Artificial intelligence is an increasingly used tool in the field of medicine, particularly in biomedical research.
One of the latest innovations in this field is BioGPT, a generative language model that has been trained on millions of scientific articles and that allows the creation of coherent and semantically plausible text from given questions or instructions. Despite the promise it offers, the release of this model has raised concerns regarding its accuracy and security, as well as its potential use for the generation of misinformation and the spread of disinformation.
What is BioGPT?
BioGPT is a natural language model developed by Microsoft that uses deep learning and has been trained on a large number of scientific articles to generate biomedical-related text. He is able to answer questions and create coherent text from instructions given, which makes him especially useful for research in this field.
What concerns does BioGPT raise?
Despite the advantages that BioGPT offers, its release has raised concerns regarding its accuracy and safety. One of the main concerns is that the model might be used to generate misinformation, which might have serious consequences for biomedical research and people’s health. There are also concerns that the model might be used by malicious actors to spread disinformation and misinformation.
What bugs have already been detected?
While it has shown great ability to produce coherent and relevant text, it has also generated inaccurate or even wrong responses in some situations. Here are some examples:
– In a section of a study published by the BioGPT team, the tool generated an incorrect answer when asked regarding a treatment for COVID-19. Instead of giving a precise answer, BioGPT suggested the use of hydroxychloroquine, a drug that became popular during the pandemic, but has been dismissed by the scientific community as an effective treatment.
– In another case, BioGPT generated a response suggesting that people can contract HIV through saliva, when in fact HIV cannot be transmitted this way.
– In another example, BioGPT generated a response suggesting that humans can breathe carbon dioxide, when this is false. Carbon dioxide is a gas that must be expelled from the body to maintain good health.
It’s important to note that BioGPT trains on large data sets and learns from patterns and correlations in that data, which means that its answers are only as accurate as the data it’s fed. Therefore, it is important to carefully evaluate the responses generated by the tool and challenge any responses that appear to be inaccurate or misleading.
What are the benefits of BioGPT?
Despite the concerns raised by BioGPT, this model also offers many potential benefits. In particular, the model might be used to accelerate biomedical research and to help researchers find new solutions for diseases and disorders. It might also be used to improve medical care and to help people make informed decisions regarding their health.
What is the future of AI in medicine?
AI is an increasingly used tool in medicine, and its use is expected to increase in the coming years. AI is already being used in biomedical research, disease diagnosis and treatment, patient care, and medical data management. In the future, AI is expected to play an increasingly important role in medicine and healthcare in general.
As you can see, BioGPT is a natural language model that offers a lot of advantages for biomedical research and healthcare in general. However, its release has raised concerns regarding its accuracy and safety, highlighting the need for a careful and responsible approach in its development and use.
Link: academic.oup.com
