AI and Machine Learning predictions for 2022

As the end of 2021 is approaching rapidly, a lot of predictions are made for the new year. For Ai and machine learning I have read many predictions for the coming coming year, but they all boil down to one thing: AI and data are becoming more and more important and will start raising more ethical questions. I think Forbes’ top 10 AI predictions are quite good and hammer home what developments are taking place. What I also like is their backtesting of their 2021 predictions, in other words, how much of what has been predicted has become true. Here are are my three highlights based on their predictions

1. Language matters. Language is a complicated beast because of its ambiguity and many flavors, however huge strides have been made over the last years. Since the advent of Transformer technology language models have taken a giant leap in their accuracy and applications. However much work still has to be done and the world of Natural Language Processing (NLP) is growing rapidly based ont he money invested in parties like HuggingFace and Cohere.
2. The rise of synthetic data and video analytics. Synthetically generated data is a fast growing field that promises to help solve current challenges in data sharing of sensitive data (think of patient training data) and labelling data. According to Gartner synthetic data will account for 60% of all data used in the AI development in 2024 so I expect huge developments for this in 2022. Video is becoming an important data source in the digital world but is still relatively underserved in the analytics world. I expect big developments in applying AI on video, both in analytics as in editing and processing. If you combine synthetic data with video one enters the exciting arena of synthetic media where companies like Synthesia.io are growing very fast.
3. Ethics and responsible AI are taken more seriously. Ethics and responsible AI have on the top 10 prediction lists for the last five years, but it turned out that that a lot was happening at the level of defining AI design principles. This is not a bad thing in itself, but is not very useful for a data scientist in their daily work. For 2022 I expect that the operationalization of responsible AI will become more tangible by the increasing feature sets in frameworks like IBM’s Explainable AI and Microsoft’s Responsible AI initiative like InterpretML. and the rise of MLops platforms like UbiOps and FIddler that have responsible AI features built in the workflow of their products. A lot of large companies have been guilty of ethicswashing but it is not instilled in their DNA yet; still I believe that data ethics will become a core component of every AI project. Last development is the rise of external, independent AI Ethics institutes like DAIR (founded by ex-Google Timnit Gebru) and the Montreal AI Ethics Institute that will help drive the public discussion about AI ethics and governance forward.

In sum, a lot is expected to happen in 2022 and all these predictions have in common that becomes more and more important to understand what AI is, how it works and how it impacts our society. AI Literacy will become one of the skills of the future. I am curious to find out what your thoughts are about the coming year so please leave them in the comments below or contact me directly. I wish you all the best for 2022