Today I’m going to propose a framework for determining whether a being is conscious. I call it the Furmanek Test (just like the famous Turing Test). The test is based on identifying paradoxes among problems that can be proved unsolvable.

General Idea

According to Wikipedia, “a paradox is a logically self-contradictory statement or a statement that runs contrary to one’s expectation. It is a statement that, despite apparently valid reasoning from true or apparently true premises, leads to a seemingly self-contradictory or a logically unacceptable conclusion.” The crucial part I consider here is that the statement is contrary to one’s expectations.

I propose that we use “having expectations” as a clue whether something is conscious or not. There are many reasons why it’s useful:

• A problem may have an unexpected result that we don’t consider a paradox. This might be because we didn’t have expectations (in that case the solution is “surprising” rather than “unexpected”) or our expectations were not reasonable (we made mistakes or we didn’t have enough knowledge of the domain to formulate viable expectations)
• Expectations can be explained and justified to some extent. We can explain why we expect something to happen by providing examples or analogies even though we may not be able to define the reasons for having expectations formally. Just like with feelings, we can describe them even though we may not be able to define them scientifically
• Paradoxes are not trivial. It takes some maturity and knowledge to really understand paradoxes and why they are “unexpected”
• It’s typically easy to solve the problem (or show that there is no solution). The challenge is not in solving the problem. It’s rather in explaining why the solution is strange/surprising/unexpected and how to actually solve the paradox (which typically takes formalizing a domain model or changing assumptions)

That being said, we can use paradoxes to analyze the reasoning. We can’t easily formalize what constitutes a paradox, but we can often tell whether something is a paradox or not.

Furmanek Test for consciousness

The test goes as follows:

1. We define a set of simple problems. Some of them are unsolvable, some of them can be easily shown that there are no solutions, and some of them are paradoxes
2. We ask the being to identify which problems are paradoxes
3. We assess the answers. If the being identifies all paradoxes and doesn’t have any false positives, then we conclude it’s conscious

Simple as that. Let’s now discuss some important remarks:

• We should have many problems (for instance a hundred), 10% of which are paradoxes. This is to make it very unlikely that the being identifies the paradoxes by pure chance
• Obviously, the problems should be novel ones so that the being doesn’t know them from the Internet (which is especially important if we’re talking about an AI-based being that is trained on the publicly available materials)
• Paradoxes should be widely accepted as such. Not many question that the Barber Paradox is a paradox while quite a few suggest that Wild Card Poker paradox is not a paradox at all

Obviously, just like with an IQ test, we have a general idea how to perform the test, but we need to make sure that the actual implementation of the test is of a high quality.

Discussion

There are many reasons why this test may not work well. I consider only some of the arguments here.

Consciousness versus human-like consciousness

The first thing to consider is whether this test identifies consciousness in general or some human-specific consciousness. It may be quite reasonable to assume that while we perceive some problems as paradoxes, other beings may not agree. The concept of paradox is based on “expectations”. If some other beings have different expectations, then the paradox isn’t a paradox anymore.

This can be obviously generalized to cultural-dependent consciousness. We may have different expectations based on our culture, language, senses, education, age, or even if we write left-to-right. The actual test should be as resilient to these aspects as possible.

Levels of consciousness

Second, it’s questionable whether humans’ consciousness is a subset of mammals’ consciousness is a subset of animals’ consciousness is a subset of Earth-beings’ consciousness. This test may be very narrow in terms of that it actually determines the human-like consciousness instead of the general one.

Formulation of the problems

We need to formulate the problems in a way that they don’t suggest assumptions or rely on some specific reasoning. For instance, with the barber paradox, we explicitly ask if “the barber shaves himself”. We don’t ask something based on maths (like “is there a set of people that shave themselves”) or require specific formal formulation. At the same time, the problems must be defined using some language that shouldn’t impact the answer. Generally, we need to solve very similar issues as with the IQ tests.

Asking to identify paradoxes or not

In the test formulation, I specify that we ask the being to identify the paradoxes. However, we may not be that specific. We may ask the being to analyze the problems and share some thoughts. If they do realize there are paradoxes, then we may differentiate between the “levels of consciousness”.