Main 2021 Developments and Key 2022 Trends in AI, Data Science, Machine Learning Technology
Tags: 2022 Predictions, AI, Carla Gentry, Data Science, Doug Laney, Kate Strachnyi, Kirk D. Borne, Machine Learning, Predictions, Tom Davenport, Trends
Our panel of leading experts reviews 2021 main developments and examines the key trends in AI, Data Science, Machine Learning, and Deep Learning Technology.
The question we asked our experts was
What were the main developments in AI, Data Science, ML in 2021 and what key trends do you see for 2022?
Some of the important topics mentioned were AutoML, Automation and RPA, Applications: more and better, Data mesh, data fabric, and data-centric approaches, Deep Learning and PyTorch, GPT-3 like systems, Edge devices, External data sources, Low-code and no-code AI, MLOps, Responsible AI, Talent gap and staffing, and Quantum computing/machine learning.
Here are our expert opinions, with contributions from Marcus Borba, Kirk Borne, Tom Davenport, Carla Gentry, Doug Laney, Pierre Pinna (IPFConline), Ganapathi Pulipaka, Kate Strachnyi, and Mark van Rijmenam.
Agree? Disagree? Please comment below!
Marcus Borba, Global Thought Leader & Influencer in AI, Machine Learning, Data Science.
Currently, AI is being used in most areas of knowledge, and is rapidly evolving and expanding into several new domains, such as multimodal applications and smaller devices, making technological development grow at a pace never seen before. By 2021, the concern for transparent and trustworthy AI has caused the concept of responsible and explainable AI to grow in importance helping to understand how AI works. Another area that is seeing strong growth is hyperautomation, using multiple interconnected technologies, such AI and machine learning, IoT, RPA and augmented analytics, improving human-machine collaboration and enabling better productivity.
For 2022, I believe Quantum Machine Learning will be the next big thing. Quantum ML is the intersection between quantum computing and AI, and will enable the creation of more powerful machine learning and AI models. With the development and support of Machine learning with quantum computers by large technology companies, it will be possible to have resources accessible through cloud models. The growing evolution and adoption of Intelligent Process Automation will provide companies with greater agility, adding more value than just saving time, improving operational efficiency and helping to simplify processes. Low-code and no-code AI solutions will also increasingly enable companies to start using AI, in order to develop AI models faster and at lower cost.
Kirk D. Borne, Data Scientist @DataPrime_ai, Chief Science Officer. Global Speaker. Founder @LeadershipData. Top #BigData #DataScience #AI #IoT #ML Influencer. PhD Astrophysics.
Big trends in 2021 include rapid expansion of AI / ML in medical applications, finance applications, customer-facing applications, logistics/supply chain applications, and the Metaverse (gaming applications). i.e., Applications, Applications, Applications, as organizations are demanding more productivity and value from their data, AL, ML, Data Science assets and investments.
For 2022, this continues, but the talent gap is providing increasing headwinds that may be an impediment to this increasing momentum. Consequently, we can expect organizations to opt for more low-code/no-code deployments (such as AutoML) to harness and maintain the growing AI/ML momentum, which will further differentiate the AI leaders from the AI laggards, which supports the old adage "getting things done is better than getting things perfect".
Thomas H. Davenport, Distinguished Professor, Babson College; Visiting Professor, Oxford Said Business School.
One of the major developments in 2021 was the rise of MLOps to monitor models once they are in production. In part this was necessitated by the pandemic, in which predictive demand and supply models for many companies were no longer accurate. Detecting model drift is an important MLOps capability. But the rise of MLOps is also an outgrowth of increased dependence within many companies upon machine learning models to run their businesses, and keeping track of models and how they perform is becoming key to business success.
This will continue in 2022, and companies will use AutoML, MLOps, and other tools and process improvements to professionalize ML development. They will also make more clear "who does what" in AI development and deployment, creating specialty roles for model creation, model deployment, model monitoring, and so forth. These tasks can't all be left up to data scientists.
Carla Gentry, Data Scientist and Data Advocate at Zuar.
As Covid continues to be a factor in all we do, companies have pivoted on staffing, remote versus in-house, data in the cloud or on-site. One thing that has become clear is the fact that we are at a crossroads with data and its ability to make a difference. Data sprawl has become a real and costly problem inside organizations, and it is hurting innovation. Hakkoda survey,
"6% of business and IT leaders labeled their data organization and processes a dumpster fire."
2022: we will continue this data siloed path since each department will always have its own agenda and needs. Companies that can bring multiple data platforms together will be vital for businesses who do not have the talent, time, or ability to tie all their data together to make executive decisions. Throwing good money at bad ideals is no longer acceptable, ROIs must be attained. Let us embrace innovative technology, but let us also keep in mind that data itself is useless unless you do something with it!
Douglas B. Laney, Innovation Fellow, Data & Analytics Strategy. Author of Infonomics — CIO Magazine's must-read book of the year.
In 2022, the major emphasis in AI, Data Science and ML will be tapping the ever-growing prevalence of external data sources. Shifts in global and local economies due to the Covid-19 pandemic and the discrepant, ever-changing national and local responses have rendered the rudimentary analytic models of many organizations useless, or worse - detrimental to the business.
Subsequently, businesses have been compelled to scrap their trend-based analytics models that rely on the company's own historical data in favor of driver-based models that consider leading indicators of performance. Instead of continuing to stare at their own navels, businesses now are compelled to look outside the organization to diagnose, predict and prescribe customer behavior, inventory levels, supplier availability, talent needs, equipment behavior, competitor response, partner capacity, and so forth.
This means that the data scientist's and AI developer's new best friend (if not someone quite as sexy) is an individual dedicated to researching and acquiring new high-value alternative data sources. 2022 will be The Year of the Data Curator.
Pierre Pinna, CEO IFFConline, Digital Transformation Consulting.
The year 2021, for the IPFConline team, was marked by the "war" of OpenAI GPT-3 type systems in natural language processing, models which are increasingly powerful and data-intensive. And also by the release of an extraordinary new paper on GFlowNet Foundations from Yoshua Bengio team, proposing a big advance in enabling causal discovery & reasoning in machine learning.
For 2022, we believe that the way of the AI story will be the desire to advance in the interpretability of models. And so, that researches will move more and more towards hybrid models between symbolic AI (with high level of understanding for us humans) and actual statistical deep neural networks. A solution that will allow the trust of the actors for a trustworthy adoption of responsible systems.
Dr. Ganapathi Pulipaka, Chief AI HPC Scientist, Speaker, Bestselling Author.
PyTorch Lightning enabled deep learning research community to spend more time on research and less time on engineering. This year, it advanced as a popular deep learning framework to read and reproduce the research on distributed hardware and high-performance computing machinery. PyTorch Lightning has been a force to reckon with on GPUs, TPUs, and CPUs. Natural language processing advanced in 2021 with PyTorch Tabular. PyTorch Tabular works on Pandas dataframes in deep learning, as there is no ready-to-use library like Scikit-learn for deep learning. It runs on PyTorch and PyTorch lightning.
In 2022, we can expect federated quantum machine learning with distributed training across several quantum computing machines with quantum convolutional neural networks. The research on graph neural networks continues to accelerate in 2022. Enterprises continue the adaption of PyTorch, TensorFlow, Python, and reinforcement learning algorithms such as DQN for various industries in 2022.
Kate Strachnyi, Founder, DATAcated.
Main 2021 Developments
- We saw more companies understanding the value of data and embracing a data-informed culture.
- Innovations within data platforms make it easier for the business to understand their data and reduce time to insights and data-driven decisions.
- Rise of data mesh, data fabric, dataware and a move towards data centricity; moving away from being app /model centric.
- Increased demand for data scientists who can handle the creation, handling, and assessment of large synthetic datasets; because of the innovations within the metaverse.
Dr Mark van Rijmenam, Future tech strategist, keynote speaker, 3x author & entrepreneur.
Artificial intelligence and machine learning will play an increased role in automating our jobs. For the non-coders or non-data scientists, this will result in more advanced robotic process automation that will enable employees and managers doing office work to further automate their own work without knowing how to code. As a consequence, employees will be able to get more work done because most of the monotonous jobs can now be automated. However, if employees and management without coding expertise can use AI and ML to automate their jobs, of course, those with these skills can also apply the same tactics.
Of course, this can be linked to using GPT-3 to write code faster but in 2022 we will also see more advanced, automated and intelligent hacking. Increasingly, hackers will turn to AI to get more work done, penetrate organisations faster and steal more data. At the same time, the people defending their companies will also turn to AI so more and more both hackers and IT security staff will use AI, resulting in a battle taking place at the speed of light.
Check also AI, Analytics, Machine Learning, Data Science, Deep Learning Research Main Developments in 2021 and Key Trends for 2022, published earlier this week.
- AI, Analytics, Machine Learning, Data Science, Deep Learning Research Main Developments in 2021 and Key Trends for 2022
- AI, Analytics, Machine Learning, Data Science, Deep Learning Research Main Developments in 2020 and Key Trends for 2021
- Main 2020 Developments and Key 2021 Trends in AI, Data Science, Machine Learning Technology