An AI agent could be designed to enable a company to handle different types and sources of crisis. AI could be leveraged to manage complex and simultaneous tasks that require vigilance and precision. But as in my previous article, I can also see some drawbacks and ethical concerns.
Pre-AI Crisis Planning and Management
When I was more active as a communications consultant, crisis management planning and training was one of the services my partner and I would offer our clients.
One of our clients, a global logistics firm, initially engaged us to assist them during the launch of their IPO. Later on, they also enlisted our services to develop a crisis management program for their rapidly growing multi-billion dollar company.
The crisis management program that was devised for the global logistics firm had been modeled after an international airline company from the Middle East. It took us months to formulate and when it was finally done, it’s tangible form was that of several stacks of documents more than a meter high plus several rolls of charts with diagrams.
Of course, for presentation purposes before the board, that’s just what it looked like. The documents were organized into packets for designated crisis response groups corresponding to different crisis types. Each packet contained step by step instructions in flip books which, if executed properly, would enable the entire organization to manage the crisis.
Ideally, a crisis management program should translate into crisis preparedness. This is something that can only be achieved by regularly training and assessing everyone in the organization with respect to the type of crisis and their role in the crisis management program.
In reality, companies in certain situations relegate crisis management programs to some back room or storage area and forget that one ever exists until a crisis is imminent. In situations like that, it would be great if at least half a dozen people in the organization remembers the barebones of the program.
Looking back at my experience in helping craft a crisis management program, I could remember wishing for there to be someway of intelligently automating an organization’s crisis response.
For sure, there might have been some kind of technology capable of it but perhaps nothing quite as capable and as affordable as the kind of AI technology that we have now.
AI In Crisis Management
Advancements in AI are at a point where it can not only understand human language, but it also be equipped to directly interact with the real world.
So much so, it is possible to conceive of an AI that would be enabled to spot an emerging crisis and autonomously deploy an appropriate response to either abate it or manage the aftermath.
In the case of the logistics firm that we had created a crisis management program for, I imagine AI agents handling a number of crucial tasks in the event of a sudden natural disaster in several key ways:
Rapid Data Analysis and Insights. AI systems can quickly process large volumes of data from various sources like weather reports, news, and internal systems. This allows AI to identify emerging threats, detect anomalies, and provide real-time insights to crisis managers. AI-powered sentiment analysis can gauge public reaction and help craft effective communication[1][4].
Automating Repetitive Tasks. AI chatbots and virtual assistants can automate responses to frequently asked questions from customers, freeing up human responders to focus on critical tasks. AI can also automate resource allocation, team coordination, and other repetitive decision-making during a crisis[1][2].
Enhancing Communication. AI enables delivering consistent, targeted crisis updates across multiple channels like social media and chatbots. It can provide real-time information to employees, customers and the public, fostering transparency and trust. AI-powered translation can also help communicate with diverse stakeholders[1][5].
Predictive Capabilities. By analyzing past crisis data and current indicators, AI can predict the likelihood of certain threats occurring. This allows crisis managers to proactively mitigate risks and prepare more effective response plans. AI models can also simulate different crisis scenarios to test and refine strategies[1][4].
Continuous Learning and Improvement. After a crisis, AI can analyze the response to identify areas for improvement. Machine learning algorithms can detect patterns and recommend changes to strengthen future preparedness. This allows crisis management to evolve and adapt based on lessons learned[1][2].
However, an AI agent should be seen as a tool to augment human expertise rather than replace it entirely. Maintaining human oversight and judgment is crucial for making nuanced, ethical decisions during a crisis. Integrating AI into crisis management requires careful planning, robust data, and continuous refinement to realize its full potential[2][3].
Citations:
[1] https://rtslabs.com/disaster-preparedness-logistics-ai
[2] https://www.linkedin.com/pulse/ai-driven-decision-making-crisis-management-andreas-wadstr%C3%B6m
[3] https://sbi.sydney.edu.au/dial-ai-for-assistance-leveraging-ai-for-supply-chain-resilience-and-crisis-management/
[4] https://www.emergingtechbrew.com/stories/2024/05/20/technology-ai-supply-chain-crises
[5] https://www.crises-control.com/blogs/crisis-management-ai/
Current State of AI
A groundbreaking development in AI has been the ability of robots to learn through observing the actions of humans, rather than just through training by humans or machine learning. This allows robots to more easily adapt to real-world settings and perform physical tasks by watching how they are done. [2]
Agentic AI represents a shift from reactive to proactive AI systems. Agentic AI agents exhibit autonomy, can set their own goals, and take independent actions to achieve those goals without direct human intervention. This allows AI to be more proactive in areas like environmental monitoring or financial portfolio management. [3]
Combining agentic and multimodal AI, where AI systems can understand and generate content across different modalities like text, images, and speech, opens up new possibilities. This could enable “no-code” development of computer vision applications through natural language prompting. [3]
Generative AI Reality Check. As organizations move beyond the initial excitement around generative AI like ChatGPT, they are facing a “reality check” on the limitations of these systems, such as output quality, security/ethics concerns, and integration challenges. This is leading to a more tempered, realistic understanding of what AI can and cannot do. [3]
AI advances include breakthroughs in robot learning, the rise of autonomous and proactive “agentic” AI agents, the power of multimodal AI, and a maturing perspective on the real-world challenges of deploying generative AI at scale. These developments are shaping the trajectory of AI technology and its practical applications.
Citations:
[1] https://indatalabs.com/blog/ai-latest-developments
[2] https://www.koombea.com/blog/7-recent-ai-developments/
[3] https://www.techtarget.com/searchenterpriseai/tip/9-top-AI-and-machine-learning-trends
[4] https://ai100.stanford.edu/gathering-strength-gathering-storms-one-hundred-year-study-artificial-intelligence-ai100-2021-1/sq2
[5] https://www.simplilearn.com/tutorials/artificial-intelligence-tutorial/artificial-intelligence-applications
An AI Crisis Agent as Crisis Spokesperson?
With AI generated avatars now a reality, I can imagine crisis briefings being handled by an AI crisis agent designed to appear an assuring presence for those affected while also be capable of disarming incendiary ploys lobbed by detractors.
But I can also see the kind of backlash it could create when it comes across as inhuman and unrelatable. Especially in instances where the crisis could involve the loss of human life or the suffering of a large group of people.
AI agents can provide valuable support in crisis management, but it cannot replace empathy and nuanced decision-making.
As one source states, “While AI undoubtedly empowers those who use it, the value of human expertise and decision-making in crisis management is irreplaceable.” The “unique qualities that humans bring to the table include the ability to tailor crisis management messages to different stakeholders, deliver speeches with authenticity and empathy, and navigate complex situations that may arise during a crisis,” notes the EC-PR article[1].
Another source notes that “AI lacks the capacity for empathy and human connection. Ongoing conversations during a crisis will inevitably require human intervention to address concerns and, many might say, it morally *calls *for a human to provide reassurance. AI cannot replace the nuanced communication requirements that PR teams use in high-stakes situations,” according to the EC-PR article[1].
While an AI agent can automate responses to customer inquiries and gather data efficiently, “AI cannot know [the nuances of a crisis] (yet).” As one expert states, “PR teams are better positioned to assess the severity of a crisis and decide which tasks can be delegated to machines, simultaneously identifying those that necessitate human judgment and intervention,” according to EC-PR[1].
Using an AI agent as a “support tool rather than a decision-maker” can help inform responses, enhance outcomes, and protect brand reputation. However, “caution is advised when trying to fully automate crisis communications, as the human element remains essential.” As another source warns, “one insensitive response from a bot to a concerned customer could make a crisis worse, so I personally would advise caution when trying to use AI to scale crisis communications for companies,” states the EC-PR article[1].
In short, a balanced approach is recommended, where “AI augments and supports the work of human PR teams, but does not replace their critical role in crisis management,” according to EC-PR[1].
Citations:
[1] https://ec-pr.com/how-should-you-use-ai-in-crisis-management/
[2] https://www.prweek.com/article/1879124/ai-anticipate-crisis
[3] https://www.nixxis.com/ai-a-game-changer-in-crisis-management/
[4] https://bryghtpath.com/crisis-management-artificial-intelligence/
[5] https://spectrum.ieee.org/artificial-intelligence-quotes
