Don’t let ChatGPT distract from the value of generative AI in healthcare

ChatGPT applications in healthcare are few and far between, addressing transactional activities that solve annoyances rather than anything meaningful or strategic (see Exhibits 1 and 2). Health plans and healthcare providers (HCPs) must therefore exercise caution applying ChatGPT. They should use the energy of the moment to refine and accelerate their roadmaps to build new generation products and services based on generative AI (GAI), which is broader in its capabilities. Spending calories on ChatGPT will be wasted as the technology matures, newer modalities become real, and adoption increases. Instead, enterprises should start piloting features to enable users to use GAI with enterprise governance.

ChatGPT, a natural language processing application developed by OpenAI and powered by generative large language models (LLMs), was launched in November 2022. It created a significant buzz, forcing Big Tech to accelerate AI plans and excited consumers to use it for interesting purposes, including generating code, writing book reports, penning love letters, and authoring legal responses. It even passed the US medical licensing exam, but it remains inconsistent, unreliable, and tends to hallucinate. That is because ChatGPT is a statistical tool used to predict language, without understanding it, through mapping the probability with which words follow other words.

The possibilities of ChatGPT in healthcare across the value chain are limited, but enterprise-grade and governed GAI can drive real transformation

The healthcare value chain has been broken. It was developed to address 20^th-century challenges and has shown it is grossly inadequate to meet 21^st-century opportunities. So, the industry has chosen to throw emerging technologies at the value chain instead of recrafting them from the bottom up. ChatGPT is the latest emerging tech to go after the value chain. While it is unlikely to affect it in any meaningful way, Exhibit 2 shows a few tactical use cases worth exploring in the near term.

Exhibit 1: ChatGPT has some tactical use cases that will help with efficiencies, but GAI has the real potential

Source: HFS Research, 2023

The broader GAI use cases are relatively sustainable compared to ChatGPT, but their true power will remain constrained unless the value chain is reengineered. Short of that, it is likely the use cases will limit the need for manual interventions, improve the quality of underwriting, and align benefits to the real needs of consumers.

The HCP value chain, however, continues to be relatively contemporary in its ability to address 21^st-century priorities. HCPs are generally technology laggards, either using it to improve efficiencies or improve patient experience and engagement. Exhibit 2 highlights a few tactical ChatGPT use cases that enhance patient interactions (communications, medical records) while also driving some automation to HCP and health plan communications concerning appeals (utilization reviews, claims) that can reduce the administrative burden.

Exhibit 2: Health systems and hospitals appear to be able to leverage ChatGPT tactically, but the real difference is with GAI

Source: HFS Research, 2023

In comparison, GAI use cases are transformational and sustainable. They can enhance surgical planning and execution and ensure the right type and level of care, and they will likely move most administrative tasks away from manual interventions.

GAI can be a significant innovative disruptor, and some innovators are attempting to disrupt

Healthcare is a target-rich environment. For entrepreneurs willing to leverage GAI, it is likely going to be a generational bonanza. Yet we need to temper the expectations that the industry will embrace GAI solutions, given the entrenched mindset of enterprises driven by a culture of risk averseness. Still, there are signs of users exploring creative use cases with ChatGPT, but they reflect the beginnings of a movement for the broader GAI to impact healthcare. In Exhibit 3, we highlight six players using ChatGPT in interesting ways to connect with healthcare stakeholders.

Exhibit 3: The world of GAI entrepreneurs is growing rapidly as use cases expand, as this sample of six companies suggests

Source: HFS Research, 2023

The efficacy of the GAI solutions or their adoption remains to be seen. If the past trajectory of other emerging technologies is any indication, we will likely see many failures before we see success. However, success will be sustainable, disrupt legacy, and make way for better healthcare to be delivered and consumed.

Expand possibilities through GAI; don’t be constrained by the ChatGPT hype

Technologies can be fickle; when the initial hype dies down as industry applicability dilutes, they join the cemetery of the obsolete like the rotary phone or cassette tape. However, the underlying driver (for rotary phones, it was communication, and for cassette tapes, it was storage) is powerful and will sustain and drive innovation.

GAI is beginning to power the next-generation AI algorithms across multiple modalities (see Exhibit 4), and it will drive a variety of sustainable industry applications.

Exhibit 4: The generative AI landscape continues to evolve as AI models get smarter, yet there is a bias toward addressing a modality vs. the entire spectrum of delivery channels

Source: HFS Research, 2023

The energy behind the technology is often seen in the investments it attracts, and that is true for GAI. In 2020, according to CBInsights, a quarter of the $220 billion private equity funding of GAI went to healthcare and life sciences. A critical indicator of the possibilities of GAI in healthcare is beyond what is fast becoming the basic chat functions of ChatGPT.

The Bottom Line: Leverage the excitement of ChatGPT. Craft a practical roadmap of embedding GAI to drive disruptive innovation that meets the needs of the next generation of health consumers.

The excitement of ChatGPT will die down, driven by the improving sophistication of GAI and the lack of sufficient use cases. While that is not news, enterprises must continue to refine their GAI plans, invest wisely, and reengineer their processes to improve adoption as they drive to better outcomes across the triple aim of care—reducing the cost of care, improving health outcomes, and enhancing the experience of care.