descent from the higher primates is remi-
niscent of what is called the day–age the-
ory. This is the interpretation, by old
earth creationists (among others), of the
six days of creation in Genesis 1 as refer-
ring to ages of the earth’s history. Just as
there are six daysof creation, there are six
eonsof Earth’s history: Hadean, Archaeo-
zoic, Proterozoic, Paleozoic, Mesozoic
and Cenozoic. The problem with such an
equating of days with ages is that the last
three “days”—the Paleozoic, Mesozoic
and Cenozoic—comprised between them
a bit over half a billion years; while the
entirety of Earth’s history is about 4.5
billion years. Thus, the last three day/ages
between them comprise only 1/9 of
Earth’s history.
There is a long history of attempts to
reconcile either science with the Bible or
the Bible with science. These include
Hugh Ross explaining the resurrected
Jesus appearing suddenly in the locked
room where his disciples are hiding (John20:26) as Jesus moving through different
dimensions and numerous attempts to
find natural, scientific explanations for
the ten plagues against Egypt in Exodus.
Whether it’s the day–age compromise,
Jesus traveling through different dimen-
sions, scientific explanations for the mira-
cles of Exodus, or Swamidass’s attempt to
reconcile Adam and Eve with human
evolution, attempts to fit science into the
Bible or the scriptures into science do vi-
olence to them both.REVIEWS
56 SKEPTIC MAGAZINE volume 25 number 1 2020There are a lot of books about
artificial intelligence. The interlibrary
site Worldcat lists over 36,000. Ama-
zon claims to have over 20,000 for sale.
Many contain histrionic titles, such as
Life 3.0: Being Human in the Age of Artificial
Intelligence; You Look Like a Thing and I
Love You: How Artificial Intelligence Works,
Why It’s Making the World a Weirder Place;
and especiallyThe Age of Spiritual Ma-
chines: When Computers Exceed Human In-
telligence.Melanie Mitchell’s new book is
more modestly titled, but it is, in my
opinion after surveying much of this liter-
ature, the most intelligentbook on the
subject. Mitchell is Professor of Com-
puter Science at Portland State Univer-
sity as well as External Professor and
Co-Chair of the Science Board at the
Santa Fe Institute. And, unlike most ac-
tive practitioners in the field, her evalua-
tion of the current state of AI and its
prospects is measured, cautious, and
often skeptical.
The book begins with an introduc-
tion (“Prologue: Terrified”), a personal
story of how she became involved with
AI, inspired by Douglas Hofstadter’s
Gödel, Escher, Bach: An Eternal Golden
Braid. Through a mixture of luck, audac-ity, and persistence, Mitchell first became
Hofstadter’s research assistant and then
a doctoral student under him. Decades
later, in 2014, at a Google conference she
attended with him, she learned that Hofs-
tadter was upset that one AI program has
defeated the world chess champion and
another has generated a music “composi-
tion” indistinguishable from (even judged
better than) a genuine composition by
Chopin. Hofstader’s concerns inspired
her to write about the pursuit of human-
level AI (and beyond).
The book is divided into five parts:
Background; Looking and Learning;
Learning to Play; Artificial Intelligence
Meets Natural Language; and The Bar-
rier of Meaning. Although Mitchell
states up front that the book isn’t in-
tended to be a general survey or history
of AI, she still manages to tell enough of
its history—especially of its hubristic in-
auguration in 1956 by John McCarthy,
Marvin Minsky, Allen Newell, and Her-
bert Simon—to put today’s enthusiastic
optimism in perspective.
As Mitchell explains, one of the first
branches in the pursuit of AI was artifi-
cial neural nets (ANN)—the foundation
of today’s deep learning algorithms. Sheprovides an example of such an ANN, the
“perceptron” designed to “learn” how to
recognize hand-lettered digits. The illus-
trative grid is 18x18, and each square has
one of four shades: white, light gray, dark
gray, and black. Curiously, she doesn’t
mention the fundamental problem of
even such a relatively modest ANN: the
number of different possible inputs in this
case is 2 to the 326th power ( 2326 ), or:
136,703,170,298,938,245,273,281,389,1
94,851,335,334,573,089,430,825,777,27
6,610,662,900,622,062,449,960,995,2
01,469,573,563,940,864
Such a program is effectively untestable.
(This is true for the even more limited
8x8 binary grid, which she discusses a
little later in a different context. In that
case, the number of possible cases is
only 2 to the 64th power ( 264 ), or
18,446,744,073,709,551,616.)
This problem will come back to
haunt today’s astonishingly successful
deep learning programs, as MitchellFarrar, Strauss and Giroux,- 336 pp. $28.
ISBN 13:9780374257835
Ten Years Away...
and Always Will Be
A reviewof Ar tificial Intelligence: A Guide for Thinking
Humans by Melanie MitchellREVIEWED BY PETER KASSAN