Earlier last week, I saw several reports on the Fortune article, “MIT Report: 95% generative AI pilots at companies failing. I wanted to explore what the report uncovered to further understand how the overwhelming hype of AI in the media around what AI can do and what it will replace, could be so disconnected from the reality of what is happening in the daily operations of business.

I am a daily user of generative artificial intelligence (e.g., GenAI). I find myself turning to ChatGPT (my current GenAI product of choice) over the traditional ‘Google Search’ often, but not always. Some of this effort, is to be sure that I am moving along with the pace of the product’s (e.g., ChatGPT) development. While some of the effort is because I do find the responses to be a faster aggregation of available data and information on the internet than I could aggregate on my own from what we have been using over the last decade and more, commonly referred to as a ‘Google Search’.

In the traditional ‘Google Search’, we might put a word or phrase in the search bar and in return, a few sponsored ads would be at the top. Then, you would see a list of ranked items that may or may not be the most reputable, reliable and valid sources. As a user, you need to be able to decipher which are the reputable, reliable, and valid sources. You may read through a few links to get a sense of what the selected page says about the entry to determine where to search next. This could go on for as long as you would like to ‘dig in’ to the topic.

Now, with ChatGPT, and similar GenAI platforms, the large language model or “LLM”, performs that aggregation for the user and presents an answer. In the paid version, I often will have links associated with the returned response. I will not take the response seriously without a corresponding link that I will check for reliability and validity. However, this is my own personal principle, and could vary user by user.

I have found that ChatGPT can aggregate and present me with enough preliminary information that can take me to the next step in my knowledge development process to answer the question I have in mind. Whether it is to determine where to go for dinner or how many nurses there are in the world, I treat the responses as a preliminary start and not the end point.

Now, I have a PhD in nursing informatics where I studied the use of data, information, knowledge and wisdom in the context of nursing and healthcare delivery. I have worked in and studied health information technology for over 20 years now. So, my mind is automatically primed for this type of behavior. Yet not everyone will have these tendencies. Therefore, it is important to note there will be potential vulnerabilities where the data and information presented could misinform the user(s).

I share this because while AI is becoming increasingly more prevalent, it is a tool. Tools have specific purposes and use cases. We know what the stethoscope can and cannot do in helping to assess the human body. We know what calculators can and cannot do to help in preparing medications for administration. We know what the pulse ox probe can and cannot tell us to help in managing the respiratory system. The stethoscope, calculator, and pulse ox probe are technologies. They are technologies we have accepted into standards of practice. They have specific use cases and have reliability and validity to allow for standardization across care providers to offer data and information that can be used to generate knowledge about the patient’s condition.

We are in the early stages of discovering the use cases for generative artificial intelligence (GenAI) in healthcare. The report from MIT further demonstrates how the technology alone is not the transformative agent. Technology also requires the identified use case for the intended people and associated processes. This is no different than any other technology that has been adopted into healthcare. I say that to reduce the intimidation factor of the term and unknowns.

Nurses are experts in knowing how to give nursing care. They know what they need to do for their patients. The same is true for every other healthcare professional role group. Thus, this means that you have the opportunity to assist in determining the best use cases for GenAI in collaboration with others.

More to come on this topic as I review the MIT report in the next post….

Have a great week,

~ Dr. Kelley