Uploaded August 2026 | Updated September 2026, 3 days ago
Decades of cancer data sit in repositories around the world, holding answers no one has had time to look for. This is the story of one of them — and of what happened when a AI system and a team of cancer researchers went looking together.
Working side by side, Ai2 and the Paul G. Allen Research Center at Providence Swedish Cancer Institute used AutoDiscovery to surface a hypothesis no one was asking about: that invasive lobular carcinoma, a breast cancer subtype long written off as unresponsive to immunotherapy, carries a stronger immune signature than anyone recognized.
This is how we believe AI for science should be built — in the lab, with the scientists who will use it, held to their standards of evidence.
The Swedish Cancer Institute is now deploying AutoDiscovery, to use on their own protected research data.
Read the full story: allenai.org/blog/swedish-autodiscovery-partnership
Decades of cancer data sit in repositories around the world, holding answers no one has had time to look for. This is the story of one of them — and of what happened when a AI system and a team of cancer researchers went looking together.
Working side by side, Ai2 and the Paul G. Allen Research Center at Providence Swedish Cancer Institute used AutoDiscovery to surface a hypothesis no one was asking about: that invasive lobular carcinoma, a breast cancer subtype long written off as unresponsive to immunotherapy, carries a stronger immune signature than anyone recognized.
This is how we believe AI for science should be built — in the lab, with the scientists who will use it, held to their standards of evidence.
The Swedish Cancer Institute is now deploying AutoDiscovery, to use on their own protected research data.
Read the full story: allenai.org/blog/swedish-autodiscovery-partnership
