Preparing for Legal Super Intelligence

Clients have been increasingly asking how we view artificial general intelligence (AGI) and artificial super-intelligence (ASI). We wanted to share conversations we are having with the broader legal community in case they are helpful and also so they can challenge and expand our thinking on the topic. Sam Altman notes that systems “that start to point to AGI are coming into view,” and we believe we need to start preparing as a society and an industry for increasingly powerful AI.

We define legal super intelligence as an autonomous AI that can handle any negotiation, litigation or adjudication in any practice area as well or better than the top expert in that domain. Concretely, in these three areas that means:

We believe this definition captures the core capabilities of legal superintelligence across the main functions of the legal industry. In future articles we will discuss the implementation details, how we believe such a system evolves and might be governed, and how such a system could transform the legal industry. We strongly believe this will be a transition for the better.

We believe the legal industry will make meaningful progress towards legal super intelligence in parallel with artificial general intelligence and far earlier than general super intelligence. We define artificial general intelligence and super intelligence as follows:

We believe AGI is rapidly approaching but specialized super intelligent systems will be created much sooner than generally super intelligent systems. For example, we have super human driving in the form of self driving cars, specialized AI for a specific task, for before we will build humanoid robots that learn to drive cars. In a future article we will discuss in more details how we see these different systems evolving and interacting.

It’s incredibly difficult to predict exact timelines given the complexity of research and of defining exact milestones. However, the majority of the frontier model companies we work with tend to define timelines in a roughly similar range:

We believe AI is improving much quicker than most of the world expects. Regardless of the exact definitions of AGI and ASI the models are already powerful enough that they will reshape society and we need to start preparing for the implications. Specifically for the legal industry, it means that the current state of incremental change and investment will be rapidly replaced by a fundamental transformation of the way we work. To that end, it requires thoughtful planning by Firm Innovation teams and proactive guidance from teams like ours at Harvey.

The legal profession must confront fundamental shifts in associate training, law-school curricula, and the skills lawyers need to stay competitive. Below is a summary of the topics from the various conversations we’ve been having with folks in the legal industry. The progress of AI is an incredibly deep and nuanced topic so we don’t expect to be able to cover everything and we know in a couple months a lot of what we write will likely change. However, we want to share our thoughts in case others find them useful.

Below are some of the topics that have emerged from our conversations with CIOs to date, alongside the ones we see as critical:

We would love to hear from CIOs, partners, associates, law students, and Innovation teams on other questions top of mind for you as it relates to legal AGI. These questions and feedback will guide the rest of our series on this topic, so please share them with us on Harvey's LinkedIn.