Microsoft: The path to medical superintelligence

Dear Readers,

The future of medicine doesn't start with a prick, but with a bang: Microsoft presents the AI Diagnostic Orchestrator(MAI-DxO), a system that shows for the first time how machines can diagnose better than doctors.

What makes MAI-DxO so special is that it is not a single AI, but a consortium of several specialized agents that question and criticize each other and work out a diagnosis together - like an interdisciplinary team of doctors, only digital. These “machine conferences” act model-agnostically, i.e. independently of the underlying language model.

In Today’s Issue:

All the best,

The AI revolution in healthcare starts with a powerful blow: Microsoft unveils a new era of “medical superintelligence”!

In its latest article, Microsoft describes how the Microsoft AI Diagnostic Orchestrator (MAI-DxO) performed significantly better than human doctors in tests with 304 complicated cases from the New England Journal of Medicine.

The trick: Microsoft does not rely on a single AI, but on the interaction of several AI agents that question each other like a team of medical experts, exchange hypotheses and arrive at a diagnosis together. These “consortia of machines ” combine clinical knowledge with diagnostic logic - model-agnostic, i.e. independent of the language model used.

This is a real quantum leap for the AI community: instead of just generating texts, these systems argue, weigh up and prioritize like human experts. This opens up a new chapter - that of decision-capable AI in the healthcare sector that not only helps, but also acts.

What does this mean for hospitals, research - and for all of us as patients? And how do we ensure that these systems work safely and fairly?

Why it matters: These AI diagnostics show for the first time that machines can make expert-level medical decisions - quickly, scalably and across models.

It is a foretaste of a healthcare system in which precision, efficiency and accessibility are redefined by AI.

Source:

🔗 https://microsoft.ai/new/the-path-to-medical-superintelligence/

AdQuick unlocks the benefits of Out Of Home (OOH) advertising in a way no one else has. Approaching the problem with eyes to performance, created for marketers with the engineering excellence you’ve come to expect for the internet.

Marketers agree OOH is one of the best ways for building brand awareness, reaching new customers, and reinforcing your brand message. It’s just been difficult to scale. But with AdQuick, you can easily plan, deploy and measure campaigns just as easily as digital ads, making them a no-brainer to add to your team’s toolbox.

Learn more now.

A new paper shows how AI stimulates itself to deeper logical thinking through repeated duels in zero-sum games. In the multi-agent environment, models develop strategic behavior that goes beyond mere reaction. The highlight: multiple moves promote long-term planning. This opens up new avenues for autonomous problem solving - e.g. in negotiations, simulations or game theory.

With AlphaGenome, DeepMind presents a model that can read and interpret extremely long DNA strands - including previously elusive regions. It predicts genetic risks and disease predispositions much better than previous methods. The new architecture combines biology with scalable AI and could become a milestone in personalized medicine.

A team is proposing a new architecture in which AI takes into account not only the current state, but the entire history - a real advance over classic Markov models. This makes AI more robust in dynamic environments, such as stock market forecasts or robotics in changing scenarios. AI learns: the past counts.