4 AI trends for the future: Ubiquitous, affordable, customisable, trustworthy

Autonomous cleaning robots roam the malls and streets, robot servers deliver food in eateries, and robot librarians help with shelving books in the right places – these are just some examples of how AI has permeated daily life today.
 

And the prevalence of AI seems set to increase. Customer experience service provider Servion Global Solutions predicts nearly 95 per cent of customer interactions will be AI-powered by 2025. Investments in AI continue to rise as well – Stanford Institute for Human-Centred Artificial Intelligence’s AI Index Report 2022 found that private investment into AI totalled US$93.5 billion in 2021, more than double the amount of the previous year.
 

“We are seeing actual use cases of AI solving real problems,” says Sascha Giese, Head Geek™ at IT monitoring software provider SolarWinds®.

AI across the public sector

“AI can be a great advisor,” Giese says. “It can compute far more information in a fraction of the time a human would need and identify connections that we might not see.”
 

For example, higher education institutions use AI to personalise learning for students when they engage in virtual lessons. It can detect students struggling with a topic and automatically alert teachers, allowing educators to tailor their lessons accordingly. Financial institutions too can benefit, with AI-powered credit scoring helping to bring typically unbanked populations into the fold. Meanwhile, the healthcare sector is using AI to provide more patient-centric care.
 

Governments can also wield AI to great effect. Giese explains how SolarWinds integrates machine learning capabilities in its Database Performance Analyzer. The programme can help organisations automatically identify anomalies in the behaviour of databases by continuously monitoring performance in real time. This allows them to detect and resolve problems quickly, improving overall efficiency.
 

Beyond databases, AI can also help IT teams with their day-to-day tasks. In the SolarWinds® Observability Platform, AI is used to detect anomalies and implement solutions, Giese says. Once it identifies a problem, it can help to automatically execute simple fixes like restarting the programme. If the issue requires human intervention, the AI software will then flag it to the relevant IT teams, Giese explains.

Cheaper AI

One reason for the gradual increase in AI adoption is likely falling costs, Giese says. In 2017, training an image-classification system could cost over US$1,000, according to Stanford Institute’s report. In 2021, that number fell to a mere US$4.60 – a decline of over 60 per cent. With the cost of AI-powered robotic arms falling by nearly 50 per cent in the past five years, AI-powered hardware, too, is seeing a decrease in cost.
 

With AI becoming more affordable, doors are open for smaller businesses that would like to adopt such technology – a phenomenon governments worldwide are encouraging.
 

Singapore, for instance, launched a national AI programme named AI Singapore (AISG) in 2017. This programme was created to build an AI ecosystem in the nation by promoting AI innovation, research, and adoption across the industry.
 

One way this programme has been used is through the 100 Experiments initiative. In this programme, AISG receives problem statements from organisations keen to adopt AI solutions but are unsure of how they do so. AISG then works alongside researchers and engineers to develop a solution to these problems. As part of this programme, AISG also provides up to S$250,000 (US$177,930) per project to allay the costs.
 

Customisable AI

Most organisations today will operate one software programme or another in their daily work, whether used for customer relations management, resource management, leads handling in sales, or other functions. Many of these programmes come integrated with the ability to generate reports out of the box, Giese says.
 

But Giese points out problems often arise when organisations require customised solutions. The advancement of AI today may help address such issues with solutions such as generative AI.
 

Generative AI refers to programmes using existing text, audio or image data to create new content. Most recently, an AI-generated artwork took first place in the digital category at an art show as part of the annual Colorado State Fair.
 

Such technology can significantly benefit government agencies and businesses in a variety of ways. Data visualisation, for example, would be made simpler if users could input the type of data they would like processed into a textbox without having to source and manipulate the data points on their own.

Regulations to improve trust

Contrary to its name, AI is sometimes not intelligent at all. It has been accused of racism and sexism in the past, and, on one occasion, an AI chatbot even began spouting Nazi talking points. What’s left after these incidents are trust issues, as users start to doubt the efficacy and usability of AI tools, Giese says.
 

Legislation written to regulate AI is one way to improve the training of such tools, which will consequently increase trust in the technology. Giese highlighted efforts by the European Union to manage AI, for example. In 2021, the European Commission introduced a legal framework for AI to address the technology’s risks. This framework segregates AI systems into four risk levels, with different regulations applied.
 

AI technology classified as an ‘unacceptable risk’ will be banned altogether. This could include systems and applications that circumvent users’ free will, according to the European Commission. Meanwhile, AI classified as ‘limited risk’ will require specific transparency obligations, like ensuring users are aware that they are interacting with a machine.
 

The EU is not alone in its mission to create stricter frameworks to guide the use and adoption of AI. Across the world, countries like the UK, USA and China are also instituting their own laws. Stanford Institute’s AI Index Report also found AI-related laws have increased in countries around the world – from one in 2016 to eighteen in 2021.
 

As AI becomes more prevalent, affordable, advanced and regulated, its potential to improve the work of governments and organisations is great indeed.