My Top Eight AI Predictions for 2022
Here are my top 8 Predictions for the development of AI in 2022.
1) Applied AI / Application Prediction
We will see more use of AI being applied to solve some of the most challenging problems affecting humanity.
- Medical Applications – The Covid pandemic has opened the door to using AI across many aspects of healthcare, from supply chain to drug discovery. This will now continue as we now have a much wider healthcare crisis with the backlog of non-covid medical procedures and diagnosis required. This is compounded with an ageing population in many counties that only looks to make matters worst. Delivering healthcare at home will becoming more mainstream, with technology providing more services remotely at home, especially monitoring and diagnosis.
- Sustainability and Green Tech – With more real focus on climate change and the challenges of population growth on the food industry, AI technologies will be used to provide a range of solutions here. From automated agriculture and energy conservation to designing new products and manufacturing methods that reduce waste.
2) Research Prediction
However, we are also seeing how the use of algorithms have been impacting our humanity. Using data analytics to determine our human behaviours and emotions, technologies have been used to predict all manner of things, from the best time to send us a special offer, to what news articles to show us to persuade us to potentially change our voting in general election. As with so many powerful technologies, they can be used for a range of purposes, both good and bad. Making algorithms work in a more considered and balanced way will be a great milestone for the AI community to achieve in 2022/2023.
We will also see significant advances in generative algorithms, able to create new products and inventions that will empower us and accelerate our own abilities to innovate. Not only will AI generate new artistic works, from books, music and videos, but it will produce new scientific discoveries.
3) The AI Investment Landscape
There is always money for good ideas, however, we have seem little real innovation in the AI startup ecosystem recently, with the natural focus being on applying tried and tested AI capabilities on different datasets and applied problems. This is going to change over the next couple of years, with another wave of AI innovations coming to the forefront of the startup lineup. Technologies such as Federated Learning and Adaptive Learning will come into play as we build larger AI systems that span much wider than our previously very narrow targets. Continued focus on Personalisation, User Privacy, Ethics and AI Governance will also play into those startups that get funded in 2022.
4) Related Emergent Technologies – Quantum Computing
We will see more and more advancements with Quantum Computing this year. The Big Tech firms are racing to lead in this field, as it is potentially the next wave of computational advancement, able to disrupt many areas of computing, from Artificial Intelligence, to Cyber Security and Optimisation. Quantum Startups are beginning to be acquired by the big tech firms, and this will only continue to increase over the coming few years. Quantum computing in the cloud is already being made available by all of the cloud platforms.
The potential of quantum is astonishing, especially for the performance gains to train a very large deep neural network. Not only will this significantly reduce the power consumption of training these huge networks, but the speed increase will be a game changer to allow data science researchers to try different architectures and configurations so much quicker than we can currently achieve.
There are many other related emergent technologies that will also play a part in 2022. From Augmented and Virtual Reality and the Metaverse, to humanoid robotics and human-to-machine interfaces. Edge computing (IoT) and 5G will also enable increased mobile AI applications, especially in the transportation and supply chain sectors.
5) Privacy vs Personalisation
Data Privacy – Consumers are becoming more aware of the use of their own consumer data, including the meta data they generate by interacting online shopping and social media platforms. This data is driving alot of the recommendations that are produced, filtering products that it is believed to be most acceptable / desirable.
Service Personalisation – The demands of consumers continues to increase, with us all expecting a personalised service, tailored to our own specific needs. For companies to be able to deliver this at scale requires AI to perform the individual analytics to enable and facilitate such capabilities. This requirement spans many industries, from financial services to retail and business services.
There is a natural trade-off between privacy and personalisation, that has until now required users to sacrifice some of their privacy to supply data that can be used to generate a more personalised experience. The balance of this relationship will shift this year, with various technologies making it easier to provide a personal experience for users without having to share huge quantities of personal data.
6) ML Ops Improvements
The development of ML models, from initial prototype or pilot into live production models at scale, require not only extensive data science and data engineering skills, but also significant ML engineering to deliver not only the model into production at scale, but to enable the appropriate monitoring and re-training needed over time. ML Ops, like its name-sake DevOps, is all about creating the tools, frameworks, processes and pipelines that can automate these tasks.
Over the last year or so we have seen improvements in this area, however, this seems to be a very bespoke part of the data science practice and I believe we will seen further focus and advances on this in 2022.
7) AI Driven Productivity
AI will drive more automation and productivity improvements for companies of all sizes in this next 12 to 24 months. The benefits to automate standard workflows, processes and tasks have been highlighted during the last two years of the pandemic. Digital transformation and migration to the cloud will further open opportunities for data insights, automation and predictive models that AI technologies enable.
Intelligent Process Automation, a fusion of standard Robotic Process Automation (RPA) with AI capabilities is making complex automation a reality now for companies large and small.
This automation will also move more into our personal and private lives too in 2022.
8) The Next AI Winter Prediction – 30mins to Midnight
Using the doomsday clock analogy, I believe we are 30mins to midnight for our next AI Winter. Why do I say this? Well yet again we have concentrated our focus on just one area of Artificial Intelligence, that being Deep Learning. We do this despite their being a number of warning signs about its depth of capability. The technology has a number of challenges. It produces potentially brittle models. It lacks understanding and reasoning that is required for true AI. While huge models like GPT-3 have shown some outstanding abilities, its real NLU is limited.
While we might have a risk of another AI Winter, I believe there are enough AI researchers and practitioners that understand we need a more balanced and broader perspective on AI technologies to ensure we do not suffer from being one dimensional or short sighted. However, it is all of our responsibility to control the hype and educate people about the need for more emphasis on other parts of the AI spectrum.