In a report released yesterday, the National Association of State Chief Information Officers (NASCIO) weighed in on the implications of artificial intelligence (AI) technologies for state CIOs.
“NASCIO will leave the job of predicting the future of AI (and whether it’s a threat to mankind as we know it, or the key to a utopian society) to other groups,” the report explained. “Instead we will look at the current applications of AI in government as well as some possibilities. As government IT leaders and employees look to AI to change the work they do, let’s look at how states can and are using this kind of technology.”
NASCIO turned to its annual state CIO survey to demonstrate the growing prominence of AI.
In just one year, the number of survey respondents who listed AI as the emerging technology that will have the greatest impact in the next 3-5 years nearly doubled. In addition to being viewed as transformative, government employees also view AI, and other emerging technologies, as a positive, not a threat. “The majority of public sector employees seem to hold the view that technology will only improve their work, not replace it,” the report explains.
The report explains that automation through AI can help the government in four different ways. Citing a 2017 Deloitte report, NASCIO said that AI can help relieve, split up, replace, and augment. To put it simply, investing in AI technologies enables work to be shared among people and computers, it frees up a worker’s time to focus on more complex tasks, and it enables governments to clear backlogs and provide better citizen services.
Government uses of AI go well beyond personal assistants like Amazon’s Alexa and chatbots to answer simple customer service questions, though NASCIO recognizes that both technologies could help state governments improve customer service and lower costs. AI can also help tackle one of the larger technology challenges facing the public sector – the dearth of cybersecurity experts. State governments struggle to hire qualified cybersecurity professionals, both because there aren’t enough trained professionals to meet demand and because the talent pool is lured into the private by higher salaries. However, AI offers a unique opportunity to augment a state’s existing cybersecurity staff.
“Because state CISOs (Chief Information Security Officers) often struggle to find qualified security staff at the state level, they may find that turning to artificial intelligence and machine learning is one solution while applying advanced cyber analytics,” the report explained. “As the amount of stored data grows, and available qualified workforce fails to keep up with demand, AI security software can detect threat patterns and allow the professionals to deal with most needed human-led tasks. Software that includes machine learning may be able to automatically respond to the continued sophistication of attacks and learn new patterns.”
When implementing AI, state governments should not rush to use the new technology simply for the sake of using an emerging technology. Rather, implementation must be thoughtful and well planned.
“In planning for AI, an organization must define a problem (not just look for one so that they can use AI) and express how the new systems will create efficiencies and innovations,” the report said. “Then the organization must show the data and opportunities that make AI a viable solution. Finally, key performance indicators must be identified so that the systems and the impact on the organization can be properly evaluated.”
NASCIO also identified three levels of autonomy a new AI solution should go through during deployment. Initially, humans must regularly monitor the new solution, ensuring it is performing properly. Next, the system can run on autopilot. Humans should still work with the solution, but don’t need to monitor it quite as frequently. Finally, the AI solution should be run autonomously. Even in the last stage, the AI system should still interact with humans and other systems. Even in the final stage, NASCIO stresses the importance of having a human monitor the system occasionally and step in when necessary.
When it comes to implementing AI, state CIOs need to be involved in every step of the process. CIOs looking to modernize should be ready to discuss disruption to the workforce, as well as security and privacy concerns and requirements.
Additionally, CIOs should insist on a roadmap for AI implementation. “As with many projects in the CIOs office, developing a roadmap can mean the difference between an ad-hoc approach to AI deployment full of unexpected problems, and a well-designed project,” the report said.