“AI is almost the master, the king, of these disruptive digital technologies...It is now having all the digital enablers coming to him or her, with data and capabilities to make use of
Richard Raj, head of digital and mobility at healthAlliance, sees a raft of opportunities in the immediate use of AI in the enterprise.
However, as he points out, it is the ongoing convergence of AI with related technologies such as IoT and robotics where exponential growth is happening.
“The convergence of AI with robotic process automation, real-time data feeds from IoT, would likely solve many problems and highlight new opportunities of business value we are not aware of.”
City councils, for instance, could soon be getting on-mass IoT data from things like manholes or city lights, drainage sensors, and CCTV.
“Very soon, if not already, they could be letting AI assess volumes of data and figure out new trends and even ask questions that have not been asked before, to provide citizens improved services at much lower costs,” says Raj.
“AI is almost the master, the king, of these disruptive digital technologies...It is now having all the digital enablers coming to him or her, with data and capabilities to make use of,” adds Raj, at a recent CIO roundtable discussion in Auckland on ‘building an AI-driven business’.
As per any technology, we need to understand the privacy and any other ethical issues around its use
The technology and digital leaders talked about their various stages of AI deployment, from ‘dabbling’ to deployment of key initiatives.
The conversations also touched on critical issues organisations need to contend with, such as ethics around the use of AI and ensuring access to skills that can work on these technologies.
Inside the networks
Bob Friday, CTO of Mist, which Juniper acquired early this year, looks at the benefits of using AI inside wired and wireless networks.
“Similar to self-driving cars, we are only just starting to see how AI and machine learning (ML) have delivered rapidly improving automation capabilities across network operations,” says Friday.
“Besides the seamless detection of faulty hardware, cables, and software, another significant improvement has been a far better mobile experience for employees and consumers alike,” he states.
Similar to self-driving cars, we are only just starting to see how AI and machine learning have delivered rapidly improving automation capabilities across network operations
According to Friday, “Real-time AI/ML network optimisation has created the ability to significantly raise the quality of services here, driven by the data provided by real-time, end-to-end visibility of individual internet connections.”
He says this is done through real-time AI/ML network optimisation, driven by the data provided by real-time, end-to-end visibility of individual internet connections.
John Pye, Director, Digital Strategy and Architecture at the University of Auckland, says that while AI has been gaining momentum over time in the research areas of the university, his team is now seeing opportunities for AI and predictive analytics in both learning and teaching and in administration.
“We have developed a student digital strategy,” says Pye. “A key part of this is to have a data informed understanding and to provide appropriate support – both personal and digital to a diverse student community.”
We see opportunities for AI and predictive analytics in both learning and teaching and in administration
He also shares that, “One area that is being explored is adaptive learning, which can provide a personalised digital experience that is appropriately paced. This is an area where AI has exciting potential.”
Vaughan Robertson, Group Manager - Technology Strategy at Beca, says the company has had “past dabbling” on AI and in different parts of the organisation.
One project was around using conversational digital agents to solicit feedback for planning purposes, and this provided them with a workable use case.
Beca has also had some success with using computer vision to evaluate quality levels of manufacturing production units. “I would categorise this as proof-of-concept - and is certainly close to [becoming] commercially viable.”
Richard Raj of healthAlliance, meanwhile, says enterprises can tap AI in predictive modelling as part of technology maintenance and services.
“This could involve using AI on upstream indicators of a potential incident that may have to do with infrastructure, an application. or any other service that the technology team provides,” he explains.
He says this can apply to healthAlliance, which provides services for many applications.
Very soon, if not already, city councils could be letting AI assess volumes of data and figure out new trends and even ask questions that have not been asked before
“Predictive analysis is not something that can be easily or economically done manually for evaluating all the upstream data and predicting incidents based on multiple causal factors,” he states.
According to Raj, there are tonnes of data that have to be assessed and AI can help look at upstream indicators of potential incidents and failures.
“Then it can activate self-recovery automation to avoid the incident altogether, hence mitigating the customer/clinical impact that would have been caused by that incident.”
He cites a theoretical example: An upstream indicator finds that the server, where coincidentally the information is being written, is about to get full because of multiple applications using the same service. At the same time, three of these applications are going into the peak time use, and the clean-up cycle is much later. The system could send an exception text to a technical person.
He further adds that a fully automated option to recover the server disk space issue could be automatically activated and the incident actually never happens because the application is being pointed to the new location.
The AI-enabled system also sends a report to the technical team to say this is an area they should look further into for improvements.
In these situations, “We are not reducing resources, we are making people do value adding exercise rather than doing mundane tasks,” explains Raj. “It makes people’s work more interesting and gives better value to our customers.”
Rather than going backwards and analysing security events, you can start thinking and predicting and actually stay ahead… We see AI turn the hunted around to be the hunter in cybersecurity
Gerhard Nagele, Service Director at Spark New Zealand, shares how AI has impacted the focus of their workforce.
He says one example is through the use of bots to automate a lot of manual processes that do not necessarily need human input.
By automating all those process you improve accuracy, he states. “You link one database to another database, it is machine-to-machine-to-machine and there is no potential human error.”
This, Nagele says, is very important especially with very complex processes that are run through in the IT industry.
“Naturally you would have to do a lot of data cleansing and making sure you have accuracy of data on the front end.”
“By doing so, you create huge cost savings but also get staff to do higher quality tasks,” he adds. “They take up new roles managing new technologies.”
AI and cybersecurity: From 'hunted' to 'hunter'
Nagele stresses security has to be built in any of these business processes, whether internal or external. “It just has to be table stakes.”
AI, meanwhile, has another impact on cybersecurity.
“What we used to do and still do in some areas of cybersecurity is we look back and analyse [trends], but by that time, the next [cybersecurity event] has already happened,” he explains.
“Whereas with AI, you start to think like the bad guy. We predict and by looking forward, we can set our new goals so that the bad guy doesn’t get through that front door anymore.”
“Rather than going backwards and analysing events, you can start thinking and predicting and actually stay ahead,” he states. “We see AI turn the hunted around to be the hunter.”
The discussions also veered towards the human component of AI, such as ensuring businesses have data professionals in their teams - whether internal or through partner organisations - who can also work across business functions.
Gerard Naish, head of ICT at Ports of Auckland, says setting up an information and analytics team was instrumental in enabling the organisation to undertake a raft of initiatives around AI.
We leveraged statistical machine learning algorithms to identify influential factors in the development of truck service peak hours.
One is around using established data platforms and models that adopted machine learning techniques to support port operations management initiatives.
“For example, we leveraged statistical machine learning algorithms to identify influential factors in the development of truck service peak hours,” he shares.
“We are now in the process of applying the findings to build machine learning models to predict short-term service demands and enable our gate service teams to proactively manage the demands. Completing this project may reduce waiting times for trucks and improve customer satisfaction.”
Naish says the Ports of Auckland also leveraged Azure AI services to explore the potential of using intelligent chatbots to help information/knowledge sharing.
For example, with the chatbot ‘Ask Amy’, a prototype, a user may use speech or enter text to tell the chatbot what information he or she is seeking.
According to Naish, these could be about IT projects, business metric definitions, analytics delivery and governance, analytical reports and dashboards.
“The chatbot can interpret and process natural languages and match the best answer to the question."
Supporting people through change
Roxanne Salton,Head of Digital Strategy and Delivery at Mercury, shares the energy company’s initiatives towards deploying emerging technologies such as AI.
“We firmly believe that each business should design a model that suits their needs and is fit for purpose. I don’t believe that one-size-fits-all in a space where technologies are still emerging.”
Mercury, she says, finds that a Centre of Excellence (CoE) is a great model to deliver the initial phase of intelligent automation.
Mercury has established an Intelligent Automation Centre of Excellence as an enterprise service delivering services like advanced analytics, robotic processing automation (RPA), digital conversations (chatbots), and future technologies like AI, IoT, and machine learning.
The secret to success is ensuring that we work with our people, have the right culture and governance, and being clear about the outcome.
“The CoE model works as it delivers scalability, speed, and consistency especially as these technologies are still emerging and have a degree of risk associated to them,” says Salton.
The centre also allows Mercury to maximise their investment through standardisation of approach, policies, talent management, data and knowledge management and governance around these technologies.
They also see improved performance through a single point of contact for technology platforms, investment, and reporting of value.
“There is, in turn, better control, orchestration and reduction of operational complexity,” she says.
“We have started our journey implementing a few automation platforms and we will be growing our capabilities over the coming year,” adds Salton.
“We are finding that the secret to success is ensuring that we work with our people, have the right culture and governance, and being clear about the outcome. Completing all of these, is a strong workforce management plan supporting people through the change.”
'AI as a service'
Elizabeth Coulter, head of Commercial Services at Noel Leeming, works with SMBs in her current role.
“SMBs seem to be focusing on technologies that will help them to deliver their core business, rather than the shiny new toy,” observes Coulter.
“Generally, they seem to be focussed on the basics such as applications that support their area of expertise, websites, e-commerce sites, and productivity suites. Some organisations are focussing on better BI and reporting, but not necessarily using AI to do this.”
She notes that SMBs are more likely to use AI where it is provided as part of their current SaaS products, where the benefit is demonstrated as part of a product they already use.
“New Zealand has a number of companies popping up providing AI for small business - or ‘AI as a service’,” she states. “It will be interesting to see how these businesses grow in this space.”
Citizen data scientists
So how can organisations build capabilities in working on AI and machine learning?
Bob Friday notes that while every organisation is different, enterprises looking to deploy AI solutions across networks will generally require a unique blend of technical skill sets across several roles.
Growing your own passionate, domain experienced, motivated ‘citizen data scientists’ seems like a far better proposition than competing in the market for the few candidates
“For starters, data scientists are needed to design and build the ML models, data engineers to build the data pipelines and databases, while domain expertise will be required to help understand the data and to do feature engineering for the models,” says Friday.
Friday says they can come from a combination of both internal resources, and partnerships with vendors.
Vaughan Robertson of Beca says the data science talent gap is a vital issue for organisations.
“Growing your own passionate, domain experienced, motivated ‘citizen data scientists’ seems like a far better proposition than competing in the market for the few candidates who appear to be only on a short stay to the next, more exciting opportunity,” he states.
He says that Beca has a growing internal “community of interest” in the data and analytics area. “They are populated by a wide variety of staff, including very young, clever engineers and software developers.”
The panellists share related concerns around the ethics of using AI.
“As per any technology, we need to understand the privacy and any other ethical issues around its use,” says Elizabeth Coulter of Noel Leeming.
Vaughan Robertson says early this year, Beca did a substantial amount of work on the ethics questions around AI and linked it to enterprise risk management.
“We aligned the development of AI based systems and processes with our risk management system and working through the range of stakeholders and functions impacted,” he explains.
“Subsequently we have reviewed the European Commission’s Ethics Guidelines for trustworthy AI and decided that the seven key principles enunciated in this framework is a good, practical, and effective model to follow rather than attempting to cover all the vagaries ourselves.”
Richard Raj of healthAlliance stresses discussions on AI should also factor in the risk of bias.
“A lot of our data is historic and has been captured with biased built in,” he points out.
“For instance, if the recruitment process is automated, and AI is watching and machine learning from historical data, it may conclude that it will ignore non-European or difficult names.”
“Thus, if you are basing AI on old historical data, there may be perceptions of human bias or undocumented rationale that we have to overcome,” he says.
Sign up for CIO newsletters for regular updates on CIO news, career tips, views and events. Follow CIO New Zealand on Twitter:@cio_nz
Join the CIO New Zealand group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.