The recent launch of new generative artificial intelligence (AI) tools, such as ChatGPT has spurred debate among consumers about the role of AI in our lives and AI in the patient experience. While it may seem that AI is a relatively new innovation, it has played a role in healthcare for over 50 years, dating back to when Stanford researchers developed an AI program to treat blood infections.
AI gains traction in the public discourse and new opportunities for using AI in healthcare open up, it’s an important time for the healthcare industry to focus on the patient experience by taking another look at how it uses AI and virtual assistants, what consumer perceptions are of the technology, and what steps should be taken to inform and educate patients.
On the heels of our recent report on consumer perceptions of AI, we used the DISQO CX Platform to take a deeper dive into patients’ attitudes toward AI-driven healthcare and gain insights into how providers can effectively navigate consumer concerns to enhance the overall patient experience.
When asked about their knowledge of AI in healthcare, 22% of respondents admitted that they did not have any knowledge, and 37% said they only had limited knowledge, indicating that public awareness and engagement are largely lacking.
One bright spot in the results revealed that 64% of respondents who reported having experienced AI-driven healthcare (12%) said that experience was at least “somewhat positive.”
When asked about AI’s reliability in terms of performing different tasks, 60% of respondents reported that they believed AI is extremely or very reliable when used to schedule appointments. However, this number dropped to 38% for AI used in performing administrative work (e.g., writing chart notes, managing cases), 30% for AI used in treatment (e.g., providing personalized treatment plans, robot-assisted surgery), and only 27% for AI used in diagnosis (e.g., reading X-rays, forecasting kidney disease).
Similarly, although 58% of respondents reported that they were either extremely or very comfortable with AI scheduling appointments, only 39% shared that comfortability when it comes to AI performing administrative tasks. Furthermore, almost half of respondents were slightly or not at all comfortable with AI being used in treatment (45%) and making diagnoses (49%). These results suggest that much remains to be done by medical professionals and AI technology providers to gain the public’s trust in AI-driven healthcare.
Echoing the skepticism about AI’s reliability, 60% of respondents reported that it is extremely or very important for them to be notified when AI is used in their diagnoses and treatment. On the other hand, only 35% of respondents felt that way about AI scheduling appointments, indicating greater trust in AI performing simpler tasks that don’t require medical expertise.
Today, younger patients are more likely to say that vendors should be held responsible for mistakes made in healthcare by AI. As this population matures along with AI, and awareness of the technology’s use increases, so too will the burden of accountability rise for AI.
Although it is very common for patients to Google their symptoms, 69% of all respondents selected either “slightly likely” or “not at all likely” when asked “How likely or not are you to ask AI (e.g., ChatGPT, Google Bard) for medical advice?”
When asked about their gender preference for a healthcare chatbot or AI virtual assistant, 52% of respondents had no preference. Intriguingly, among those with a gender preference, male and female respondents both favored chatbots or AI virtual assistants of their own genders.
When asked “What are the biggest benefits you see of AI in healthcare?”, many respondents cited diagnostic values, lower costs, and efficiency. In general, they viewed AI as a tool to free up medical professionals to perform more important tasks, but they still wanted human supervision.
“I think many nurses and doctors are overworked, and likely many of the mistakes that they make is due to being overworked, so if AI were to relieve their workload by performing some of the simpler/less sensitive tasks, it could help them a lot. I can also see AI handling data and maybe recognizing patterns that could help with a patient's treatment.”
-Female, age 21, $125,000 to $149,999 household income
“AI could help confirm or narrow down a diagnosis, but the actual diagnosis should come from the expertise of the doctor. It could also take notes and keep records organized, but records would still need to be reviewed and proofread because there are still flaws.”
- Female, age 42, $40,000 to $44,999 household income
Respondents also shared their concerns about AI-driven healthcare, particularly when it comes to its accuracy, security, and lack of personal touch. Many respondents were worried about AI making mistakes and medical professionals being overly dependent on AI.
“[I have] lots of concerns around mistakes and misinformation, including the question of who’s responsible and is someone overseeing what AI is doing or telling people about their health.”
Female, age 31, $90,000 to $94,999 household income
Another concern shared by many is that AI-driven healthcare will dehumanize the patient experience and the medical field.
“AI can be programmed to accomplish a lot in the healthcare arena, but the one thing it is not (yet) capable of, is a personal, human instinct with regard to each patient. For example, a hernia can be diagnosed by a bot as long as the information is transferred correctly; only another human can instinctively react to a patient's distress, sadness, or anxiety.”
Female, age 74, $65,000 to $69,999 household income
Overall, 45% of respondents said that they thought AI will make healthcare better and that percentage went up to 57% for people aged 25 - 34. Twenty-seven percent believe that AI won’t change anything in healthcare, and about the same amount think it will make it worse, showing that AI technology and healthcare providers need to partner on demonstrating the value AI brings to patients.
This study provides valuable insights into patients’ perspectives on AI-driven healthcare. In general, researchers found that there is a lack of public awareness of the use of AI in healthcare. Although patients are comfortable with AI scheduling appointments and performing some simple administrative work, they do not currently find AI technologies reliable when it comes to diagnoses and treatment, and they desire to be notified of its involvement.
Many patients see the potential of using AI to liberate medical professionals from paperwork and lower medical expenses. However, it is generally believed that human supervision is needed to prevent errors, and lots of work remains to be done to improve the accuracy of AI technologies. Many argue that as human touch, compassion, and instincts are crucial components of healthcare, they are cannot be replaced by machines.
Despite these concerns, the public is relatively optimistic about the impacts of AI on future healthcare. To promote the adoption of AI-driven healthcare, medical professionals, healthcare groups, and AI vendors must make further efforts to mitigate possible harms, increase transparency and accountability, and engage the public in dialogue about ethical innovation.
DISQO used the company’s CX platform to survey 1,896 respondents (ages 18 to 75) who had visited a doctor within the last year on March 6, 2023.
Learn how you can use DISQO's CX platform to test and learn about your patients' experiences.