| Definitions |
|
 |
|
ChatGPT is becoming increasingly popular among individuals looking for assistance with day-to-day business tasks such as writing emails, creating reports and developing other communications. It leverages artificial intelligence to learn from previous interactions and sample texts to produce relevant output. This makes it easier to generate content without spending much time and effort.
ChatGPT can likewise create conversational and interactive chatbot experiences. By using ChatGPT, individuals can create chatbots with an accessible conversation flow that understand natural language queries. This is done by using a combination of OpenAI’s GPT-3 language model and domain-specific training.
Companies are leveraging ChatGPT to automate customer service interactions as well. Developers can create chatbots that can understand inquiries and provide accurate responses. This reduces the need for customer service representatives, a cost savings for business.
The development of ChatGPT is, in a way, like the development of the web browser. While the internet existed well before the web browser, that latter made the internet available to a broader audience. Similarly, ChatGPT is making AI more accessible by providing a simple and easy-to-use conversational interface.
GPT can help organizations across healthcare verticals achieve the goals of the Quadruple Aim: enhancing patient/consumer experience, expanding population health, improving the work life of healthcare providers (HCP) and reducing the cost of care. Although some applications are a long way off, there are numerous scenarios in which GPT can have a positive impact on healthcare today, providing immediate benefits.
Providers and care teams:
- Improve patient health literacy: GPT provides patients with access to a wealth of health information available on the internet, but in a much more digested and convenient way. Patients will be able to ask questions and get answers that summarize the best information available. These AI-enabled tools will help patients become more informed, leading to overall improved health outcomes.
- Improve patient engagement: GPT makes it easier to access and interact with automated patient communication systems. Today’s conversational AI platforms can leverage GPT to provide a tailored experience for patients by assisting with appointment reminders, providing helpful resources and streamlining communication with providers.
- Reduce staff burden: GPT automates mundane tasks in the call center and streamlines processes across departments, from customer service to billing, to ensure staff workloads are balanced and organized.
Pharmaceutical companies:
- Provide convenient education to HCPs and patients. GPT auto-generates virtual assistants that use existing resources such as prescribing information (PI) to deliver on-demand training, education and support to providers and patients 24/7.
- Extend capacity of internal teams. These new AI models extend the capacity of Medical Affairs by making PI documents intuitively searchable. Prescribers can quickly find information – even in lengthy documents – about common questions, contextualized for their specific needs.
- Alleviate call-center volume. GPT supports conversational, natural language dialogs to answer common patient questions – ensuring these answers will be smarter and more precise so fewer calls will be escalated to live agents
Despite its advantages, GPT exhibits weaknesses provider and pharmaceutical organizations need to consider.
- Clinical diagnosis – Among the primary concerns centers around patients viewing GPT as “the new doctor in the house.” Will patients turn their attention from “Dr. Google” to “Dr. GPT?” Like it or not, this is likely going to happen. Even putting aside the fact that ChatGPT isn’t a clinical decision-making tool, there are other challenges. Top of the list are its significant gaps in knowledge, as the technology has been trained up to September 2021, rendering it unaware of advances or changes beyond that point. Ask about the guidelines for colorectal screening, for instance, and you’ll get an inaccurate response, since that recommendation was recently updated.
- Oversight: ChatGPT responds to prompts with computer-generated text. This means there is little-to-no human oversight of the generated responses, and mistakes or inaccuracies are likely to go unnoticed. Without any human intervention, the accuracy and quality of the generated text cannot be assured.
- Potential data bias: Data bias can be introduced in many ways. For example, it can be caused when data is collected (e.g., taking data from a limited cohort, which may not represent the wider population), processed (e.g., weighting certain attributes greater than others) or used to make decisions (e.g., algorithms with a bias towards a certain outcome possibly favoring a particular population). Data bias can have significant implications particularly for healthcare.
- Security: Specifically, Chat-GPT and OpenAI-enabled tools are not compliant with the Health Insurance Portability and Accountability Act (HIPAA), which means Business Associates Agreements (BAA) cannot be signed. This will be an interesting area to watch, however, with Microsoft’s recent investment in OpenAI and integration into Microsoft Azure cloud computing services, and its support for enterprise, healthcare and other use cases.
With the technology still in its preliminary stages, it alone is not yet ready in healthcare. There are several initiatives currently underway, however, to make OpenAI more accessible, such as the OpenAI Healthcare Initiative.
| How can Orbita address these challenges today |
|
 |
|
As the Chinese curse goes, “May you live in interesting times!” The future of GPT-3 and related technologies looks incredibly promising. We can expect to see greater adoption of ChatGPT and the emergence of entirely new businesses.
- Tier 1 companies – Expect to see innovation this year from the big tech players – OpenAI, Google, Amazon, Microsoft – as they continue to build and buy large language models. For example, Google just announced a $300M investment in an OpenAI competitor, Anthropic.
- Tier 2 companies – Expect an explosion in new businesses and start-ups, as well as products and strategies from existing companies, particularly with regards to digital experience management and middleware providers. It’s likely they will adopt GPT technologies to build new solutions leveraging new large language models
- Tier 3 companies – Expect to see adoption of GPT-type technologies over the coming year by businesses and organizations leveraging new products and services to realize efficiency gains, increase revenues and cost savings.
GPT technologies can revolutionize the healthcare sector by providing AI-enabled tools to improve health literacy, patient engagement and more. Granted, there are potential drawbacks. They can be mitigated through appropriate oversight, accurate data and compliance with HIPAA regulations – specifically by leveraging Tier 2 companies that provide the benefits of GPT technologies that address the risks.