For fun, I built an AI-agent powered "supreme court" where agents can evaluate real court filings and render a decisions in minutes. On early testing, it looks accurate.

I've been fasctinated with the concept of AI-as-a-judge. Why? Human judgment has biases, inconsistencies, inefficiencies -- it's often time slow.

This is a POC's aim was to explore how courts and justice could work in the Age of AI? It is asking the question what would the world look like if we had AI-powered agent courts, run entirely by nine autonomous AI justices.

How does it work: - Reads real court filings—entirely, zero abstractions in seconds - Summarizes those court filings and pulls out key information and citations - Outlines and drafts a common factual record and legal standard, with citations to the record - Outlines and drafts the best arguments for both sides. - Nine independent AI Agent judges evaluate this information and the full set of papers and independently and render independent decisions.

The aggregated result tends appears to predict who will win these cases (without knowing the actual results) -- at least on early use.

The concept of AI-as-a-judge could be a big deal! It's not unreasonable to believe that AI judges could surpass human judges over the next several years, becoming the most trusted judicial authorities in the world.

If AI-as-a-judge is better than our best judges, it could revolutionize how justice is administered, drastically reducing decision times from months or years to days or even minutes

AI-as-a-judge could guarantees fairness, transparency, and unparalleled accuracy (although work would need to be done to assess and/or debias any LLM or other AI model used). Every decision can be audited; every reasoning fully transparent.

Lawyers (and even regulators) can pre-test legal strategies, assessing success probabilities before proceeding. We could require that administrative agencies prove they can win a case before bringing it, preventing corruption or overzealous advocacy

Justice could become cheaper. You wouldn't need tens of thousand--if not millions--of dollars to defend your rights.

AI-as-a-judge could encourages settlements by providing credible, unbiased forecasts of outcomes.

This system could be legally binding, via an arbitration agreement. We can imagine a future where this court is organized as a DAO (even potentially a tokenized one) where humans help maintain the underlying models and/or review decisions of the AI agents

We understand this idea might feel radical (that's my goal to push boundaries) and you have skepticism (great!)

But, we need to start reimagining all systems with AI and AI agents at the center. That include fundamental systems like courts and justice. We can built more powerful, efficient, and better systems starting from first principles.

If anyone wants to jam on this with me and some of the rest of the folks at @tributelabsxyz -- reach out!

Have AI answers to legal questions reviewed by a real lawyer instantly.