Structured Computational Prowess vs. Human Irrationality

human hand touching robot hand

Considering its ubiquity, marketing departments clearly believe the mere mention of AI makes their solutions seem more robust and cutting edge. It’s as if the very invocation suggests a semi-infallibility—an ‘if the machine says so, it must be right,’ mentality.

Part of this may be our outsized notions of our own intelligence. If a machine is as smart as we are, the logic goes, but can absorb, sort, and cull infinitely more data in a set amount of time, we’re golden!”

Unfortunately, we’re not as bright as we think we are (sorry) and it’s our creative irrationality, not our logical input/output computational ability, that distinguishes us. Herbert L. Roitblat, author of “Algorithms Are Not Enough; Creating General Artificial Intelligence,” identifies the difference between ‘path problems’ and ‘insight problems.’ At the former, AI excels. At the latter, it sucks. Per Roitblat:

Insight problems generally cannot be solved by a step-by-step procedure, like an algorithm, or if they can, the process is extremely tedious. Instead, insight problems are characterized by a kind of restructuring of the solver’s approach to the problem. In path problems, the solver is given a representation, which includes a starting state, a goal state, and a set of tools or operators that can be applied to move through the representation. In insight problems, the solver is given none of these.

In business tech, the goal is to automate and optimize, which means turning every problem into a path problem that is solved by following an approved procedure. Simply move step by step to arrive at an appropriate result. These problems can also be prime targets for automation, which reduces costs. Under this scheme you either automate, or you ensure employees follow the same process, which, to be frank, idiot-proofs the enterprise. Just find individuals who can follow the procedure as opposed to those who discover novel solutions. 

With path problems, the solver can usually assess how close the current state of the system is to the goal state. Most machine learning algorithms depend on this assessment. With insight problems, it is often difficult to determine whether any progress at all has been made until the problem is essentially solved.

In the enterprise, the path problem approach also ensures employees “show their work.” An employee discovering a solution via a eureka moment cannot be duplicated. An employee reaching a goal via regimented steps can infinitely scale.

The goal of business is to maximize return. The idea of AI is so popular due to its potential to extend the range of path problems. The goal, per Roitblat, is to avoid extending it into areas where insight has significant value.

If all that is necessary for a machine learning system is to engage its analytic capabilities, then the machine is likely to exceed the capabilities of humans solving similar problems. Analytic problem solving is directly applicable to systems that gain their capabilities through optimization of a set of parameters. On the other hand, if the problem requires divergent thinking, commonsense knowledge, or creativity, then computers will continue to lag behind humans for some time.

Focusing too heavily on well-structured path problems, Roitblat observes, is like “looking for your lost keys where the light is brightest.” Opportunities for success go unexamined.

-Leonce Gaiter, VP of Content and Strategy