5 AI and machine learning trends to watch in 2022

14 Dec 2021

Image: © LizFoster/Stock.adobe.com

Verne Global’s CTO talks about how the AI landscape has changed so far and what he sees coming next for the industry.

Advancements in AI have brought about some big changes in tech in recent years, from image recognition to developments in healthcare and smart manufacturing.

Tate Cantrell, CTO of data centre operator Verne Global, said a key trend in the AI space has been its convergence with high-performance computing.

“This powerful combination is pushing the boundaries of scientific research. A great example of this is Peptone, which has developed an algorithm to work out how proteins behave and aggregate in the human body,” he said. “This research is being used to accelerate the development of vaccines and other medicines, by diminishing the need for costly and time-consuming lab work.”

There’s no question that AI and machine learning have already come so far. And with technology constantly evolving, developments in this area are set to continue.

As we look towards 2022, Cantrell gave SiliconRepublic.com his predictions for what to expect from AI and machine learning in the coming year and beyond.

AI models will get larger

AI and machine learning models uses huge volumes of data and Cantrell said these models will continue to expand and draw on even greater data sets to make increasingly accurate decisions.

“For example, the continued evolution of OpenAI’s large-scale generative pre-trained transformer (GPT) models is one to watch. These very powerful language models can perform a vast array of natural language processing tasks – answering questions, summarising texts and much more,” he said.

“The latest model, GPT3, is 100 times the size of its predecessor, with 175bn parameters, making it capable of writing articles that are considered indistinguishable from those written by humans. There’s no doubt that this represents a massive leap forward and has the power to recalibrate how tasks are divided between humans and machines.”

AI will spur advances in pure science

Cantrell also said that AI has the potential to trigger interesting advancements in the pure science fields such as pure mathematics.

“Advancements in pure mathematics often require inspiration in the form of recognising a new pattern. And where patterns are the target, artificial intelligence is a terrific tool for the task,” he said.

“We might even see humans start to rethink how they make decisions, based on their observations of how machines work. Indeed, we’ve already seen this in the world of chess where machines have had huge tactical influence over how the game is played, perhaps evidenced most clearly in the playing style of the current world champion and possibly the greatest chess player ever, Magnus Carlsen.”

There’ll also be advances in cybercrime

While many AI developments can benefit society, advancements in technology can be a double-edged sword and one of the more unwelcome trends that is expected to grow in the future is cybercrime.

Cyberattacks are already on the rise and growing more complex, and 2022 and beyond are likely to see more bad actors use AI to hone and personalise their attacks on enterprise infrastructures.

“In turn, we’ll also see their would-be victims invest in AI-based solutions to either thwart these attacks or, in the case of some governments, hack back or even take pre-emptive action. This will be an arena of considerable activity,” said Cantrell.

Sustainability will be key

Another key trend across many areas of technology is sustainability, both within the industry itself and in terms of how tech can be used to increase sustainability.

AI and machine learning are no different, with recent research name-checking AI as an important technology for sustainability in the coming years. From major fashion brands to small start-ups, many companies are already using AI to create a more sustainable world.

But in terms of the sustainability of the tech itself, Cantrell said the need to report on environmental, social and governance goals may cause some companies to investigate the carbon footprint of their AI programmes.

“If these models are located in data centres or in on-prem locations that draw power from grids burning a high proportion of fossil fuels, the results might not be very palatable,” he said.

We’ll see a more strategic-thinking AI industry

AI and machine learning are terms that have become so synonymous with emerging tech that one trend Cantrell has noticed in recent years is a certain ‘space race’ mentality from industry players.

“They are all largely fixated on innovation without always thinking too much about whether these innovations are necessary or deliver ROI,” he said. “In essence, AI innovators have been expanding the complexity of their simulations to fit the availability of computing resources, not necessarily to meet market demand.”

However, this is a trend he sees turning around in 2022. “The industry will get a bit more strategic, rethinking the size and focus of these models and simulations, rather than just expanding for the sake of expansion,” he said.

“This includes applying more scrutiny to the resources required to run these models. Supercomputers aren’t just expensive to deploy and maintain, they also consume enormous amounts of power. Rising energy prices across huge swathes of the world means the cost of training these models is going up all the time.”

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

Jenny Darmody is the editor of Silicon Republic

editorial@siliconrepublic.com