Detecting Intelligence

The Turing Test and Other Design Detection Methodologies

George D. Monta˜nez

Machine Learning Department, Carnegie Mellon University, Pittsburgh PA, U.S.A.

Keywords:

Turing Test, Design Detection, Intelligent Agents.

Abstract:

“Can machines think?” When faced with this “meaningless” question, Alan Turing suggested we ask a dif-

ferent, more precise question: can a machine reliably fool a human interviewer into believing the machine is

human? To answer this question, Turing outlined what came to be known as the Turing Test for artiﬁcial intel-

ligence, namely, an imitation game where machines and humans interacted from remote locations and human

judges had to distinguish between the human and machine participants. According to the test, machines that

consistently fool human judges are to be viewed as intelligent. While popular culture champions the Turing

Test as a scientiﬁc procedure for detecting artiﬁcial intelligence, doing so raises signiﬁcant issues. First, a

simple argument establishes the equivalence of the Turing Test to intelligent design methodology in several

fundamental respects. Constructed with similar goals, shared assumptions and identical observational models,

both projects attempt to detect intelligent agents through the examination of generated artifacts of uncertain

origin. Second, if the Turing Test rests on scientiﬁcally defensible assumptions then design inferences become

possible and cannot, in general, be wholly unscientiﬁc. Third, if passing the Turing Test reliably indicates

intelligence, this implies the likely existence of a designing intelligence in nature.

1 THE IMITATION GAME

In his seminal paper on artiﬁcial intelligence (Tur-

ing, 1950), Alan M. Turing, the brilliant, perse-

cuted, and posthumously celebrated hero of Bletch-

ley Park (Copeland, 2012) asked a seemingly simple

question: “Can machines think?” His answer was that

while this question was too vague to admit an exact

answer, it could be replaced by a precise and more

meaningful question: can a machine trick a human

judge into believing it is human? To answer the sec-

ond question, he described a variant of a popular party

game, the imitation game. In his version of the game,

two subjects (one human and the other a computer)

are placed in separate rooms. Questions are asked of

the subjects, and responses are typed out on sheets

of paper (or displayed on screens). The machine’s

“goal” is to fool the judge into believing it is the hu-

man, and if it can reliably do so, we posit that the

machine is intelligent. This game and variants of it

are now known as the Turing Test (French, 2000).

While challenges to the Turing Test abound (Say-

gin et al., 2003), this paper presents a novel critique

of the Turing Test in the spirit of a reductio ad ab-

surdum. I will argue that the Turing Test procedure

cannot be viewed as scientiﬁcally valid by most con-

sistent scientists and that even its necessary assump-

tions have problematic consequences. This is shown

in three steps. First, the core logic and assumptions

of the Turing Test are drawn out. Second, it is shown

that if the Turing Test is a valid and reliable indicator

of intelligence, then this implies a designing intelli-

gence behind physical nature with high probability.

Lastly, even if we discard the Turing Test itself as un-

reliable, the assumptions necessary for any test like it

to function allow for intelligent design to be formu-

lated as scientiﬁcally sound methodology. These con-

sequences are explored and we are either forced to re-

ject the test and its underlying assumptions, or reeval-

uate the scientiﬁc standing of a theory the mainstream

academic consensus has declared pseudo-scientiﬁc.

Before we begin, let us return to Turing’s famous

paper. The question “Can machines think?” raises

a logically prior question, namely, could we even in

principle distinguish between machines that think and

those that do not? Without an afﬁrmative answer to

the second question, the ﬁrst question remains unan-

swerable. Turing proposed the imitation game as a

method of distinguishing between thinking and non-

thinking machines, between intelligent and unintelli-

Montañez, G.

Detecting Intelligence - The Turing Test and Other Design Detection Methodologies.

DOI: 10.5220/0005823705170523

In Proceedings of the 8th International Conference on Agents and Artiﬁcial Intelligence (ICAART 2016) - Volume 2, pages 517-523

ISBN: 978-989-758-172-4

517

gent causes. For the Turing Test to have any value,

there must be some empirical basis for distinguishing

between intelligent and unintelligent machines. To

this we now turn.

2 SIGNS OF INTELLIGENCE

The Turing Test consists of three core components:

1. A human judge (or judges),

2. An unknown message sender (machine or hu-

man), and

3. A set of artifacts produced by the sender (e.g.,

words on a screen).

The job of the judge is to decide, based on observa-

tion of the artifacts, whether the sender is a human or

machine. Other telltale signs and clues are removed,

such as vocal articulation, physical appearance and

differences in how likely the sender is to be human

or non-human. This leaves the artifacts as the only

observable evidence available to discern the nature of

the sender. In the language of graph theory, the chain

from sender to judge forms a Markov chain,

S → W →

where S is the true sender,W are the words that appear

on the screen and

S, a function of W, is the estimate

from the judge as to whether the sender is a computer

or a human. The test forms a Markov chain by de-

sign, since Turing’s choices of placing the sender and

receiver in different rooms and not allowing vocal-

ized communication were purposefully made to re-

move all side information. Because of this Marko-

vian structure, the data processing inequality (Cover

and Thomas, 2006) ensures that

S contains no more

information about S than given by W.

The set of artifacts, W, comprising the sum to-

tal of information regarding S available to the judge,

must either contain information that allows for reli-

able discrimination between intelligent and unintel-

ligent senders or not. If it does not, then the Tur-

ing Test cannot work even in principle, since there

can exist no reliable means of distinguishing be-

tween dumb causes (machines without intelligence)

and smart ones (agents with intelligence). If no dis-

crimination is possible between artifacts sent by unin-

telligent causes and intelligent causes, then one can-

not distinguish between unintelligent computers and

intelligent computers, which are the cause of the arti-

facts in question.

Formally, I(S;

S) ≤ I(S;W), where I(·;·) denotes the

mutual information between two random quantities.

On the other hand, if artifacts can convey informa-

tion sufﬁcient to allow for discernment of intelligent

causes, then the Turing Test might possibly work,

since it opens the possibility of determining whether

the cause of an artifact was intelligent or unintelligent.

The test was proposed to be exactly such a discrimi-

nator: when a judge repeatedly believes the machine-

made artifacts come from a human, this is offered as

an indicator of computer intelligence (i.e., an intel-

ligent cause). Thus, the purpose of the Turing Test

as a test for intelligence rests on the assumptions that

there is a difference between the powers of unintelli-

gent systems and intelligent agents and that such dif-

ferences can be reliably reﬂected in physical objects.

Furthermore, reliable discrimination must be possible

based on physical artifacts alone.

The logic underlying the Turing Test is laid out as

follows:

1. Intelligent causes, which we call agents, can be

distinguished from unintelligent causes based on

their effects.

(a) Intelligent computerscan be distinguished from

unintelligent computers based on the artifacts

they produce.

2. Minds cannot be directly observed, while physical

objects and artifacts can.

3. Physical artifacts by themselves are sufﬁcient for

discrimination between intelligent agents and un-

intelligent causes.

(a) Based on the observation of physical artifacts,

a procedure can distinguish between intelligent

and unintelligent computers.

4. The Turing Test is proposed as a reliable proce-

dure for performing such discrimination.

Denial of the any of the points (except 2) renders

the Turing Test unworkable and invalid. Furthermore,

if some procedure allowed us to perceive minds di-

rectly then the TuringTest would no longer be needed.

Point 3 may be stated more strongly as “Physical ar-

tifacts by themselves, without any knowledge of the

true sender and her attributes or capabilities, are suf-

ﬁcient for discrimination between intelligent and un-

intelligent causes,” given that the test depends essen-

tially on the judge not having side information about

the sender. This becomes important when discussing

the connection to intelligent design.

ICAART 2016 - 8th International Conference on Agents and Artiﬁcial Intelligence

518

3 INTELLIGENT DESIGN

CRITICISMS OF STRONG AI

Intelligent design advocates have publicly voiced

skepticism to the prospects of Strong Artiﬁcial Intel-

ligence and computational persons (Kurzweil et al.,

2002; Egnor, 2014; Richards, 2011; Larson, 2015;

Torley, 2014; Smith, 2015). Humans, we are told, are

unique in their intellectual abilities and mind cannot

be reduced to computation; thus, mere computational

devices can never become persons nor ever become

truly intelligent beings. The Turing Test, insofar as a

machine could pass it, would not work as advertised,

since a machine could not be intelligent in principle.

Thus, skepticism towards the prospect of Strong AI

has been openly voiced within the intelligent design

community.

However, we will see that this perceived incom-

patibility actually obscures a striking similarity be-

tween the Turing Test and the intelligent design re-

search program.

4 UNEXPECTED IMPLICATIONS

Given that the Turing Test is rooted in the assumed

capability of physical structures to carry observable

signs of intelligence, we come to an unexpected con-

nection. Intelligent design theory posits that features

of the natural world, such as biological organisms,

bear the hallmarks of being caused by an intelligent

agent and that by observation of physical structures

one can reliably discriminate between designed and

undesigned objects. If the Turing Test is a scientiﬁ-

cally and philosophically sound test then we cannot

rule out that this test might be applied to artifacts

other than words. Indeed, it takes special pleading

to prevent the test from being applied to other kinds

of objects, such as poems, musical scores, works of

architecture and molecular machines. Given the lit-

tle contested fact that humanity has a strong tendency

to attribute design to structures in nature, this would

further indicate that an intelligence behind nature has

already been judged as present: when judges repeat-

edly attribute artifacts to an intelligent mind, the Tur-

ing Test says the cause behind the artifacts is indeed

some form of intelligence.

While one could personify natural selection as an

intelligent agent, we would also need to account for

cosmological structures, which cannot as easily be as-

cribed to the powers of Darwinian natural selection

(though some have argued that this is also the case,

e.g., (Smolin, 2004)). However, invoking natural se-

lection as an intelligent agent acting across all space

and time (wise, omnipresent and eternal) brings us

remarkably close to the Charybdis of classical the-

ism. Claiming natural selection as a god-like intelli-

gent mind seems a poor way of avoiding the idea that

natural systems are the result of an intelligent mind.

Doing so essentially concedes the point at issue, even

if arguing against its interpretation.

A more sober rejoinder is that just because hu-

mans repeatedly mistake artifacts as being the result

of an intelligent mind does not mean they are. This is

true; but then what becomes of the Turing Test itself?

The test is based on the premise that if humans repeat-

edly mistake artifacts as being the result of an intelli-

gent agent then they likely are. However, if humans

consistently judge the cause of observed artifacts to be

intelligent, and yet this does not force us to consider

the cause intelligent, then a successfully passed Tur-

ing Test becomes meaningless: fooled human judges

would serve as no indicator of prior intelligent activ-

ity. Thus, an inherent tension exists between the logic

of the Turing Test and unexpected consequences that

follow from its validity.

Taking a step back, we might argue that while the

applied logic of a generalized Turing Test is not suit-

able for discriminating between intelligent and unin-

telligent causes, other methods may be. While this

avoids the problem of having already conﬁrmed the

likely existence of design in nature, it leaves open the

possibility of design inference in general. If physical

objects can reliably transmit signals of a designing in-

telligence, even absent of knowledge concerning their

origins or the additional capabilities and attributes of

their true causes, then research into methods of em-

pirical design detection cannot be ruled out a priori.

Furthermore, such methods might one day be applied

to artifacts of nature, such as biological systems and

cosmological structures. If reliable methods of de-

sign detection are developed, then we cannot rule out

the possibility that natural objects themselves might

contain such signs of intelligence. If they did, intel-

ligent design theory would not only give a sound sci-

entiﬁc methodology but also a true theory of origins.

If such possibilities cannot be entertained then neither

the methodology nor the logic underlying the Turing

Test can be treated as sound.

We now see that the Turing Test and intel-

ligent design methods attack different variants of

the same underlying problem: detecting intelligent

causes through the examination of generated artifacts.

While there are differences between the two methods,

the similarities are striking. A solution to one prob-

lem set provides a solution to the other, since reli-

able methods of design detection can be applied to

words written on a computer screen as easily as to en-

Detecting Intelligence - The Turing Test and Other Design Detection Methodologies

519

gineered objects like machines. In both cases, uncer-

tainty surroundsthe true sender and his or her capabil-

ities. In both cases, physical objects are examined to

determine whether the generativecause was an intelli-

gent agent or merely an unintelligent physical system.

Fundamentally, the Turing Test is applied intelligent

design: a special case of design detection applied to

computational artifacts. Thus, the scientiﬁc validity

of one endeavor is intrinsically tied to the scientiﬁc

validity of the other.

5 POSSIBLE OBJECTIONS

Let us now propose some arguments that might sal-

vage the Turing Test and its underlying assumptions.

5.1 The Language Objection

It may be argued that the textual and verbal aspects of

Turing Test are inseparable from it and that general-

ized Turing Tests that rely on other types of artifacts

such as machines, artwork, or other non-verbal com-

positions are not possible, or at very least, are not of

equal validity to the classical form of the Turing Test.

Furthermore, it can be argued that because nature is

not a written text, then we can safely prohibit the ap-

plication of Turing’s logic to aspects of nature, spar-

ing one from potentially embarrassing consequences,

such as seeing design in the cosmos.

We have already brieﬂy mentioned that this re-

striction has all the appearances of special pleading.

If verbal artifacts in isolation can carry sufﬁcient in-

formation to allow for agent discrimination, why can’t

other artifacts? Why not machine code or machine de-

signs? We can recognize the work of genius in a su-

perbly constructed machine or an emotionally mov-

ing portrait as readily as we can in written text. It

is difﬁcult to argue otherwise. For example, if we

encountered a planet where non-verbal beings were

busily constructing skyscrapers, vehicles, and sculp-

tures, would we hesitate to ascribe intelligence to the

beings simply because they could not speak? Do we

normally accuse the mute of not possessing minds?

Words, while possibly sufﬁcient, do not appear to

be necessary to convey the presence of an intelligent

mind.

Nature is replete with awe-inspiring marvels

of engineering, containing self-adapting and self-

replicating code systems, all following the math-

ematically elegant (and surprisingly comprehensi-

ble (Wigner, 1995)) forces of physical law. How-

ever, even ignoring the apparent genius of natural ma-

chines and systems, our objection is defeated by the

presence of an actual textual system in biology it-

self: DNA (Hood and Galas, 2003). DNA forms the

basis for a chemical language, which is transcribed,

read, and directs the construction of objects within

the cell. The four bases, A-G-C-T, act as chemical

letters, forming “words” that convey instructions and

regulate processes. Thus, if strings of text are suf-

ﬁcient for conveying intelligence, then the language

objection cannot exempt biological artifacts from the

application of intelligence detection methods.

5.2 Known Origin of Biological Systems

One might also argue that since we know how biologi-

cal systems were produced, namely through unguided

evolution over millions of years, then we can inde-

pendently rule out any “false positives” caused by the

potential application of Turing-like tests to the natural

world.

Such an argument fails; if knowing the mechani-

cal causes of an artifact rules out the possibility that

it was produced by an intelligent mind, then comput-

ers could never be judged as intelligent. We know,

down to the transistor and logic-gate level, the me-

chanical workings of computer systems. Any artifact

produced by a modern computer can be traced back

to exact sequences of binary pulses and voltage oscil-

lations in logic gates. Thus, if the objection were to

hold, computers cannot in principle be ruled intelli-

gent, since any hint of design in their artifacts would

be immediately overruled by our knowledge of their

mechanical origins, turning all such inferences into

false positives. This objection applies to intelligent

design and the Turing Test with equal force.

5.3 Objection of the Unknown Agent

It might be further argued that we only recognize de-

sign and intelligence by analogy to human capabilities

and that we can recognize purported artiﬁcial intelli-

gence only as far as it parallels human intelligence.

Because artiﬁcial intelligence developed by man will

be constructed with the explicit goal of mimicking

man, we have some hope of being able to detect it.

However, were the intelligent mind extraterrestrial,

immaterial (e.g., a spirit), or otherwise wholly un-

known, then we may not be able to detect design in

its artifacts or infer its intelligence. Thus, one might

maintain that the Turing Test can reliably detect cer-

tain types of intelligence, namely agents of man-made

origin (human or computer) that are sufﬁciently hu-

man or human-like, but that intelligent design meth-

ods cannot reliably detect the actions of non-human

intelligent agents in nature, precisely because they are

ICAART 2016 - 8th International Conference on Agents and Artiﬁcial Intelligence

520

not human-like. I will call this the objection of the un-

known agent.

This is superﬁcially plausible as an objection,

since it is true that an unknown mind might not dis-

play its intelligence in ways that are humanly perceiv-

able or detectable by our tests. However, it might;

being unknown, we cannot rule out this possibility.

Furthermore, if we had independent evidence that the

mind in question did in fact generate artifacts similar

to those produced by humans, then the force of the ob-

jection would vanish. Even barring such independent

evidence, the degree to which the unknown actor pro-

duced artifacts similar to those produced by humans is

the degree to which our tests might (correctly) detect

its intelligence. Thus, in regards to intelligent design

in nature, we have an empirical question: how simi-

lar are the artifacts of nature to human artifacts? The

fact that springs (Shin and Tam, 2007), gears (Bur-

rows and Sutton, 2013), compasses (Qin et al., 2015),

boolean logic networks (Robinson, 2006), digital

codes (Hood and Galas, 2003) and other human in-

ventions have been found to preexist in biology at

least suggests the possibility that a mind behind natu-

ral phenomena would be sufﬁciently similar to human

intelligence to allow for its detection through observa-

tion of its engineered artifacts.

Lastly, the objection, at its core, concerns false

negatives, not false positives. The objection is basi-

cally that an unknown agent may escape detection,

not that an unknowing system might be mistaken as

intelligent. Given the danger of falsely attributing in-

telligence to unintelligent systems, this is exactly the

form of bias we want. Furthermore, if we did detect

an unknown intelligence behind a set of objects, the

detection itself would be evidence in favor of it be-

ing sufﬁciently similar to human intelligence, since

such tests only output positive classiﬁcations when ar-

tifacts are sufﬁciently similar to those anticipated by

humans.

Therefore, the possibility of unknown agents does

not create sufﬁcient separation between the Turing

Test and other methods of inferring design to warrant

any distinction.

5.4 The Interrogation Objection

Lastly, there remains a seemingly powerful argument

that can be raised against the full equivalence of the

Turing Test to other design detection methods: the

call-and-response nature of the Turing Test seems ab-

sent from other methodologies, and thus might be

used to validate one while invalidating the other. The

Turing Test presupposes that one can interrogate the

subject in question, gathering speciﬁc responses to

speciﬁc questions, whereas one cannot demand spe-

ciﬁc answers from nature.

To ensure this argument does not point to a dis-

tinction without a difference, we must specify exactly

what it is about the call-and-response that makes it

uniquely suitable to convey the existence of intelli-

gence, and ensure that this same quality cannot be at-

tained by other means. One such feature could be that

in responding to posed questions, a system is forced to

respond to and overcome unanticipated or novel chal-

lenges. However, the same can be said of biological

systems, which have shown adaptive architectures for

overcoming a variety of ﬂuctuating and novel envi-

ronmental challenges. When viewed from this per-

spective, we have systems which are given input con-

ﬁgurations (e.g., questions from a judge or environ-

mental structure in nature) and create response conﬁg-

urations that convey intelligence (e.g., an apropos re-

sponse or clever environmental adaptation). Further-

more, nature can be probed for speciﬁc answers in the

form of experiments and tested hypotheses, thus mim-

icking, albeit imperfectly,the call-and-response struc-

ture of the Turing Test. Thus, this distinction might

not present any real difference.

Examining this critique in greater depth, we see

that even given the call-and-response structure of

the Turing Test, one cannot fully control the re-

sponses given, thus lessening the force of this objec-

tion. While an interrogator is free to ask the system

in a Turing Test any question, he cannot be guaran-

teed that the response will be sufﬁcient or even rel-

evant. For example, consider the responses of Eu-

gene Goostman, a computer system claimed to have

passed the Turing Test by fooling a small set of human

judges into thinking it was human (Auerbach, 2014).

The following excerpts were taken from transcripts of

conversations held with Goostman (Auerbach, 2014):

Judge: Why do birds suddenly appear?

Eugene: Just because 2 plus 2 is 5! By the way,

what’s your occupation? I mean - could

you tell me about your work?

...

Judge: It is ok, I get sick of sick people. How

is your stomach feeling today? Is it upset

maybe?

Eugene: I think you can’t smile at all. I bet you

work in a funeral agency.

Clearly, the output of such a system may have lit-

tle or no correlation to the input questions. When the

responses are unrelated to the questions asked, we are

essentially given an arbitrary collection of sentences,

having no say in their contents, similar to the situa-

tion in nature. However, even such a rigid collection

Detecting Intelligence - The Turing Test and Other Design Detection Methodologies

521

might persuade judges, if it displayed empirical signs

of intelligent thinking. A one-way communication

channel, such as a mayday signal or SETI-received

broadcast from an alien civilization, may still contain

sufﬁcient information to conveyan intelligent agent at

work. It is difﬁcult to argue that only two-way com-

munication can convey intelligence when such coun-

terexamples are easily found.

However, if the interrogative nature of the Tur-

ing Test is a necessary aspect for intelligence detec-

tion and there exist no other means of attaining the

same property, then we would be forced to demote

full equivalence to a mere uncanny similarity.

6 STRUCTURED ARGUMENTS

Having stated the main arguments and some possi-

ble objections, we will now review the arguments in

a more structured manner. Let P denote premises, C

conclusions and R corollaries.

1. The ﬁrst argument can be stated as follows:

P1. The Turing Test procedure is based solely on

information contained in observable artifacts.

P2. That such artifacts could reliably indicate the

causal action of an intelligent agent entails that:

A. The causal powers of intelligent agents differ

from those of unintelligent systems in empiri-

cally detectable ways; and

B. Signals of these differing causal powerscan be

reliably contained in physical objects, absent

of knowledge of their origins or characteristics

of their generative systems.

P3. Nature contains physical objects of unknown

origin.

C1. From (P1)-(P3), it is possible that natural phys-

ical objects may reliably indicate the presence

of an intelligent cause in empirically detectable

ways.

R1. Reliable signals of an intelligent cause may ex-

ist in natural objects such as biological systems

or cosmological structures.

2. The second argument can be stated as:

P4. The Turing Test is a reliable test for detecting

the activity of intelligent agents.

P5. The Turing Test is scientiﬁcally legitimate.

P6. The primary objective of intelligent design the-

ory is to develop reliable tests for detecting the

activity of intelligent agents.

C2. (P4) and (P5) together entail that developing re-

liable tests for detecting the activity of intelli-

gent agents can form a scientiﬁcally legitimate

pursuit.

C3. From (P6) and (C2), the primary objective of

intelligent design theory might represent a sci-

entiﬁcally legitimate pursuit.

R2. From (P4) and (P5), there exists at least one re-

liable, scientiﬁcally legitimate test for detecting

the activity of intelligent agents.

R3. Design inferences in general cannot be unsci-

entiﬁc.

3. The third argument can be stated as:

P7. Successfully passing the Turing Test indicates

the probable existence of an intelligent cause.

P8. Passing the Turing Test requires that physical

artifacts be repeatedly mistaken for the prod-

ucts of an intelligent cause.

P9. Natural cosmological and biological objects

have repeatedly been mistaken for the products

of an intelligent cause.

C4. From (P8) and (P9), the cause of natural cosmo-

logical and biological objects has successfully

passed the Turing Test.

C5. From (P7) and (C4), the cause of natural cos-

mological and biological objects is probably in-

telligent.

7 CONCLUSION

Where does this leave us? For the Turing Test to

work, one must be able to distinguish intelligent

causes from unintelligent causes based solely on ob-

servable artifacts. But this leads to the conclusion

that intelligent design cannot be simultaneously dis-

regarded, since its methodological structure rests on

the same foundation. Furthermore, if the Turing Test

is a reliable procedure for detecting intelligence, then

the cause of biological origins is likely an intelli-

gent mind, having passed a generalized Turing Test

since the time of Cicero (Cicero, 45BC). Appealing

to strong materialism, known mechanistic origins and

unknown agents does nothing to remedy the situation.

This forces us to either reject the reliability of the Tur-

ing Test and its underlying logic, or face what many

would consider a creeping creationism.

To allow for the possibility of intelligent design

would be to deny a scientiﬁc consensus, and we are

often reminded that true scientists and intelligent lay-

men cannot deny any fact established by the consen-

sus of experts. To do so is derided as a strong form

denialism, for suggesting the majority of scientists

might be wrong on a well-studied issue. Surely such

could not be the case (Ioannidis, 2005; Baker, 2015).

ICAART 2016 - 8th International Conference on Agents and Artiﬁcial Intelligence

522

REFERENCES

Auerbach, D. (2014). A computer program ﬁnally passed

the Turing Test? slate.com/articles/technology/

bitwise/2014/06/turing

test reading university did

eugene oostman ﬁnally make the grade.html/.

Accessed: 2015-11-10.

Baker, M. (2015). First results from psychology’s largest

reproducibility test. http://bit.ly/1bk6qDF. Accessed:

2015-05-13, nature.com/news.

Burrows, M. and Sutton, G. (2013). Interacting gears syn-

chronize propulsive leg movements in a jumping in-

sect. Science, 341(6151):1254–1256.

Cicero, M. T. (45BC). Of the Nature of the Gods. Book II,

chapters XXXVII, XLIV, and XLVII.

Copeland, B. J. (2012). Turing: pioneer of the information

age. Oxford University Press.

Cover, T. M. and Thomas, J. A. (2006). Elements of infor-

mation theory. Wiley & Sons. Section 2.8.

Egnor, M. (2014). Steven Novella Doesn’t Trust His

Computer One Bit. evolutionnews.org/2014/12/

steven

novella092181.html. Accessed: 2015-05-11.

French, R. M. (2000). The Turing Test: the ﬁrst 50 years.

Trends in cognitive sciences, 4(3):115–122.

Hood, L. and Galas, D. (2003). The digital code of DNA.

Nature, 421(6921):444–448.

Ioannidis, J. P. (2005). Why most published research ﬁnd-

ings are false. PLoS medicine, 2(8):e124.

Kurzweil, R., Gilder, G. F., and Richards, J. W. (2002). Are

We Spiritual Machines?: Ray Kurzweil vs. the Critics

of Strong AI. Discovery Institute.

Larson, E. J. (2015). Group Delusions Aside, Sentient

Robots Aren’t on the Way. evolutionnews.org/2015/

03/sorry

nature gr094741.html. Accessed: 2015-05-

11.

Qin, S., Yin, H., Yang, C., Dou, Y., Liu, Z., Zhang, P., Yu,

H., Huang, Y., Feng, J., Hao, J., Hao, J., Deng, L.,

Yan, X., Dong, X., Zhao, Z., Jiang, T., Wang, H.-W.,

Luo, S.-J., and Xie, C. (2015). A magnetic protein

biocompass. Nat Mater, advance online publication.

Article.

Richards, J. (2011). I, for one, welcome our new robot

overlords. aei.org/publication/i-for-one-welcome-our-

new-robot-overlords/. Accessed: 2015-05-19.

Robinson, R. (2006). Mutations change the boolean logic

of gene regulation. PLoS biology, 4(4).

Saygin, A. P., Cicekli, I., and Akman, V. (2003). Turing

test: 50 years later. In The Turing Test, pages 23–78.

Springer.

Shin, J. H. and Tam, B. (2007). Unleashing of a biologi-

cal spring. In Magjarevic, R. and Nagel, J., editors,

World Congress on Medical Physics and Biomedical

Engineering 2006, volume 14 of IFMBE Proceedings,

pages 2824–2827. Springer Berlin Heidelberg.

Smith, W. J. (2015). Machines Will Always Be

Things, Never “Persons”. evolutionnews.org/2015/04/

machines

will a095201.html. Accessed: 2015-05-11.

Smolin, L. (2004). Cosmological natural selection as the

explanation for the complexity of the universe. Phys-

ica A: Statistical Mechanics and its Applications,

340(4):705–713.

Torley, V. (2014). Why a simulated brain is not conscious.

http://www.uncommondescent.com/intelligent-design/

why-a-simulated-brain-is-not-conscious/. Accessed:

2015-05-19.

Turing, A. M. (1950). Computing machinery and intelli-

gence. Mind, pages 433–460.

Wigner, E. P. (1995). The unreasonable effectiveness of

mathematics in the natural sciences. In Philosophical

Reﬂections and Syntheses, pages 534–549. Springer.

Detecting Intelligence - The Turing Test and Other Design Detection Methodologies

523