The 2023 American Society of Clinical Oncology (ASCO) Annual Meeting highlighted a growing wave of artificial intelligence innovation in cancer care, with ChatGPT-powered chatbot platforms in oncology drawing considerable attention from clinicians, researchers, and patient advocates alike. These AI-driven tools represent a meaningful shift in how information, communication, and clinical support may be delivered across the oncology landscape.
Key Takeaways
- ASCO 2023 featured notable presentations on ChatGPT-based chatbot platforms designed for oncology settings.
- AI language models demonstrated potential in patient education, symptom triage, and clinical decision support.
- ChatGPT tools showed promise in improving access to cancer-related information for patients.
- Researchers identified significant limitations, including hallucination risks and lack of real-time data integration.
- Clinical oversight remains essential before large-scale deployment of AI chatbots in cancer care.
ChatGPT-Powered Chatbot Platform Presented at ASCO 2023
At ASCO 2023, several research teams introduced prototype and operational chatbot systems built on large language models (LLMs), most notably OpenAI’s ChatGPT. These platforms were designed to interact with cancer patients in natural language, offering responses to questions about diagnoses, treatment options, side effect management, and supportive care resources. The platforms aimed to reduce the burden on oncology teams by handling routine informational queries at scale.
One prominent demonstration involved a chatbot integrated into an oncology clinic’s patient portal, allowing patients to submit questions between appointments and receive immediate, contextualized responses. The system was trained on curated oncology literature and institutional guidelines to help reduce the risk of providing generic or misleading information. Presenters noted that initial pilot data showed high patient engagement rates, with many participants rating the chatbot’s responses as clear and helpful.
Importantly, the AI chatbot tools presented at ASCO 2023 Annual Meeting were not positioned as autonomous decision-making systems. Instead, they were framed as augmentation tools — designed to complement, rather than replace, the expertise of oncologists and multidisciplinary care teams. Each system included escalation pathways to ensure that complex or urgent clinical concerns were flagged and routed to human clinicians promptly.
How AI Chatbots Are Being Applied in Oncology Clinical Settings
AI chatbots are being applied in oncology clinical settings across several functional areas, ranging from patient-facing support to administrative and research applications. In patient communication, these tools help individuals understand their diagnosis, prepare questions for appointments, and navigate the often complex landscape of cancer treatment. This is especially valuable for patients in rural or underserved areas where immediate access to specialist guidance may be limited.
Beyond patient support, clinicians have explored chatbot applications for literature synthesis, clinical trial matching, and documentation assistance. For instance, an oncologist might use an LLM-based interface to rapidly summarize recent publications on a specific cancer subtype or identify trials for which a patient may be eligible. The ChatGPT use in oncology clinical settings 2023 extended to generating structured clinical notes and drafting patient-facing summaries of complex pathology reports, tasks that are time-intensive when performed manually.
Oncology teams also experimented with chatbots as symptom-monitoring tools during active treatment. Patients undergoing chemotherapy or immunotherapy could report symptoms via a chat interface, and the system would assess severity based on standardized criteria, offering guidance on whether to seek immediate care or manage symptoms at home. This approach aligns with broader trends in remote patient monitoring, which has accelerated significantly since the COVID-19 pandemic demonstrated the feasibility of digital-first cancer care models.
ASCO 2023 Findings: ChatGPT Tools for Cancer Patients and Research
Research presented at ASCO 2023 examined the accuracy and utility of ASCO 2023 AI language model tools for cancer patients across several key domains. Studies evaluated whether ChatGPT could correctly answer questions about cancer staging, chemotherapy regimens, immunotherapy eligibility, and survivorship care. Results were mixed but informative: the model demonstrated strong general knowledge of common cancers and well-established treatment protocols, but showed inconsistencies when queried about rare malignancies or newly approved therapies.
One study assessed ChatGPT’s responses to 100 oncology-related patient questions and compared them to answers provided by board-certified oncologists. The AI performed competently on approximately 60–70% of questions, producing responses that reviewers deemed accurate and appropriately nuanced. However, responses to questions involving dosing specifics, drug interactions, or emerging biomarker-guided therapies were flagged as requiring significant clinical correction. These findings reinforced the importance of positioning AI as a supplementary resource rather than a standalone authority.
On the research side, the ASCO 2023 artificial intelligence chatbot cancer research presentations explored how LLMs could accelerate literature reviews and hypothesis generation. Scientists noted that ChatGPT could synthesize broad bodies of evidence in minutes — a task that might otherwise take researchers days — though the outputs required careful expert review to confirm accuracy and relevance. Several investigators proposed hybrid workflows in which AI-generated summaries serve as starting points for deeper, human-led analysis.
| Application Area | Demonstrated Capability | Noted Limitation |
|---|---|---|
| Patient Education | Clear explanations of common diagnoses and treatment pathways | Inconsistency with rare cancers or new approvals |
| Symptom Monitoring | Standardized triage support during active treatment | Risk of under-triaging atypical symptom presentations |
| Clinical Documentation | Draft generation for notes and patient summaries | Requires clinician review before use in medical records |
| Research Synthesis | Rapid literature summarization and hypothesis support | Potential for hallucinated citations or outdated references |
| Clinical Trial Matching | Preliminary eligibility screening based on patient profiles | Limited integration with real-time trial databases |
Limitations and Considerations of AI Language Models in Cancer Care
Despite the enthusiasm generated at ASCO 2023, experts emphasized that AI language models carry meaningful limitations that must be carefully managed before widespread clinical adoption. Chief among these is the phenomenon known as “hallucination,” in which LLMs generate plausible-sounding but factually incorrect information. In oncology — a field where clinical decisions directly affect patient survival — inaccurate outputs can have serious consequences. Researchers stressed that robust validation frameworks must be established before any AI chatbot is deployed in a patient-facing capacity without active supervision.
Another critical concern involves data currency. ChatGPT models are trained on datasets with a fixed knowledge cutoff, which means they may lack awareness of therapies approved or guidelines updated after their training period ended. The rapidly evolving nature of oncology — where new targeted agents, immunotherapies, and biomarker-driven protocols emerge frequently — makes this a particularly acute problem. Institutions deploying AI tools must implement regular model updates or supplementary retrieval systems to ensure responses reflect current standard-of-care recommendations.
Equity and access considerations also emerged as central themes. While AI chatbots have the potential to improve health literacy and support underserved populations, they also risk widening disparities if implementation favors digitally connected patients in well-resourced health systems. Additionally, privacy and data security concerns must be addressed, particularly when chatbots collect or process sensitive patient health information. Regulatory alignment with frameworks such as HIPAA in the United States is non-negotiable for any clinically integrated AI tool.
- Hallucination risk: AI models can generate confident but inaccurate medical statements.
- Knowledge cutoff: LLMs may lack awareness of recently approved therapies or updated guidelines.
- Equity gaps: Digital divide may limit access for vulnerable or underserved patient populations.
- Data privacy: Integration with patient records requires strict compliance with healthcare privacy regulations.
- Regulatory uncertainty: The FDA classification of AI chatbot tools in clinical contexts is still evolving.
Speakers at ASCO 2023 broadly agreed that the path forward requires multidisciplinary collaboration — between oncologists, data scientists, ethicists, regulators, and patients — to develop AI tools that are not only technically capable but also clinically safe, equitable, and trustworthy. The consensus was optimistic but measured: the technology holds real promise, but responsible deployment demands rigorous standards, transparent validation, and sustained human oversight.
Frequently Asked Questions
What types of ChatGPT chatbot tools were highlighted at ASCO 2023?
Presentations at ASCO 2023 highlighted tools designed for patient education, symptom triage during active cancer treatment, clinical documentation assistance, literature synthesis, and preliminary clinical trial matching. Most platforms were built on large language models and positioned as supplementary aids to clinical teams rather than autonomous decision-making systems. Oversight by trained oncology professionals was considered essential in all demonstrated use cases.
Are AI chatbots safe to use for cancer-related medical decisions?
Current evidence suggests AI chatbots can provide useful informational support, but they are not yet considered safe or reliable enough to guide independent clinical decisions in oncology. Risks include generating inaccurate information, lacking awareness of the latest treatment guidelines, and missing nuanced patient-specific factors. All AI-generated content in a clinical context should be reviewed and verified by a qualified healthcare professional before being acted upon.
How are oncology institutions managing the regulatory challenges of AI chatbot deployment?
Institutions are approaching regulatory compliance by limiting chatbot roles to informational support, maintaining clinician oversight, and implementing audit mechanisms to detect and correct inaccurate outputs. Privacy frameworks such as HIPAA must be satisfied before any patient data is processed through AI systems. The FDA continues to develop guidance on how AI-based clinical tools should be classified, validated, and monitored, making ongoing regulatory engagement a priority for health systems exploring these technologies.
