AI, analytics drive 4IR in the public sector
South Africa's public sector is not yet prepared for the fourth industrial revolution (4IR). Both in form and function, the public sector is still in a 3IR world.
This is according to Pierre Schoonraad, chief director of research and development at the Centre for Public Service Innovation.
Schoonraad will be presenting on 'How are AI and analytics driving the fourth industrial revolution (4IR) in the public sector?’, at the ITWeb Business Intelligence Summit, to be held from 3 to 5 March, at The Forum in Bryanston.
There are several reasons for this, says Schoonraad.
Many officials don’t yet understand the nature and potential impact of the technologies associated with the 4IR, the new role of data and analytics, or the rate of change we’re experiencing.
Furthermore, most officials are not sufficiently skilled to exploit the benefits of the 4IR, and organisational structures still adhere to an outdated bureaucratic model, which means they are overvaluing outdated competencies and undervaluing new competencies such as data literacy and coding. In addition, the quality, availability and format of our own data remains a significant inhibitor.
Finally, the State IT Agency's (SITA) business model is still embedded in a 3IR world where connectivity is determined by email and legacy system use.
Analytics is leading
When looking at local reports and case studies, it’s tricky to distinguish between AI use, simple data collection, automation and use of frontier technologies. Locally, very little, if any AI, is currently being used, he adds.
Analytics in various forms is implemented across sectors, mainly for planning, budgeting, monitoring and evaluation. StatsSA, Treasury and the DPME (Planning, Monitoring and Evaluation) are leading in this regard.
Certain systems, such as the health referral app Vula, PDUs (pharmacy dispensing units), average speed prosecution, Shotspotter and drone-based detection, do use automation and basic algorithms, but don’t yet apply more sophisticated machine learning, adds Schoonraad. Adding machine learning to existing solutions, such as online school enrolment or booking systems, could prevent us from making the same mistakes year after year, he says.
“Globally, we see the growing use of AI and analytics. Besides military and intelligence applications, AI is used in education, health (such as for cancer detection and diagnostics), transport (like Seoul’s TOPIS system that can predict traffic behaviour and is used to schedule preventative maintenance), law enforcement (facial recognition and mass surveillance being the most notorious), in procurement and for disaster prevention and management.
By applying AI to simplify processes and procedures or completely replace outdated procedures, government can achieve quick wins, he adds. Many services that are linked to 'life events’ such as births, deaths, going to school, and suchlike can be automated.
There are significant benefits in the health sector, too, says Schoonraad.
“Some studies have shown that AI can detect breast cancer up to two years earlier than normal diagnostics. IBM’s Watson can provide doctors with additional treatment options. Community health workers can also be better equipped to correctly refer patients (the locally developed Vula app uses basic AI to classify vision-related and other problems). AI linked to IoT devices can make home-care possible for conditions that previously required hospital observation.”
He cites some examples of benefits that have been realised internationally. “Canada’s `bomb-in-a-box’ AI solution provides a risk-based oversight service for airfreight cargo. Finland’s AuroraIA is in the process of providing access to seamless, integrated 24/7 services while anticipating the future service needs of citizens.”
In addition, natural language processing and speech recognition hold significant positive promise for improving access and creating better service experiences for marginalised communities and the disabled. “Bots in various forms can play a significant role in this regard,” he says.
Internationally, there have also been significant gains in the use of AI for policing and law enforcement.
“However, this is raising a number of ethical questions about the kind of democracy and level of privacy we value. Government’s role in championing the auditing of algorithms and limit unintentional bias due to limited training data or poor practices, will become more important."
During his presentation, Schoonraad will engage the audience on the enabling environment for AI and analytics in government and point to use cases that South Africa should explore. He will also touch on ethics and the role of government in regulating its own and private-sector AI.