situation is, albeit it loses the sport , AlphaGo will win the sport ruthlessly, because it can do nothing but play Go!

For instance , military knowledge are often utilized in business, and biological knowledge also can be utilized in economics.


Know it, but do not know why

The current AI is to summarize knowledge from an outsized amount of knowledge . This crude "inductive method" features a big problem:

Don't care why  


Scams like Ponzi schemes take full advantage of this!

  • It uses super high returns to draw in leeks, then allows everyone who gets up early to participate within the transfer of money;
  • When onlookers discover that each one participants have actually made money, they will simply sum it up as: historical experience shows that this is often reliable.
  • So more and more people were jealous and joined until at some point the liar ran away.

When we use logic to deduce this matter, we will draw the liar's conclusion:


  • Such a high return doesn't conform to the laws of the market
  • Steady profit without losing? i do not got to take the high risk of high return? Doesn't seem reasonable
  • Why does such an honest thing fall on me? Doesn't seem right

It is precisely because the present AI is made on "inductive logic", it'll also make very low-level mistakes.

  • Left: The occlusion of the motorcycle makes the AI mistake a monkey for a person's .
  • Middle: The occlusion of the bicycle caused the AI to mistake the monkey for a person's , and therefore the jungle background caused the AI to mistake the handlebar of the bicycle for a bird.
  • Right: The guitar turns the monkey into a person's , and therefore the jungle turns the guitar into a bird
The image above shows the effect of a guitar on PS during a photo of a jungle monkey. This causes the deep network to mistake monkeys for humans and guitars for birds, presumably because it believes that humans are more likely to hold guitars than monkeys, and birds are more likely to seem within the nearby jungle than guitars.

      It is precisely due to inductive logic that an outsized amount of knowledge must be relied upon. The more data, the more universal the experience summarized. AI isn't a fresh thing, it's been developed for decades! Below we introduce the three most representative development stages.