The world of data science is evolving quickly. In the last year we have been coping with a new pandemic, COVID-19. This resulted in a new applications of data science: fighting a worldwide crisis. Many initiatives were set up, something that we could not have predicted a year ago. But in this article, we will try to look into the future and give some trends that will be seen in 2021.
1. AI is being implemented on a larger scale
An unexpected trend that we saw this year, was the application of AI and machine learning on a large scale. When Covid-19 started to spread, companies, governments, hospitals, and other institutions had to look into new technologies in order to fight the virus. Examples are making predictions about the number of people being infected, or helping to find a cure for the virus.
Because of the experience that organizations now have, and because of what they saw happening in the world, AI will be more commonly used in 2021. Where data science was mostly used by large enterprises, 2021 will open the doors to data science for small and medium enterprises as well.
2. More awareness for ethics and risks: the use of explainable AI
In the past years, deep learning is being used in an increasing number of sectors. Deep learning can be dangerous as well, since the number of deepfakes is growing exponentially. There has become lots of public awareness and attention on the ethics of AI in the last year, especially with documentaries such as Netflix’s The Social Dilemma. Because of this, ethics are becoming increasingly important in AI, that is where Explainable AI comes into place.
Imagine that you are going to the bank to request a mortgage, and your banker says no because the deep learning model tells him so. Of course, you would like to know why you are not eligible for a mortgage, and what you have to do to apply for one. To understand these decisions, an algorithm must be transparent, understandable and explainable. This means that the banker should be able to explain on which criteria a decision was based. This can be done with the support of decision tree methods. But what we will see more in 2021 are tools that make algorithms explainable, such as XAI.
The Data Science QuickScan
TMC Data Science launched the QuickScan to help companies kick-start or accelerate their data science activities. During this scan, two or three of our Data Scientists will come to your organization, map out your business activities and provide you a clear plan for Data Science and AI within just a few days. With over 20 data scientists, there is always someone with the right know-how and specialization to take an organizations’ data science activities to the next level. The QuickScan is an opportunity to explore the potential of data science without having to make a major investment in time or costs.
3. Research and businesses will collaborate more intensively
In the past years, lots of research has been conducted in the field of AI. Many insights about the possibilities of this technology have been gained, but only a small part of the organizations is actually using it. To stimulate this, several collaborations have been set up by businesses, governments, and non-profits. By joining forces, research will develop faster and businesses who want to start with AI can use the expertise of the other parties. Examples of such collaborations are The Netherlands AI Coalition and the AI Innovation Center. Next year we will see even more initiatives and collaborations in this field.
4. Natural language processing will be used more
Speech recognition performances are equivalent to humans nowadays due to deep learning. Thanks to this technology, Google Home or Amazon Alexa’s are found in many living rooms. These technologies are based on Natural Language Processing (NLP), a field focusing on pushing the understanding of human language, like how humans communicate and how to respond to that.
OpenAI is one of the organizations that does a lot of research on this topic. The artificial intelligence research laboratory which was co-founded by Elon Musk, has developed a new technology: GTP3 (Generative Pre-trained Transformer 3). This technique is being used more and more and it basically uses pre-trained algorithms to generate texts. For example, it can write, summarize, and create texts or even computer code. The first iteration of GPT has been showed while not providing the code, solely to create awareness about the dangerous of such algorithms. Especially in our society of fake news, it is important to control the consequences of such a generator in social media. The new version, GPT3, is part of the new business model that enables them to control and monitor how and by whom it is used.
Because of the popularity of speech recognition technologies, the research and development on NLP will increase even further in 2021. The only thing in the way of its success, is that companies are struggling to find the right professionals for investigating these techniques.
Director Data Science | Nanotechnology | Physics, Netherlands
Tel: +31 (0)6 52 89 25 70