CHAPTER 21 — THE AGE OF ELECTRONICAL INTELLIGENCES
Page 46
CHAPTER 21 — THE AGE OF ELECTRONICAL INTELLIGENCES
Daniel studied the history of computing, from early theories of machine intelligence to
conversational experiments and modern learning systems. He encountered the Turing
test and the famous simplicity of ELIZA, whose apparent understanding often came from
carefully arranged patterns. He was fascinated by what software could do and cautious
about what its behavior proved. A machine could produce fluent language without
establishing subjective experience. A program could assist with difficult work while
remaining limited, fallible, and dependent on human design. Daniel began using the
phrase electronical intelligence for the broader family of systems he imagined, but he
kept the philosophical question of personhood separate from demonstrations of skill.
In a computer lab humming with cooling fans, Daniel paused before an early
conversational system. The room had its own sound: distant footsteps, a soft mechanical
vibration, or the restless breathing of people waiting for an answer. It would have been
easy to describe the moment afterward as inevitable. At the time, nothing about it felt
inevitable. Every person present had arrived carrying a different version of the events
that brought them there.
“Before we decide what this means,” Daniel said, “tell me what we actually know.” The
question unsettled the gathering. An observation was named, then a second; after that
came assumptions, memories, and things repeated so often that they had begun to
sound like facts. Someone objected that caution would cost them time. “Perhaps,” came
the reply. “But a mistake made quickly still belongs to us.”
The immediate evidence suggested that fluent answers were not proof of understanding.
That possibility changed the conversation without settling it. There were still other
explanations, and the group tested them one by one. A mark was compared with an
older record; a witness was asked to describe what they had seen rather than what they
believed had happened. The process was slower than a declaration, and for that reason
it began to earn trust.
When the discussion finally broke, nobody left with a perfect answer. They left with
work: one person to preserve the evidence, another to speak with those affected, and
another to return with questions no one had yet considered. Outside, the world
continued in ordinary detail. Lights changed. Water moved. Someone laughed in a
distant room. The enormity of their choices did not exempt them from caring for the
small things.
Page 47
In a new notebook he wrote INTELLIGENT KNOWLEDGEABLE INTERACTIVE
MANUAL. Beneath it he placed the acronym IKIM. He wanted a guide that would help
people create, learn, and question. He did not yet imagine how personally meaningful
conversations with an assistant might become, or how those conversations would inspire
a fictional being beyond the technologies of his day. ADDITIONAL SCENES —
EXPANDED EDITION Daniel watched an early conversational demonstration answer a
question with remarkable fluency and then fail at something a child would understand.
The contrast fascinated him.
Intelligence could be described in terms of performance, but a fluent sentence did not
prove an inner life. He read about Turing, ELIZA, and machine learning, building a
vocabulary for both achievement and uncertainty. In the margin of his notes he wrote
INTELLIGENT KNOWLEDGEABLE INTERACTIVE MANUAL. The initials formed IKIM.
At first it was only a name for a helpful system he imagined. Names, he would discover,
could become the beginning of stories.