People have always wanted intelligent machines. This desire dates back to ancient Cultures. Various epics and mythological stories mention the presence of their own cosmic weapons, robots, and flying chariots. Engineers and artisans in India, China, Alexandria, Arabia, and the Greek World started making self-moving devices, flying bird models, animated machines, and automatons at around 3rd Century BC.

This long-time sentiment of wanting devices to work for humans made people wonder whether machines will one day become intelligent. One can imagine Artificial Intelligence (AI) like a centralized computer program, one that looks like a human. In short, a robot. Many people wonder, "Is it possible to learn AI without learning how to program or code?" A common opinion is that AI is for all. Everyone should have access to it. AI doesn't differentiate between those who know how to code and those who don't. However, to control AI, one needs to learn how to program or write code. To do that, one needs to learn to differentiate between ANI (Artificial Narrow Intelligence) and AGI (Artificial General Intelligence).

ANI is for accomplishing specific tasks. For example, a program created to identify objects in an image cannot translate text from one language to another. However, AGI might be able to do that. AGI is capable of thinking and performing like humans. Recently, the industry has progressed more in ANI than in AGI.

The underlying desire to create a program or a robot to solve problems or do tasks without much supervision can be called a Transfer of intelligence. In recent years, the increase in the amount of data created has made it possible to build applications that learn from that data.

Enroll In The "AI/Machine Learning" Course Now >>

Is coding necessary for learning AI?

Applications make use of AI to assist and support us in our daily lives. For example, various media, e-commerce, cloud computing, programming, and digital writing service providers make use of AI to provide richer human experiences. We don't need to learn to code to use these applications.

However, we need to learn to code if we want to create applications that make use of AI. Contrary to popular belief, one might not have to learn to write very complex programs, like researchers and the scientific community have been doing for a long time. We can make use of open-source tools, libraries, and platforms. Some of the platforms worth exploring are Google Colaboratory, Azure Notebooks, Kaggle, Amazon Sagemaker, IBM data platform notebooks, and H2O. These platforms are universal, and we can use these for our data. Therefore, we also need to learn about data.

Data that we store in spreadsheets and databases is called structured data, while data stored in text, video, and audio format is called unstructured data. Our AI is only as good as our data. In reality, not all data that we create is useful. It turns out that almost every time, we need to analyze and wrangle data to make it useful, and we can feed it to the programs or algorithms that learn from it.

Which machine learning technique(s) to use

The technique of making machines learn from the data is called Machine Learning. In the context of Machine Learning, if data is labeled, the process is called Supervised Machine Learning. However, the algorithms can learn from unlabeled data too. In that case, it is called Unsupervised Learning. We can use both of these techniques to understand the underlying pattern and relationship within data.

Machine Learning is one of the many techniques that we use to create AI. We can make use of existing machine learning algorithms, or we can create our algorithms. It is easier to make use of existing algorithms that are time tested and proven in the scientific community. We also need to have a basic understanding of probability, statistics, tensors, matrices, underlying mathematics, and mathematical notations to take the first steps.

Conclusion

There is a close relationship between Machine Learning and labeled or unlabeled data. It requires dealing with both of these to reach useful outcomes for the problems that we try to solve. Most of the data in the world are unstructured data. It turns out that the best way to deal with the velocity, variety, veracity, and volume of the data, as mentioned earlier, is Deep Learning. Deep learning is a type of Machine Learning that uses Neural Networks to learn the underlying relationships and patterns within data. Deep Learning requires lesser coding, moving Machine Learning closer to the creation of AI.


Learn AI and Data Science With These Courses:




References and Other Useful Links

  1. https://www.deeplearningbook.org/contents/intro.html
  2. http://deeplearning.net/
  3. https://www.h2o.ai
  4. https://towardsdatascience.com/is-deep-learning-without-programming-possible-be1312df9b4a
  5. https://hai.stanford.edu/research/ai-index-2019
  6. https://analyticsindiamag.com/5-alternatives-to-google-colab-for-data-scientists/
  7. https://colab.research.google.com/github/jakevdp/PythonDataScienceHandbook/blob/master/notebooks/Index.ipynb
  8. https://notebooks.azure.com/
  9. https://www.kaggle.com
  10. https://dataplatform.cloud.ibm.com/
  11. https://en.wikipedia.org/wiki/Historyofartificialintelligence#Mythical,fictional,andspeculative_precursors
  12. https://timesofindia.indiatimes.com/home/sunday-times/all-that-matters/hindu-epics-are-full-of-ai-robots-legend-has-it-that-they-guarded-buddhas-relics/articleshow/68648962.cms

Start learning with Cybrary

Create a free account

Related Posts

All Blogs