Cancer Grand Challenges exists to tackle some of the biggest and most important questions in all of science. The latest one is especially timely. From today, the initiative’s researchers will start a rigorous test to see if artificial intelligence (AI) can help them make potentially lifesaving progress faster. 

Through the £8m AURORA project, a team at Harvard University has developed a specialised AI “co-scientist” platform, which three Cancer Grand Challenges teams will now use and evaluate as part of their ongoing research.

AURORA’s AI co-scientists are designed to act as virtual lab assistants for flesh-and-blood researchers. They can’t make any scientific decisions on their own, but, guided by real scientists’ expertise, they can analyse vast amounts of data, generate new ideas based on hidden patterns, and prioritise the most promising experiments. 

The teams will test the technology as they investigate why some normal cells turn cancerous, how to help the immune system recognise and attack tumours, and how to improve immunotherapy for children and young people with cancer. 

“Behind every advance in cancer care are researchers asking difficult questions, testing ideas and following the evidence,” said Michelle Mitchell, Cancer Research UK’s chief executive. “AI offers new possibilities, but it takes skilled scientists to turn a promising idea into a discovery that can make a difference to people’s lives.  

“Through Cancer Grand Challenges, we’re bringing AI together with the expertise of outstanding researchers to tackle some of cancer’s toughest problems. Our ambition is to accelerate discoveries that could lead to new ways to prevent, diagnose and treat cancer.”

AI designed for cancer research

Unlike AI chatbots, which all too often “hallucinate” fake answers to difficult questions, AURORA’s co-scientists are designed to stick to real scientific evidence and recognise when things are uncertain.  

In practice, this means that they always draw from properly reviewed scientific resources and research data. And when the data needed to answer a question isn’t available, they’ll say so.

The system can also do more than just generate text. AURORA uses a different technology called “agentic” AI, which means the co-scientists have the autonomy and reasoning powers to guide themselves through tasks step by step. Researchers will still have control and oversight, but they don’t have to spend all day at the keyboard. They can ask their AI lab partners to comb through vast libraries and databases without constantly checking in.

The idea behind this approach is that it will give researchers more time to focus on the parts of their projects that need their expertise. Meanwhile, the co-scientists can use existing data to evaluate questions and identify the experiments most likely to generate new insights.

As always in science, the experiments are the most important part of the process. AURORA co-scientists could, for instance, spot a hidden pattern in a vast database and use it to suggest a new hypothesis, but scientists will still need to test it and evaluate the results.

As well as contributing to the research projects, those tests will help with evaluating AURORA’s outputs and feed back into the system itself, helping improve it so it can make better suggestions for what to investigate next.

Meet the teams

Crucially, Project AURORA is itself an experiment – one that Cancer Grand Challenges teams are especially well-suited to carry out. Rather than applying AI to abstract problems, they’re in a position to find out whether it can support research with the potential to change lives.

“A key challenge for AI in science is understanding how it can best support researchers working on real-world problems,” said Marinka Zitnik, an Associate Professor at Harvard Medical School and the lead researcher behind the AURORA system. “With project AURORA, we’re putting AI co-scientists into active cancer research programmes and rigorously evaluating its impact. 

“By testing the system against real scientific questions and experimental evidence, we hope to build a trusted, rigorously evaluated tool that helps researchers accelerate discovery and uncover new insights into cancer research.” 

Teams will apply the technology to three separate problems linked to cancer:  

1. Developing CAR T-cell therapies for children’s cancers  

The NexTGen AURORA team will use AI to explore how immune cells can be modified to better find and break down solid tumours in children and young people. Specifically, researchers will work with AI co-scientists to identify what features of cancer cells they should target and how to overcome obstacles that can stop immune cell treatments from working.  

2. Understanding what turns a normal cell cancerous  

The PROMINENT AURORA team will explore why some cells with cancer-linked genetic changes stay healthy while others turn cancerous. The AI co-scientists will help suggest explanations for researchers to test, including how age, sex and exposure to cancer-causing substances might influence this process. 

3. Helping the immune system recognise cancer  

The MATCHMAKERS AURORA team will use AI to explore how T cells recognise cancer. The researchers will test which markers on cancer cells make the most promising targets and which immune cells are best equipped to attack those cells – work that could help shape more personalised treatments.