Technically Speaking

The Official Bigstep Blog

Subscribe to our newsletter

The Current State of Artificial Intelligence - Infographic

Artificial Intelligence (AI) is a broad term that defines any machine capable of performing human functions.

The forms of AI that we encounter today are very specific, are capable of self-improving on a particular task, but this Narrow AI can only perform one or several human functions. Many things were considered AI and we now regard them as normal, non-intelligent machines:

Narrow Artificial Intelligence

Artificial Intelligence (AI) is a broad term that defines any machine capable of performing human functions.

The forms of AI that we encounter today are very specific, are capable of self-improving on a particular task, but this Narrow AI can only perform one or several human functions. Many things were considered AI and we now regard them as normal, non-intelligent machines:

  1. Pocket calculator: can perform computations, a human intelligence task.
  2. Heat Sensors: can detect heat and the presence of some living organisms.
  3. Face Recognition: can recognize a human face; since millions of phones do it each hour, people do not regard it as AI anymore.

Machine Learning

Machine Learning (ML) is a way of programming a machine by feeding it data sets in order to make it proficient at a particular task (like playing chess). Machine Learning relies on statistics and big numbers.

Training a machine:

  1. Feed the machine a data set.
  2. The machine gives a result.
  3. Correct that result and adjust the machine.
  4. Go back to step 1.

After enough data sets are put in and the machine is trained, that machine can begin to learn on its own from new data sets.

Data Sets ->>> Machine Learning ->>> Processed Data

New Data Sets ->>> Machine Learning ->>> Better Results

Deep Learning & Neural Networks

Deep Learning (DL) is a form of machine learning that focuses on learning data representations. Deep Learning uses artificial neural networks that remotely resemble the way the brain uses neurons. This becomes very useful when working with millions of images for example, because task-specific algorithms fail in this case.

AI and Humans

There are significant differences between the human brain and AI which lead many scientists to call the use of the word Intelligence in AI a misnomer.

  1. The neurons in human neural networks can connect to each in multiple ways that we don’t even begin to grasp, while the neurons in artificial neural networks connect to each other in specific ways.
  2. As a consequence, the human brain is adaptable to many new and different tasks while machine learning and deep learning can perform one task very well.
  3. Thus, the brain thinks while the machine learning only gets better at performing a specific task.

Narrow AI is very specific as described above. An Artificial General Intelligence is able to perform most of the functions of the human brain (or even more) and it is postulated that it will develop a form of consciousness.

You can regard AGI as a mastermind capable of performing most tasks that Narrow AIs perform, but correlated and at the same time. An Artificial General Intelligence (AGI) is more the kind of AI we are accustomed to from Sci-Fi. However, an AGI seems a very, very long way ahead of our current development, contrary to claims from popular figures like Elon Musk.

Some Use Cases

Let’s take a look at current applications of machine learning and get a glimpse of the future.

  1. Chatbots like Siri, Alexa and Google Now are chatbots that are helping you right now. In the future, most company support will be chatbots.
  2. Self-driving cars use IoT devices and sensors, have short-term memory compared to other machine learning technologies, and will be your future way of moving from place A to place B or even shipping merchandise. Transport services like Uber or Taxify also use machine learning to connect drivers to clients.
  3. Online fraud detection, illegal content or spam filtering is done through machine learning algorithms because there is not enough manpower to review all the content being generated, or even if there is, the time a human takes to react is too slow for the speed things are moving at.
  4. Optimization of cost efficiency, supply chain management, car or air traffic in cities, and so on and so forth is done with machine learning and big data.
  5. Customer profiling done by Netflix, Spotify, Amazon, Facebook, and Google Search is based on machine learning that paints a profile of you. This is how your news feed is created, your search is refined, and more suited music, movies, products, services or ads are suggested to you.

 

 

Got a question? Need advice? We're just one click away.
Sharing is caring:TwitterFacebookGoogle+PinterestEmail

Readers also enjoyed:

3 Real-Live Big Data Success Stories That Prove You Can Make It Happen

Have you been sitting on the sidelines, waiting to see whether others achieved that promised success with big data, or instead fell flat on their faces?…

Leave a Reply

Your email address will not be published.

* Required fields to post your comments.
Please review our Privacy Notice in order to understand how we process your personal data and what are your rights in this respect.