by Mark Ollig
The folks at the Defense Advanced Research Projects
Agency (DARPA) recently made public a proposal asking for specific models on
how humans can naturally communicate with computers.
This investigative columnist read with interest,
DARPA’s: Communicating with Computers (CwC), 57-page document.
Dated Feb. 19, 2015, and titled: DARPA-BAA-15-18
Communicating with Computers, this document describes the CwC specifications.
DARPA desires proposals demonstrating how people can
engage in two-way interaction with computers using hand gestures, eye
movements, facial and other expressions; in addition to language understanding.
“The CwC program seeks to develop technology to
facilitate communication between humans and machines, which we take to mean,
technology for assembling complex ideas from elementary ones given language and
context,” says a sentence on page 9 of the document.
There are certain CwC program tasks and case examples
within this document, outlining specific human-machine interactive scenarios.
Some situations call for a collaborative composition
involving machines and humans contributing to the creation of a poem, drawing,
and story-telling.
An example of a human-machine combined collaboration is
when the human would start a story with “Mother cried at her window.” The next
sentence would be created by the machine, which would analyze the human’s
words, and come up with an original idea of why the mother would be crying, and
create a fitting sentence to carry on the story.
The machine would comprehend and build upon additional
sentences, in order to come up with new concepts for continuing the story;
using its own reasoning algorithms.
I wish to inform my faithful readers; this column is
not being written by a machine.
The CwC document suggests evolving intelligence and
understanding, acquired by computing machines, will be analogous to how a
growing child learns from its parents or guardians.
It is hoped, by having computers easily understand
human conversation, and how we interact, that they will be better able to
contribute meaningful responses for solving problems, and accomplishing tasks.
“The sole purpose of the CwC evaluation is to steer the
program toward technology that gives computers human-like communicative
abilities,” states the document.
What will be needed are advancements in algorithmic
software coding techniques, so computers (including robots) can “evolve” into
reliable, human-assisting associates, and idea-contributing machines; rather
than being the static receiving device for input commands.
Well, there we are.
You see, everyone; once again, this all goes back to
“Star Trek.”
Yes, friends, let us recall the original series “Star
Trek,” episode 53: “The Ultimate Computer,” which aired during March of 1968.
Testing begins of an intelligent and autonomous
computing machine, called the M-5 Multitronic System.
The M-5 is physically wired into the USS Enterprise
main control computer.
The M-5 starts out innocently enough, effectively
performing simple, space-maneuvering operations of the USS Enterprise; without
human navigation assistance.
It even successfully defends the Enterprise, and
defeats other starships during a mock space battle; again, all without human
intervention.
The human crew onboard the Enterprise are both
impressed and somewhat worried about losing their jobs to a machine.
The M-5 continues to learn, and its autonomous
intellect develops to the point where it takes total control of the Enterprise.
No longer listening to the human crew, it decides on
its own which of the ship’s systems should be active or inactive.
The M-5 then seeks out and begins destroying other
space ships in the area it feels are a threat to its existence – which ends up
being all of them.
Of course, the humans onboard the Enterprise are
frantically trying to regain control of the ship. I can still hear Dr. McCoy
frustratingly say, “Fantastic machine, the M-5: no off-switch.”
In order to save the day, Captain Kirk establishes a
verbal communication tie-in with the M-5.
Using his amazing tact, and above-average reasoning
prowess, Kirk convinces the M-5 computer to feel guilty about what it has done.
The M-5 computer then shuts itself off, which allows
the humans time to “pull the plug” out of the “unconscious” computing machine;
thus disconnecting it from the Enterprise, and allowing the humans to regain
control of the ship.
A suggestion to DARPA: make sure future intelligent
machines come equipped with an emergency off-switch; just in case they try to
take over the planet from the humans.
Machines with a bit of compassion, humor, and patience
would be a nice adder, too.
Have we now arrived at the starting gate of when humans
begin creating autonomous machines capable of independent thought, human-like
conversation, original concepts, and unique problem-solving solutions?
Are we emotionally ready for computing devices and
machines which can think and communicate with us, as easily as you and I do
with other people?
Will humans ultimately end up creating artificially
intelligent machines which will become our new co-workers, bosses, friends, enemies,
soldiers, teachers, protectors, space explorers, and personal companions to
keep us company, and to assist us in our old age?
Stay tuned.
DARPA’s Broad Agency Announcement CwC (Adobe Acrobat
PDF file), is located at: http://tinyurl.com/nt4l3zk.
