IBM Watson: Final Jeopardy! and the Future of Watson – Motivation to Step Up Your Game in 2012 and Beyond :)

I just showed these two videos (one above, one below) in my Quantitative Methods for Lawyers Class here at MSU Law. I think it is important to try to let people know where things really stand with Big Data and ‘Soft’ Artificial Intelligence (i.e. just exactly what is possible today – and what is soon going to move from the high end lab to the enterprise level to the level of personal computing).

Challenge: Watch these two videos and ask yourself the following question: Am I really sure that professional fields such as law, medicine, education, etc. are not going be dialed in for a real quant and technology invasion? Marc Andreessen thinks so (yeah software is indeed eating the world).

One of the purposes of this blog is to create a venue to highlight what the rise of computation, software, big data, soft AI, etc. mean for the future of legal education and the legal services industry. Our business is going to change (has changed) and obviously there is a major disconnect between the scholarship / research and development side of our industry and the market for legal education/legal academics.

The sophisticates in legal academy are obsessed with causal inference and experimental methods. That is completely understandable as these approaches are now mainstream methods to do rigorous policy evaluation and social science work with law applications.

Let me encourage folks to think a bit more broadly about what falls in the broader set of rigorous approaches. In short, engineering is über rigorous (even if there is not an IV regression to be found).

We are entering the age of Quantitative Legal Prediction (for a historical analog see Quant Finance). Indeed, as I will argue in forthcoming work this is key to the future of our industry. Causal inference (as well as experimental methods) are not the core of what won the NetFlix Prize or what built IBM Watson. Rather, it was Machine learning, Network Analysis, Natural Language Processing, Probabilistic Graph Models, the Science of Similarity and Algorithm Construction, etc.

Here is the good news – those who have been trained in high end methods are in striking distance of the above (although it would be useful to learn (or collaborate with someone who knows) a real programming language). R is not a real programming language and STATA does not even pretend to be one.

Yesterday’s Fast Is Today’s Slow — Time to step up your game 🙂