![\"David](\"https:\/\/www.newsbeep.com\/nz\/wp-content\/uploads\/2026\/10\/David-Bau.jpg\")
08\/23\/22 \u2013 BOSTON, MA. \u2013 David Bau, Assistant Professor of Computer Science at Northeastern Khoury College, poses for a portrait on Aug. 23, 2022. Photo by Matthew Modoono\/Northeastern University<\/p>\nAsk an AI chatbot for some medical advice and it might recommend you head to the hospital when all you really have is indigestion.\u00a0<\/p>\n
One of the biggest issues with AI chatbots today is that they can still share false information, according to Northeastern experts.\u00a0<\/p>\n
And researchers still don\u2019t really understand how to get to the bottom of a chatbot\u2019s miscalculation or potential deception, because they don\u2019t have a strong grasp on how AI models produce their answers, the researchers say.<\/p>\n
Despite advancements made in computing power, higher quality training data and new reinforcement learning techniques, AI models largely remain an enigma.\u00a0<\/p>\n
\u201cWhen we use these systems, we don\u2019t know what their algorithms are,\u201d said David Bau, a professor in the Khoury College of Computer Sciences.\u00a0<\/p>\n
An AI model from a research lab such as OpenAI or Anthropic performs billions of mathematical equations to produce even one response, numbers that are so large it\u2019s often impossible for researchers to decipher. This is why AI is called a \u201cblack box,\u201d according to researchers.\u00a0<\/p>\n
The fundamental issue with AI models is that unlike traditional software systems, humans do not explicitly program them to function, added Byron Wallace, a professor in the Khoury College of Computer Sciences, who like Bau works in AI interpretability.\u00a0<\/p>\n
While humans are involved in building and training the models, once they are up and running, these systems work independently to come up with their answers to prompts, he explained.\u00a0<\/p>\n
08\/23\/22 \u2013 BOSTON, MA. \u2013 David Bau, Assistant Professor of Computer Science at Northeastern Khoury College, poses for a portrait on Aug. 23, 2022. Photo by Matthew Modoono\/Northeastern University<\/p>\n
![\"Portrait](\"https:\/\/www.newsbeep.com\/nz\/wp-content\/uploads\/2026\/10\/Byron_Wallace.jpg\")
08\/16\/17 \u2013 BOSTON, MA. Byron Wallace poses for a portrait on Aug. 16, 2017. Photo by Adam Glanzman\/Northeastern University
\nDavid Bau and Byron Wallace Wallace are members of the new AI consortium. Photos by Matthew Modoono\/Northeastern University and Adam Glanzman<\/p>\n
For the past two years, Bau has been working under the hood at Northeastern\u2019s University National Deep Inference Fabric,<\/a> or NDIF, a National Science Foundation funded lab based in Northeastern focused on AI, to get to the bottom of how AI models function.\u00a0\u00a0<\/p>\n Many AI systems in use today are based on neural networks, software algorithms designed to function loosely to the human brain. At NDIF, Bau has been putting AI through theoretical MRI machines to understand the synaptic connections that make those neural networks work.\u00a0<\/p>\n More specifically, Bau and his team do this work using open source AI models that are free and publicly available \u2013 as opposed to private models made by Anthropic and OpenAI \u2013 super computers and a specialized software library called NNsight that<\/a> shows a model\u2019s neurons firing as it answers prompts.\u00a0<\/p>\n But working to understand AI\u2019s black box component is only part of the story.\u00a0<\/p>\n Equally important is developing a \u201cwhite box\u201d for AI, which Bau describes broadly as a lie detector that can be used to help uncover when an AI is convincingly being deceptive.<\/p>\n When researchers talk about building a lie detector for AI, they are referring to being able to pick up on certain patterns in the AI model\u2019s neural network activity that determine it is tricking a user.\u00a0<\/p>\n The hope is to share those patterns with people who use these systems so they can tell when a chatbot is being deceptive, Bau said.\u00a0<\/p>\n It\u2019s becoming increasingly more important because people are using AI more regularly for life advice, medical consultation and even to go over their finances. But they may be unwittingly receiving false or misleading information, and researchers need to find a way to counter that.<\/p>\n \u201cAI has developed to a point where we have to recognize that it is a serious science, and like other sciences, it has its own safety issues,\u201d Bau said. \u201cBut it is developing so quickly that one of my concerns is that our ability to create AI is well ahead of our ability to be knowledgeable, responsible scientists about it.\u201d\u00a0<\/p>\n Wallace, whose research has uncovered racial biases in generative AI systems used in health care, added that without tools to understand how these systems work, users could be putting themselves in dangerous situations.\u00a0<\/p>\n