Ray Kurzweil predicted that in less than 30 years non-biological intelligence will be a billion times more capable than humans. This will more than strain our epistemic boundaries. Binary humans are incapable of digesting a not-to-distant future dominated by qubits and wetware. That’s why, from a contemporary vantage point, we’re dealing with an event (or perhaps a series of events) that have all of the necessary ingredients to irreversibly shatter epistemic configurations.
These capabilities will be fueled by quantum computers, which can be millions of times more powerful than the most powerful computers we have today. Calculations that are practically impossible for even the most powerful contemporary computers (since the length of time they would require is measured in thousands of years) can take quantum machines mere seconds to resolve. Being able to instantly resolve highly complex calculations in seconds offers unprecedented capabilities in many areas such as medicine, pharma, engineering and computational law applications.
We’re already in the midst of this change. Not so long ago, back in April 2014, IBM announced they had succeeded in accomplishing a critical milestone with a 4 qubit chip. About a year later, Google more than doubled that, with a 9-qubit chip. Both of these achievements were done with the requisite self-error detection threshold that makes for reliable qubits; in other words, these companies achieved not a theoretical, but a viable quantum computing state. See, State Preservation by Repetitive Error Detection in a Superconducting Quantum Circuit (J. Kelly et al. Nature 519 66-69, March 2015). Microsoft, Northrop Grumman, Lockheed, Alibaba and others are pouring massive resources getting into the quantum game. Researchers at the University of New South Wales even managed to substitute expensive materials like cesium and diamonds with silicone. It is inevitable we will see more players touting their double-digit qubits in the near future.
Quantum computing renders many of the concepts I have previously discussed here viable. For example, the concept of “adaptive learning,” introduced in my first installment of “Maximizing Representative Efficacy“becomes attractive when you’re dealing with an autonomous, intelligent entity that can efficiently learn and adapt in a dynamic environment faster than any human could.
Deep learning AI algorithms are proliferating (e.g., Watson, AlphaGo) and others already power common applications (Facebook, facial recognition; instant voice translation on Skype). If we shift the timescale Kurzweil is talking about from years to minutes, then within about 5 minutes we will have AI-based computational law applications (CLA) that can deliver rudimentary legal advice through our mobile device. And approximately 10 minutes after that, we will see a new class of AI-based CLAs built on cloud-based quantum computing architectures. These legal services will be provided by cyber(netic) entities that will take on different forms (from mobile apps to cybernetic-humanoids). It is also likely that some law firms will evolve to a hybrid practice, part developer and part cloud provider of AI-based CLAs. (I first discussed this possibility here.)
Update 12-23-2017: Quantum computing-powered computational law AI applications will be capable of handling unprecedented levels (volume and complexity) of pattern analysis and data mining functions. This computing power can enable, for example, “dynamic security,” which is an adaptive learning-based system security reconfiguration that, as its name implies, responds to and processes in real time changes in a given operational environment. (Note that this concept extends from my security-by-design presentation at the 2016 Ecommerce Best Practices Conference at Stanford Law School.) The concept of dynamic security can be better understood when viewed in the context of autonomous vehicles, be they drones, cars, ships or other type of vehicle. These vehicles operate in a heterogeneous environment (comprised of manual, semi autonomous and, eventually, other fully autonomous vehicles), which is itself subject to a wide variety of laws and regulations. They need to be able to quickly and correctly adapt to ensure their continued safe operation. In a quantum computing environment, AI applications stand to be able to more accurately and quickly identify and process critical details about their environment (see Fractal Disambiguition for AI post). With that unprecedented speed and accuracy, the vehicle’s safe operation parameters can be continuously adjusted to ensure optimal operation.
Update 11-13-2017: MIT Technology Review reports that IBM recently released a 50 qubit quantum computer, and is making a 20 qubit version available through its cloud computing platform. Google is (likely only temporarily) behind, but steaming ahead with its quantum supremacy project. While a 20 qubit platform is more powerful than a 9 qubit one (Google released a 9 qubit two years earlier), it will be interesting to see what benefit, if any, AI-based computational law applications will have running on the more powerful qubit platforms. One possible benefit could be in dealing with more complex data mining applications and augmenting (e.g., features/capabilities) in mixed reality (MR) applications.
Update 4-27-2017: Google recently revealed that on its AI applications “that utilize neural network inference, [its self-developed Tensor Processing Unit] TPU is 15x to 30x faster than contemporary GPUs and CPUs (think Nvidia, and Intel). This is significant because it dramatically impacts and accelerates the proliferation and power of deep learning AI algorithms. A corollary effect is the enhanced potential for expanding the availability of AI-based computational law applications (CLA), as it is reasonably certain that Nvidia and Intel will work hard to keep abreast, if not overtake Google’s TPU specs.
Update 1-28-2017: Open sourcing quantum computing is an important step in broadening and speeding the development of quantum apps. The D-Wave open source Qbsolv is available here. Qbsolv is capable of solving large quadratic unconstrained binary optimization (QUBO) problems. As such, Qbsolv could be well-suited for building AI-based CLAs (e.g., in relation to QUBO application to pattern analysis, see Taxt, T. and ∅ivind Due Trier. Evaluation of binarization methods for document images. IEEE Transactions on Pattern Analysis and Machine Intelligence 17(3), (1995), pp. 312–315); data mining apps, see Wang, H., B. Alidaee and G. Kochenberger. Evaluating a Clique Partitioning Problem Model for Clustering Data Mining. Tech. Rep. HCES-06-04, Hearin Center for Enterprise Science, University of Mississippi (2004)).
Update 6-2-2016: The reality of AI-based CLAs on cloud-based quantum architectures just got a bit closer: “IBM scientists have built a quantum processor that users can access through a first-of-a-kind quantum computing platform delivered via the IBM Cloud onto any desktop or mobile device.” See https://www-03.ibm.com/press/us/en/pressrelease/49661.wss