I am sure you have heard about ChatGPT. It was developed by OpenAI, a company now closely aligned with Microsoft. GPT is the acronym for a generative pre-trained transformer (GPT) – an architecture specifically designed to generate human-like responses in natural language conversations. Having studied artificial intelligence (AI) in my Murray State University courses and following its development, I was not surprised to see ChatGPT released. I must admit that I like using it and encourage you to play around with it. https://chat.openai.com/
ChatGPT, like other automated language models, uses unsupervised learning, which means it can predict the following sentence and the word given the previous context. This process helps develop a rich understanding of grammar context and general knowledge. Although ChatGPT and similar models are being used now in health records and may be coming to an EHR near you, eClinical Works and Epic have already included aspects of AI in their software. You can play around with AI clinical notes and create notes for your clinical services by going to autonotes.ai. AutoNotes | SOAP & DAP Notes in Less Than 60 Seconds and a clinician helped to develop it. There is a free version and a paid version. This software does not incorporate a business associate agreement, and it is not to be considered HIPAA compliant, unlike eClinical Works and Epic software. Also, this is not an endorsement of autonotes.ai.
Alan Turing and William McCulloch made some of the first AI learning models in World War II, as is depicted in the movie, The Imitation Game (which you should watch if you haven’t seen it). Plus, many of us have things like Roomba robots cleaning our houses, which is part of the evolution of these models, and some of us remember how well IBM’s Watson did on Jeopardy. So while Watson has primarily been a disappointment and is no longer promoted by IBM for healthcare, it was still an early demonstration and a building block for where we are today. The release of ChatGPT is a watershed moment and maybe similar to the release of the iPhone.
AI-powered chatbots and virtual assistants already exist, e.g., Woebot and Calm, and can enhance accessibility and support individuals in need. These tools offer information resources and guidance for people seeking assistance even outside the traditional services that social workers provide. Chatbots use natural language processing to understand the response to inquiries of users and appear to offer personalized recommendations and referrals for appropriate services.
Other EHR Assistant Examples:
- Generate responses to messages.
- Translate messages to a new language (using one?)
- Make diagnostic recommendations and treatment plans.
- Access info from user guides/settings/preferences with everyday language
- Use normal language to create patient schedules.
- Requesting reports generated from a phrase.
- Transform patient summary to narrative text from forms (creating a biopsychosocial?)
- Translate patient summary to patient’s native language (100+ languages)
- Natural language charge capture on mobile
- Combining AI with search and data from multiple sources,
- Voice recognition (already using?)
- Summarize the patient’s entire health record in a readable format.
- Take a picture of an insurance card and verify coverage.
- Image identification on Faxes to auto-identify the patient.
- Image identification to detect types of documents.
- Find a patient’s appointment using a picture of the patient (from a phone call?)
- Auto-generation of notes from telehealth or in-person visits
- Self-guided feedback using tone analysis.
- Real-time closed caption on telehealth with translation, including voice for native language.
- And the list goes on……
Anything that helps us do our work better and serve our clients is definitely worth examining. However, there are still questions about using AI in behavioral health and social work practice. For example, a fundamental ethical question may be what use of AI we disclose to our clients. How much of our work will then be questioned by payers and auditors? Did the software developer include social workers in developing their AI product?
Ethical considerations are crucial when implementing AI and social work. It’s essential to ensure that AI systems are transparent, fair, and unbiased. A mechanism like ChatGPT is a powerful tool, and it’s important to note that AI systems are based on the training it receives and the available data. It may only sometimes provide the most up-to-date and accurate information. Therefore, its responses must be evaluated critically so it does not perpetuate discrimination and inequalities already existing in our society.