What does it take to create an AI?

Artificial intelligence (AI) is a powerful tool in which machines learn to solve problems in a way how humans would. Artificial intelligence helps to perform tasks either too complicated to humans or inefficient.. and to be honest, it includes the majority of tasks that most of us perform at work.

However, it’s not as easy as it might sound. Even if the technology itself currently exists, most companies haven’t made a decision to proactively take advantage of possibilities enabled by emerging technologies. There are multiple reasons behind that but the most common ones are lack of money and understanding of the technology.

To create solutions based on artificial intelligence, companies need people with skills in data science, machine learning, data engineering etc (in-house or outsourced). However, such skills are still quite rare in the market and very expensive. Therefore, not all companies can or even should employ their own team of AI experts. But, as we know, artificial intelligence has become an extremely hot topic during recent years and this means that if there is demand, the supply will be created. This has opened up a new field to investigate and learn. Even though artificial intelligence has existed for more or less around 70 years, there are not yet too many university curriculums where data science can be learned comprehensively. Hence, a lot of data scientists or field specialists have acquired the necessary skills on their own. True, a lot of them already have a university degree in computer science, so the starting point for them is much higher than for others like business people, but honestly, nothing is impossible. When I was carrying out research for my master thesis, I met the Head of Data from one amazing fintech startup. He shared with me a super inspiring story how he started off as a financial analyst in the company with absolutely no single line of code written in his past life (greetings Nicholas, if you’re reading this!). But as there was a need to create a machine learning model in the company, he decided to learn it from scratch and can’t be more satisfied with the decision.

“In God we trust. All others must bring data.”

— W. Edwards Deming, statistician

But.. what is data science then and why it’s so important? Data science is basically a field in which data is extracted, processed and analyzed to generate knowledge and insights. For that, a combination of systems, algorithms and scientific methods is used. One could ask what’s the difference with what people currently do in Excel? While Excel is a great tool to work with data and you can truly do amazing things with it, data science is much wider than that. While you can only use Excel efficiently for the data analysis or visualization part of the work, in data science you also have to be able to affect the data collection, storage, cleaning, and pre-processing part. And if you use the right tools, performing data analytics and visualization becomes very automatic compared to Excel, unless your spreadsheets are interlinked and you pile it up with a lot of formulas, which however makes Excel su-u-u-uper slow and which obviously isn’t a most efficient way for working all the time.

And of course there is machine learning – a process where the good quality data collected and cleaned by the data scientists is used to train the model which as a result (some time later, not instantly) learns to make human-like decisions based on data. And that’s when artificial intelligence is created. Just as easy as 1-2-3? Well – not quite, the process might seem simple, but actually, it’s not. Usually, it takes a lot of time to actually gather all the relevant data necessary for your machine learning model. In many cases the data gathering hasn’t even been started at the moment when the business need is identified.. and then after it’s started and some time has passed, you determine problems with the data quality surface as well.. and then you understand that you need additional data sources to train the proper machine learning models.. and so on. So, as you probably can understand, it’s a never-ending process and most probably you won’t ever have a perfect quality or quantity of data.

However, this is an extremely exciting field with so much to offer! It’s one the freshest fields of science – e.g. in physics and chemistry a lot of discoveries have already been made a long time ago, so it is much more difficult and resource-intensive to advance the field, whereas, for data science it’s much different. Who knows – maybe the next big discovery in data science or artificial intelligence will be done by you? If you don’t ever try, it won’t be the case.. but what if you do?

Leave a Reply