thinking about thinking

John picked up the two double Latte Grandes and walked over to the corner table
near the fireplace where Mary was setting up the chess game. She took a pawn
of each color and concealed them in her hands before offering two fists to John.
Putting the cups down, he tapped Mary’s left hand and was pleased to see the
white piece as he took his chair.

John said “Playing chess always reminds me of the game between IBMs Deep
Blue Supercomputer and the reigning World Chess Champion at the time, Garry
Kasparov.” He glanced at the board, “d4, I think,” as he moved his pawn.

Mary said, “Me too.” Mary smiled to herself as she moved her own queen’s
pawn forward to d5. She knew that John had strong feelings about the limits of
true Artificial Intelligence and she hoped to gain an advantage by baiting him.
“That was the first time a computer won a complete match against the world’s
best human player. It took almost 50 years of research in the field, but a computer
finally was thinking like a human.”

John bristled slightly, but then realized that Mary was just poking a little fun.
Taking his next move, c4, he said “You can guess my position on that subject.
The basic approach of Deep Blue was to decide on a chess move by assessing
all possible moves and responses. It could identify up to a depth of about 14
moves and value-rank the resulting game positions using an algorithm developed
in advance by a team of grand masters. Deep Blue did not think in any real
sense. It was merely computational brute force.”

Mary reached over and took John’s pawn, accepting the gambit. “You must
admit,” she replied, “although Kasparov’s ‘thought’ processes were without a
doubt something very different than Deep Blue’s, their performances were very
similar. After all, it was a team of grand masters that designed Deep Blue’s
decision-making ability to think like them.”

John played his usual Nc3, continuing the main line of the Queen’s Pawn Gambit.
“You’ve made my point,” he exclaimed, “Deep Blue did not make its own decisions before it moved. All it did was accurately execute, the very sophisticated
judgments that had been preprogrammed by the human experts.”

“Let’s look at it from another angle,” Mary said as she moved Nf6. “Much like
a computer, Kasparov’s brain used its billions of neurons to carry out hundreds
of tiny operations per second, none of which, in isolation, demonstrates intelligence. In totality, though, we call his play ‘brilliant’. Kasparov was processing
information very much like a computer does. Over the years, he had memorized
and pre-analyzed thousands of positions and strategies.”

“I disagree,” said John quickly moving e3. “Deep Blue’s behavior was merely
logic algebra—expertly and quickly calculated, I admit. However, logic established the rules between positional relationships and sets of value-data. A fundamental set of instructions allowed operations including sequencing, branching,
and recursion within an accepted hierarchy.”

Mary grimaced and held up her hands, “No lectures please.” Moving to e6 she
added, “A perfectly reasonable alternative explanation to logic methods is to
use heuristics methods, which observe and mimic the human brain. In particular, pattern recognition seems intimately related to a sequence of unique images
connected by special relationships. Heuristic methods seem as effective in producing AI as logic methods. The success of Deep Blue in chess programming is
important because it employed both logic and heuristic AI methods.”

“Now who’s lecturing,” responded John, taking Mary’s pawn with his bishop. “In
my opinion, human grandmasters do not examine 200,000,000 move sequences
per second.”

Without hesitation Mary moved c5 and said, “How do we know? Just because
human grandmasters are not aware of searching such a number of positions
doesn’t prove it. Individuals are generally unaware of what actually does go on
in their minds. Patterns in the position suggest what lines of play to look at,
and the pattern recognition processes in the human mind seem to be invisible to
the mind.”

John said, “You mean like your playing the same Queen’s Gambit Accepted line
over and over again?” as he castled.

Ignoring him, Mary moved a6 and said, “Suppose most of the chess player’s skill
actually comes from an ability to compare the current position against images of
thousands of positions already studied. We would call selecting the best position
(or image) insightful. Still, if the unconscious human version yields intelligent
results, and the explicit algorithmic Deep Blue version yields essentially the same
results, then why can’t I call Deep Blue intelligent too?”

John said, “I’m sorry, but for me you’ve overstated your case by calling Deep
Blue intelligent,” moving Qe2. He continued, “Would you like to reconsider your
position?”

Mary moved Nc3 and said, “Of course not, I still have plenty of options to think
about alone this line.”