Once I got past the gee whiz uses of ChatGPT, like “Write the auto-biography of the notorious EMS1 columnist Greg Friese,” or give me a recipe that uses “rice, black beans, a can of tomatoes and corn that isn’t beans and rice,” I began looking for use cases to assist me with my job or assist the EMS community we serve. For example, at the Pinnacle EMS Leadership conference, I used ChatGPT to summarize books speakers mentioned and to explain management concepts I wasn’t familiar with. The more I use ChatGPT, the more use cases I see for it in my work and professional activities.
The National EMS Management Association Google group is regularly used by its members to crowdsource opinions and share resources on EMS software, patient care protocols, quality improvement initiatives and operational documents and processes. For example, a member of the Google group recently asked, “Does anyone have panel interview questions they wouldn’t mind sharing?”
I don’t have any panel interview questions to share with the group, sorry, but I did ask ChatGPT to suggest questions.
Prompting generative AI
Prompts for ChatGPT or other generative AI, like Google Bard, are the instruction, question or beginning of a conversation. The prompt directs ChatGPT to generate a response or continue a threaded conversation. Prompts can be simple, like, “suggest five interview questions to ask paramedics,” or more complex. When I write a prompt, I lean towards a more complex set of instructions that might include outlining a persona or role for ChatGPT, describing the situation and asking for a specific form of output. Generally, the clearer I am on what I need from ChatGPT, the more helpful the response.
Here is the prompt I wrote for panel interview questions:
I am interviewing applicants for a full-time job as a paramedic at a busy, private ambulance service that responds to 911 calls and interfacility transport requests. One component of the process is a panel interview with the operations supervisor, a field training officer and an experienced paramedic. Recommend five scenario-based questions the panel should ask each applicant.
A better prompt might have included the length or complexity of the questions, the persona I wanted ChatGPT to assume, such as a human resource professional, and any additional criteria, such as, “these questions should be for a new graduate” or “a paramedic with three or more years of experience.”
ChatGPT’s suggested panel interview questions
I am still dumbstruck by how quickly the ChatGPT large language model generates a response. In a few seconds, ChatGPT recommended these questions with a short introduction:
Absolutely, scenario-based questions can be very insightful for understanding how an applicant might handle real-world situations on the job. Here are five such questions:
1. Major Accident Scenario: "You arrive on the scene of a multi-vehicle accident on a busy highway. There are several injured victims, some unconscious and others in visible distress. As the first paramedic on the scene, how would you prioritize care and delegate tasks to your partner and other first responders?"
2. Pediatric Emergency Scenario: "You respond to a call about a 5-year-old child who has fallen from a second-story window. The child is unconscious but breathing, and the parents are highly stressed and panicking. How would you handle this situation both medically and in terms of communicating with the parents?"
3. Difficult Patient Scenario: "You are dispatched to a nursing home for a patient who needs transport to the hospital for a non-emergency evaluation. The patient is resistant, refusing to leave the nursing home. How would you handle this situation?"
4. High-Stress Scenario: "You and your partner respond to a domestic violence call where a woman has been severely beaten. The assailant is still on the scene and acting aggressively. Police have been dispatched but have not arrived yet. How would you manage this situation?"
5. Ethical Dilemma Scenario: "During a mass casualty incident, you have to make quick triage decisions. Two patients are in critical condition: a 70-year-old man with a severe head injury and a 30-year-old woman with multiple fractures and internal bleeding. You have only one available spot in the helicopter for immediate evacuation to the trauma center. How do you decide who to transport?"
These questions can help assess the candidate's critical thinking, decision-making abilities, communication skills, and understanding of ethical issues in emergency medical services.
I have not edited the ChatGPT suggested questions. Also, I have not previously written panel interview questions for paramedics, and it has been many years since I was interviewed for a paramedic position, so I don’t know if these are good or bad questions. But they seem plausible and like a good starting point for updating existing panel interview questions or developing a new set of questions.
What do you think of these questions? Are they plausible? Do the likely responses give you, as an employer, helpful information to make a hiring decision?
ChatGPT can continue the work
ChatGPT becomes especially useful when moving from one question, one response, to a threaded conversation. For example, I gave ChatGPT this prompt after reviewing the panel interview questions:
Create a rubric to score applicants’ responses based on the criteria of critical thinking, clinical competence and communication skills.
Recall that the initial response suggested that the questions can be used to assess critical thinking and communication skills. Without any additional prompting, ChatGPT suggested a five-point rubric from excellent to poor, along with instructions on what each rating means. Here is the rubric:
Sure, I can certainly help with that. Here's a sample rubric:
Each applicant's response can be scored based on this rubric, with a maximum of 15 points per scenario-based question. The scores can then be totaled and compared to help determine the most suitable candidate for the position.
Again, this might not be the final panel interview rubric, but it creates a starting point.
Generative AI mindset
You might have heard the technology quip, “AI isn’t going to take your job, but someone who knows how to use AI is going to take your job.” It doesn’t take much to get started with generative AI, like ChatGPT. Create an account and start writing prompts. There are also lots of articles, videos and online courses on how to write prompts.
As you get familiar with the technology, apply a generative AI mindset to your work-related tasks and projects. I've used generative AI to write quiz questions, summarize meeting notes, develop a training plan for a new employee, write a news article, explain to me the medication Ozempic, and yes, suggest a quinoa burger recipe.
But, as you plunge into generative AI, make sure to know and follow your organization’s policy for confidential department or business information. Finally, never share protected health information (PHI) with ChatGPT or other generative AI tools.
What questions do you have about utilizing generative AI in your work as a paramedic or EMS manager? What are ways you’ve used generative AI?
Learn more
3 ways providers can use ChatGPT on the job
The artificial intelligence revolution has arrived; learn how OpenAI’s most popular innovation can add value to EMS
Bonus: Use ChatGPT for interview prep
As you explore ChatGPT, give it this prompt:
I am interviewing for a job as a paramedic at a busy, urban ambulance agency. I expect to be asked a question like, "You and your partner respond to a domestic violence call where a woman has been severely beaten. The assailant is still on the scene and acting aggressively. Police have been dispatched but have not arrived yet. How would you manage this situation?" during a panel interview. Please give me advice on how to answer this question that showcases my critical thinking and decision-making skills.
What did you think of the ChatGPT response? Would you hire a paramedic that gives you this response?
EMS1 is using generative AI to create some content that is edited and fact-checked by our editors.
0 Comments