AI, Artificial Intelligence, Data, Data Analysis, Data Science, Information Technology, Software Development

The Importance of Data in Artificial Intelligence (AI)

the_importance_of_data_in_ai-1

AI effectively mimics the reasoning and thought processes of the human brain to replicate in our everyday applications. This is seen a lot in Cybersecurity with task automation and predicting future threat variants.

But just like a car, at the heart of any AI system is the fuel that it is being fed. But rather than gasoline, it’s data and lots of it. So, the focus of this article is to help you understand the key role data plays in AI.

Why Is Data so important?

Good question, here are some key reasons:

1. It’s garbage in and garbage out:

The answer you seek from an AI system is known as the “output”, and the only way you can get this output is by what the inputs are. In this case, it’s in the form of datasets. If any of these are erroneous in any way, your output will get skewed, and your results will send you in the wrong direction.  

2. What are the characteristics of a good data set?

This can be a rather subjective issue to answer, as it depends primarily upon the application of which the AI system is serving. But, in general, the following are features you should look out for when parsing through datasets:

  • It is complete: By this, there are no empty spots or cells in your datasets. Every slot has a piece of data in it, and there are no visible holes in them.
  • It is comprehensive: The datasets are as complete as they can get. For example, with Cybersecurity if your goal is to model a threat vector, then all of the signature profiles from which it emerged must have all of the necessary information.
  • It is consistent: All of the datasets must fit under the variables that it is has been assigned to. For instance, if you are modeling gasoline prices, your selected variables (natural, unleaded, premium, etc.), must have the appropriate pricing data to fall into those categories.
  • It is accurate: This is key. As you will be selecting various feeds for your AI system, you must trust these data sources. If there are chunks that are not accurate, your output will be skewed, and you will not get a correct answer.
  • It must be valid: This is crucial with time series datasets. You don’t want old data that could interfere with the learning process of the AI system when analyzing recent datasets. So, let it learn from recent data. How far back depends on your application. With Cybersecurity, for example, going back a year is typically enough.
  • It is unique: Similar to consistency, each piece of data must be unique to the variables it is serving. For instance, you do not want the same price of natural gas to fall under two different variables.

3. Not all AI systems are built equally:

With actual datasets, we often think of a long series of numbers i.e. quantitative data. But, there are also datasets in qualitative data i.e. videos, pictures, etc. 

With AI systems, these datasets are known as “Structured” and “Unstructured”, respectively. It’s important to note that not all AI systems can handle both of these sets.

But there are systems that can use both and these require little human intervention. So, it’s important to select the right dataset for your system, or your output could yield a different answer than what you imagined.

4. The issue of qulaity versus quantity:

For an AI system to learn and produce the desired outputs; it must first ingest and learn from a lot of data. It doesn’t take a long time to process this, so the question now arises: quality over quantity? The latter is always preferred.

Although it will take the AI system longer if datasets are shorter in nature, you will have some guarantee that your output will be robust and relevant. It’s not productive to feed an AI system lots of data just for the hope that it will learn something from it.

What to take away

Data is key for helping AI systems to learn effectively. But, it’s important to remember that even though you may trust your data sources, you still need to do your due diligence in making sure that the datasets conform to your requirements. 

This requires targeted testing and sampling, and possibly running smaller training exercises to ensure they are being fully optimized. This hard work will pay off in the long run.

Mikaela Pisani

Want to work with Mikaela Pisani? We are hiring :)

Other articles you might like