Cognitive AI as a Teammate: CMU Researchers Chart a Path for Human-AI Collaboration in Healthcare

As artificial intelligence moves deeper into healthcare and everyday decision-making, Carnegie Mellon University查看公司主页 researchers are exploring how AI can help people make better decisions while preserving the judgment, expertise and accountability that humans bring. The work centers on "cognitive AI," a distinct approach that draws on scientific understanding of the human mind as an information-processing system, using models of how people learn and make decisions to design systems that work effectively alongside people rather than simply imitate or replace them.

"The idea of intelligence has changed through the years, and now it is changing to be a collaborative human-AI intelligence," said Cleotilde "Coty" Gonzalez, a professor in the Dietrich College of Humanities and Social Science's Department of Social and Decision Sciences. "It's what we can do together with AI that makes us smarter, more capable and more able to do things that we were not able to do before."

Gonzalez, along with collaborators including Aarti Singh, professor in CMU's Machine Learning Department, and Anita Williams Woolley, professor of organizational behavior at CMU's Tepper School of Business, have recently published three scientific papers investigating the facets of cognitive AI and how it could partner with humans on teams.

What Cognitive AI Is — and Is Not

Cognitive AI is distinct from generative AI models such as ChatGPT or Grok, which generate text by learning statistical patterns from massive amounts of text data. It also differs from artificial general intelligence or artificial super intelligence, systems that could match and then exceed human intellect across a wide variety of tasks — concepts that remain hypothetical even if common in science fiction.

Rather, Gonzalez works on AI that draws on scientific understanding of the human mind as an information-processing system. "Everything I do is related to decision making, how humans make decisions," said Gonzalez, who is also the founding director of the Dynamic Decision Making Laboratory and research co-director of the National Science Foundation AI Institute for Societal Decision Making (NSF AI-SDM). "I want to understand how we use experience to learn and make better choices over time, and the role technology can play in that."

A defining feature of cognitive AI is its ability to learn from individual humans over time. "At first, a cognitive model may predict a person's choices with little more than chance-level accuracy," Gonzalez said. "But as it observes more decisions, it learns from that individual, step-by-step. In some controlled decision-making tasks, after only a few dozen decisions, our models have reached up to 95% alignment in predicting what a person will do."

Importantly, the goal of this technology is not to take away human decision-making, but to augment it. "A lot of conversations around AI have become twisted into the idea that we want to replace humans. But that is not what we want to do. We want to empower humans," said Gonzalez.

Designing Around the Consequences of Error

In a radiology setting, both human radiologists and AI systems can review scans for cancer (搜索)搜索该疾病. The researchers emphasize that not all mistakes carry equal weight. "Both types of mistakes are not equal," said Singh, a professor in CMU's Machine Learning Department and director of NSF AI-SDM. "If the AI fails to detect cancer, that's a bigger deal than if it raises a false-positive."

Rather than combining human and AI judgments in a one-size-fits-all workflow, Singh argues the system should be designed around the consequences of different errors. In a cancer (搜索)搜索该疾病-screening context, that might mean using AI specifically to flag cases the radiologist may have missed, rather than treating the AI as a simple replacement for the first human reader.

The clinical value of such collaboration is measurable. A study published in the Lancet Digital Health in 2022 found that when a human radiologist was paired with a machine learning AI (搜索)搜索该靶点, the team achieved a 2.6 percentage-point improvement in sensitivity over a radiologist alone. While that may seem like a small improvement, the study was conducted on nearly 1.2 million mammograms, demonstrating how even small percentage-point gains can matter at population scale.

In another example, Singh described an AI-human team tasked with providing health information to pregnant mothers with limited resources. One option might be for AI agents to call the mothers directly, but mistakes in a sensitive health context could erode public trust. "But you could change the way AI is used. Rather than talking with people directly, it could identify who's most at risk and then match the limited healthcare workers with them," said Singh. "In both examples, it's really about thinking how to combine human and AI expertise in complementary ways to get the best of both."

Where Human-AI Teams Add the Most Value

Woolley sees human-AI teams being especially useful in three areas where humans often struggle: identifying the right expertise, coordinating work and aligning goals.

"In terms of people, AI can help us figure out who has the particular kind of expertise we need to draw on in a given moment," said Woolley. Beyond storing and utilizing institutional knowledge, this can help correct mismatches within human teams where introverted but more qualified experts are overshadowed by those with more confidence or ability to speak up.

Goal alignment, Woolley noted, is an underappreciated weak spot. "I think people don't realize how often they get together in teams and the goals are not really clear or aligned, or maybe people have competing goals or incentives or different understandings of about the problem they're trying to solve," she said. "And nothing else really matters until you have that handled."

Risks, Trust and Responsible Design

The researchers are candid about the potential pitfalls of human-AI teams. "There's a chance AI could actually weaken teams if we use it poorly," said Woolley. "People may all decide they don't need to work with others anymore because they can just do everything with ChatGPT." However, she added that research to date suggests such an attitude will probably not result in better products for individuals or organizations.

Trust is another potential weakness, because humans tend to be less forgiving of AI's mistakes than they are of other humans. "Consider the robotaxis in San Francisco. Some analyses suggest that AI-driven cars perform well on certain safety metrics," Singh said. "But if there's an accident, a single incident can completely break that trust."

At the same time, the researchers argue that the answer is not to reject AI, but to design systems where humans and AI complement one another intentionally. "Some decisions may be fully automated because they're routine and well-defined, while others require human judgment, ethical reasoning or contextual understanding," said Gonzalez. "The important question is not whether humans or AI should decide alone, but how we design systems where they work together effectively and responsibly."

That means building AI systems with transparency, accountability and trust from the beginning. "We should not deny the power of technology or avoid integrating with it," Gonzalez said. "What matters is creating systems that enhance human capability, reflect human values and help people and AI make better decisions together."