It’s remarkably easy to inject new medical misinformation into LLMs
It’s pretty easy to see the problem here: The Internet is brimming with misinformation, and most large language models are trained on a massive body of text obtained from the Internet.
Ideally, having substantially higher volumes of accurate information might overwhelm the lies. But is that really the case? A new study by researchers at New York University examines how much medical information can be included in a large language model (LLM) training set before it spits out inaccurate answers. While the study doesn’t identify a lower bound, it does show that by the time misinformation accounts for 0.001 percent of the training data, the resulting LLM is compromised.
While the paper is focused on the intentional “poisoning” of an LLM during training, it also has implications for the body of misinformation that’s already online and part of the training set for existing LLMs, as well as the persistence of out-of-date information in validated medical databases.
Related Knowledge Center: Artificial intelligence
On top of a failing business model, newsrooms must navigate the powerful — and not fully reliable — new capabilities of artificial intelligence and the deployment of large language models.
State of the business model | Copyright and intellectual property
Related Knowledge Center: Information integrity
In an increasingly fractured media landscape, truth has been a casualty in the public discussion. Coverage has been warped by the business needs of struggling news organizations, and a significant segment of the public relies on bad-faith outlets that are little more than propagandists. While fact-checking has become routine, some organizations are exploring other ways to combat the trends. And misinformation spewed by AI-driven chatbots has become more of a feature than simply a bug.
Fact-checking and its detractors