Part 1 of this series on AI looked at interactivity as an imperative in risk communication and touched on how AI-technologies could potentially help risk communication systems meet the demands of users.
AI-technologies will continue to flourish in many ways with advances in communication at the forefront. AI can assist risk communicators in meeting the audience requirements of interactivity, which are an essential part of the risk communication process. Advances in risk communication processes are as important as audience characterisation, message development and trust in information sources. Yet, innovation in this area has occurred slowly, the move to online channels has been embraced but represents the tip of the iceberg.
Interactivity is a key requirement of effective risk communication. Without it, communication becomes a sender-receiver affair, an inoculation. Whether this information meets the needs of the audience, particularly regarding their perceptions of risks or feelings, is another matter. Interactivity allows feedback, further questions, and may allow the audience to engage in a process that will lead to possible convergence of opinions, or at worst an exchange in a 2-way process that the user will feel was sufficient and fair. To have this interaction occur in real time has traditionally required human-to-human interaction. Furthermore, the human would require extensive training and experience to deal not only with factual information, but with higher levels of communication relating to experience and competence, and ultimately values, beliefs and worldviews. In essence, a highly trained risk ‘negotiator’ empathetic to audience needs and concerns and able to orchestrate a fair outcome and do so time and time again. In short, nearly an impossible task at scale, a team of such personnel (agents) would struggle to address daily audiences in their 100s, never mind their 1000s. Such a system would be too expensive and time consuming. With an entire range of risk issues to deal with, multiple banks of interactive risk councillors would face an impossible task.
Even on one issue, GMOs for example, the challenge has proven costly and difficult to sustain at a suitable level of user engagement. The initiative at GMOanswers.org debuted in 2013, could never be described as a truly interactive, synchronous or on-going process of risk communication on the topic of genetic modification. Questions could be posted by users, and they would be “answered” at a later time. If built today, such a site could utilise AI to provide pertinent and relevant information to users in real time, without the need to employ teams of experts or resort to static FAQs.
Interactive is a term that has been used as a commercial catchword, to promote vague technological enhancements and improvements. Used with precision, interactive is used to distinguish computer processes that respond to user input during execution (in real time). Genuine human-computer interaction has been an ambitious goal of AI for decades. Now we are seeing the fruits of this research as the increasing algorithmic, integrative communication capabilities of AI are being realised.
Machine learning, natural language processing and sentiment analysis are being strategically deployed in a range of settings to augment or even replace live ‘agents.’ As users become more accepting of AI technology in devices such as Apple’s Siri, Amazon’s Alexa, and Google’s Home Assistant, they are conditioned to demand a level of interaction in short, natural conversations with other AI-systems. This conditioning represents a perfect setting for extrapolation of existing applications in customer service centres to fully fledged behavioural intervention technologies (BITs) that provide customised, integrative support, education and interventions through brief conversations.
The full range of risk communication objectives—health promotion, reaching consensus on controversial technologies and critical incident interventions—would greatly benefit from AI-technology integration to augment campaigns through increased dissemination and accessibility, cost-effectiveness and increased engagement through real-time interactivity.
At Reciprocom, we are working on algorithms to enhance natural language processing to better interpret user requests, to understand patterns of risk perception relating to trust, benefit, risk perception and levels of knowledge. The range of applications in Asia is almost endless, from food safety, diet and weight management through to systems capable of meeting factual requests that would otherwise require live agents, and can take up to 85% of running costs.
In part 3 we look at how interactive and AI-supported systems can engage users in a risk narrative to allow them to fully explore, understand and gain enlightenment on a pathway to behaviour change.