This century has so far been marked by incredible technological developments. That is easily seen by the impact of tech across every aspect of our lives. From economy to healthcare, tech continues to play an increasingly larger role in our society, and we can only assume that that trend will continue to grow as progress is made every day.

One of the most significant areas of advancement in tech right now is artificial intelligence. Even though it’s still at a relatively early stage, considering all the development yet to be made, AI has quickly become a revolutionary tool across multiple fields. 

A key area of application for AI so far has been STEM. Researchers in these areas are now able to solve problems that previously seemed unsolvable, making significant discoveries at a much quicker rate, thanks to AI. 

Across the next few weeks, we’ll be diving into this topic in-depth, discussing how AI is changing each major area of STEM, starting with scientific research. 

Different Views of AI (and AI Researchers)

When it comes to integrating AI in other fields, it’s not always as simple as could be expected. While AI shows incredible potential in facilitating and even exponentiating STEM research, there is still some resistance and even friction when considering the integration of this technology.

For one, some researchers demonstrate a fear, common to many individuals when considering AI, that they’ll be replaced by this technology eventually. While this may not be applicable to most professionals, it’s only natural that in some cases this is an area of concern.

The fact is that AI does present a lot of potential and is already capable of carrying out very complex tasks. Even so, we’re nowhere near a point at which something as nuanced and intricate as scientific research can be performed exclusively by AI. At this point, we can’t even assume that this will ever be the case.

At this point, we should view AI as a tool that can improve our work, helping us carry out tasks more easily and effectively. But make no mistake – AI is a very powerful tool, unlike any other that humans have created. It isn’t limited to simply doing as we instruct it as it can learn, make predictions and adapt. It’s important to understand that so that we can make the best use of it.

Still, the challenges don’t stop there. As scientists begin to consider the integration of AI into their projects, they may decide to work together with AI researchers. However, some aren’t necessarily looking for a partnership with a colleague as much as they simply want someone to program AI according to their directives. 

When choosing to work with other professionals that have a higher level of expertise in AI, researchers must look towards them as partners with whom they collaborate to solve their problems and conduct their research. 

Whether you choose to partner with an expert or to work by yourself, it could always be worth considering learning the basics of programming, AI, and machine learning. At the very least it will enrich your knowledge and help you understand the basis of your work. 

Taking Advantage of AI

One main question remains – how can we specifically use artificial intelligence to enhance the work carried out in STEM? 

To start off, we’ll look at the application of AI in scientific research, from the first steps of experiment design and hypothesis generation, all the way until the development of products and technologies as a result of this research.

As you start your research, it’s key to review the literature already available so that you can understand what’s already been found and what areas require additional work. In the end, scientific progress isn’t about any one single individuals, but rather about the collective work of scientists, piecing together to enhance our understanding of science.

One significant problem that scientists face however is the growing volume of literature that is available. While each piece is essential to the puzzle, trying to sort through and read about so many experiments has become an extremely laborious task, particularly as you’re trying to understand what may even be useful to you and the research you’re looking to conduct.

All of this can be made easier through the use of AI algorithms, which can analyze scientific papers and summarize the most important bits of information.

These natural-language processing programs, therefore, help make literature more accessible to researchers, saving a significant amount of time and effort.

You can try an example of this type of program here.

After reading the already existing papers relevant to you, comes the time to create a hypothesis and plan your own experiment. As any scientist will know, this is a crucial part of the process as it will dictate the course of the rest of the work.

AI can be used to help both to generate the right hypothesis but also to then create the most efficient experimental design to test it, from the ideal sample size to the most effective methodology.

You can even take this one step forward by applying AI to help manage the resources and workflow of your research, facilitating decisions and automating some of the work. This way, AI can even act throughout the experiment, adapting to the data as it is created. 

It may, for example, alter the parameters of the experiment and even the hypothesis according to the results achieved up to a certain point, improving the output of the experiment, maximizing the data obtained, and employing resources more effectively.

Even the resources we use today, whether they be programs or pieces of equipment, have been improved through the use of artificial intelligence. New features have been introduced to these, from microscopy to DNA sequencing tools, enhancing their capabilities. 

As tools become more advanced and we’re able to look into wider and more complex problems than before, we’re faced with an increase in the volume of data collected. Once again, we can use AI to combat this issue.

Machine learning is essential an application of AI that enables systems to learn from data rather than having to be directly programmed to carry out a specific task. Essentially, systems are able to analyze large data sets and recognize patterns within that information. 

This can be applied to scientific research and the large data clusters created during experimentation, using AI to find patterns which humans likely would not be able to find on their own given the volume of data and intricacy involved. 

Through this application of artificial intelligence, scientists have been able to make complex connections. A prime example of this is found in genetics.

In 2003 the human genome was sequenced for the first time, after 13 years of work. Today, there is an entire discipline dedicated to the study of the genetics of entire populations – population genomics. 

This field handles large sets of data, collected from thousands of individuals, so AI can be very useful to find links between their genetic information and multifactorial traits, which are controlled by several factors, or to find patterns related to people’s ancestry.

As AI identifies patterns, it is able to create models from the data. When models are used in conjunction with simulation and reasoning systems, which draw conclusions about the data through logic, researchers can create real-world simulations even of complex systems.

Nowadays, it’s inclusively possible to use AI for “inverse design”. That essentially means that artificial intelligence can be applied to discover the factors leading to a result. This would be like trying to discover what ingredients are used to bake a cake by tasting it. 

Overall, this helps reduce the need to rely on people to come up with hypotheses that explain a given result and test every single one, as AI can simulate this process and essentially work backward to discover the most probable hypothesis. 

One of the main areas benefiting from AI’s simulation capabilities is drug discovery. Learning from the data available on molecules and compounds as well as known disease targets, AI can be used to help uncover new substances with therapeutic potential.

A great example of this was the creation of a drug, fully engineered by AI, to combat Idiopathic Pulmonary Fibrosis by the biotechnology company Insilico Medicine. 

First, AI was used to analyze existing literature in search of a protein involved in the pathology of this disease that could, therefore, serve as a target for prospective therapeutics. Then, researchers used a different program to design a molecule that acted on this target specifically, thus serving as a drug against Idiopathic Pulmonary Fibrosis. 

The drug created is currently in clinical trials. You can learn more about it through this link.

Another significant example that demonstrates the modeling and simulation power of AI is AlphaFold. This is an AI program created by DeepMind able to predict the 3D structure of proteins simply from their amino acid composition. 

This has been revolutionary given that simulating protein structures has long been an immense issue for researchers, given its complexity. Until the introduction of AlphaFold, finding out the structure of protein could take researchers months or even years. Now, it’s almost instantaneous.

In this case, this application has had tremendous practical effects on our lives already as AlphaFold was used in 2020 to predict the structure of the spike protein of the Sars-Cov-2 virus, used in the development of Covid-19 vaccines.

You can find out more about AlphaFold through this link. 

These principles may also be applied on an even larger scale, enabling a full understanding of how entire systems function, including the underlying processes behind major phenomena.

A great example of this would be the application of AI’s simulation capabilities to study the human body. Through this, we may be able to gain a full understanding of how each biological system functions within our body, and eventually create faster and more precise ways to monitor our organism. 

This would allow us to control the state of our health, as well as detect and treat some diseases, like cancer, earlier and more effectively.

Overall, it’s clear that artificial intelligence is deeply impacting the way scientists approach research and their fields as a whole. Looking forward into the future, it seems that AI will continue to grow and help advance scientific innovation.

STEAM in AI and the Scientists of the Future

At the DataEthics4All Foundation, we’re focused on helping the next generation access and succeed in STEAM, Data, and AI fields. 

To this end, we started the STEAM in AI Movement, which includes three main initiatives for young people – the STEAM in AI Research and Build Program, the STEAM in AI Clubs and Chapters, and our Community.

The STEAM in AI Research and Build Program is aimed at high school students with a particular interest in Data Science, Artificial Intelligence, and STEAM, offering an opportunity for them to receive mentoring and learn through experience. 

Further, through the STEAM in AI Clubs and Chapters initiative, our foundation helps high school students start a STEAM in AI club in their school or a chapter in their region, so that they can have easier access to experiences and activities in STEAM areas.

Finally, students can also become members of our Community, which grants them access to the Ethics 1st Institute, where we host live and on-demand courses related to tech, AI, data science, and ethics. Members can also connect and interact with peers, as well as Data and AI leaders. 

With the development of AI and the growing potential of its application in other areas, gaining more knowledge and experience with it is becoming a valuable asset for young professionals. 

The STEAM in AI Research and Build Program and related initiatives could be a great starting point for students looking to pursue a career in STEAM or even AI specifically. To learn more information, visit our website or book a consultation.

Stay tuned for our next article where we’ll continue to explore this topic in depth, this time focusing on the impact of AI across several life sciences, including genetics, ecology, and zoology.