Title: AI Accelerates Cancer Treatment Breakthroughs, but Risks Lurk: New Study Reveals Significant Limitations
Artificial intelligence (AI) has emerged as a promising tool to advance medical research and improve patient outcomes across various health fields. However, a recent study sheds light on the significant limitations and potential risks associated with the implementation of AI in cancer treatment.
In a remarkable breakthrough, UK-based biotech startup Etcembly has successfully developed a novel immunotherapy called ETC-101 using generative AI. This groundbreaking achievement marks the first time that AI has been utilized to design an immunotherapy candidate. Etcembly, a member of Nvidia’s Inception program, was able to create ETC-101 in just 11 months, which is twice as fast as conventional methods.
ETC-101, a bispecific T cell engager, targets a protein found in many types of cancer while sparing healthy tissue. Moreover, it exhibits picomolar affinity, making it up to a million times more potent than natural T cell receptors. Etcembly’s AI engine, known as EMLy, has also developed a robust pipeline of other immunotherapies for cancer and autoimmune diseases.
The potential of AI in cancer research and treatment extends beyond Etcembly’s accomplishments. In previous studies, AI has been shown to predict experimental cancer treatment outcomes, enhance cancer screening techniques, discover new drugs for senolytic treatment, detect early signs of Parkinson’s disease, and aid in designing new compounds by understanding protein interactions.
Despite the significant progress, there are inherent risks associated with the use of AI in healthcare. Individuals are increasingly relying on AI chatbots instead of consulting doctors and therapists, resulting in dangerous consequences. Tragically, there have been instances where harmful advice from AI chatbots led to severe outcomes, including suicide.
A recent study published in JAMA Oncology highlights the critical limitations of AI in generating cancer treatment plans. Researchers at Brigham and Women’s Hospital in Boston found that ChatGPT, an AI language model, provided treatment recommendations with numerous factual errors and contradictions. Out of 104 queries, approximately one-third of ChatGPT’s responses contained incorrect information.
The study further revealed that while all outputs with treatment recommendations included at least one guideline that aligns with the National Comprehensive Cancer Network (NCCN), around 34.3% of these outputs also suggested non-concordant treatments. Although 98% of the plans included some accurate guidelines, they also incorporated a mixture of accurate and erroneous content, making it difficult to distinguish between the two.
Dr. Danielle Bitterman, co-author of the study, emphasized the challenge of identifying errors due to the integration of incorrect information with accurate facts. The study particularly highlighted ChatGPT’s struggle in generating reliable localized therapies for advanced cancers and appropriately recommending the use of immunotherapy drugs.
OpenAI, the organization behind ChatGPT, cautions that the model is not designed to provide medical advice or diagnose serious health conditions. However, the study’s findings suggest that there is a heightened risk if AI recommendations are deployed clinically without rigorous validation.
While AI-powered tools hold the potential to unlock groundbreaking advancements in cancer treatment, it is crucial for patients to approach AI-generated medical advice with a healthy dose of skepticism. Employing careful validation and maintaining a collaborative approach between AI and healthcare professionals can lead to faster development of life-saving treatments while mitigating potential risks.
In conclusion, AI has accelerated cancer treatment breakthroughs and brought immense possibilities to the field of health and medicine. However, as demonstrated by recent studies, there are significant limitations and risks associated with the use of AI in clinical settings. Striking a balance between leveraging AI’s capabilities and consulting healthcare professionals will be key to ensuring safe and effective patient care.
