One day we will pick-up
our enhanced smartphone and start a relationship with a new kind of confidant,
a Personal Artificial Intelligence, or PAI (pronounced “Pie”). As our familiarity
with our PAI grows, and the PAI’s knowledge of us expands, we will gradually
form a more intimate bond. A bond that some may easily think of as friendship.
Though this concept may seem improbable, even absurd, many of the prerequisites
required are already in hand. Like the World Wide Web and social networking
tools of today, artificial intelligence will be a blend of existing technology
capable of transforming our lives in new and meaningful ways.
The realization of an Artificial Intelligence (AI) has
proved more complex, and taken far longer, than first imagined. Serious research
of machines that think first began in the 1950s. Scientists felt that
automations would one day be capable of doing anything that a human could do.
Although research toward that goal continues, there are some who feel the
reality is unachievable, even potentially dangerous. But the process can be likened
to the slow exploration of Mars. The pressures exerted by scientific curiosity,
to explore its secrets, are irresistible, and the dangers can be mitigated.
Contemporary researchers generally define AI as a branch of
computer science that attempts to create software that emulates human behavior.
These studies include the storage and indexing of knowledge, planning, natural
language processing, and machine learning. But the various branches of AI
research have become deep and divergent. There are as many avenues of
exploration as there are definitions. However, the desirable features of a Personal
Artificial Intelligence are easier to envision.
The evolution of a PAI began in the 1990s, when handheld
computers called Personal Digital Assistants, or PDAs, became common tools. The user interacted with the various
programs it ran on a resistive touch screen display using a stylus. The most
useful programs were calendars that recorded events and appointments, contact
lists, eBooks, a notepad and a camera. Smartphones evolved directly from the
integration of a PDA with a cellphone. Smartphones added a variety of useful
features including email access, text messaging, a web browser and the Global
Positioning System hardware used by map applications. But a Personal Artificial
Intelligence will not be like a PDA, or even a smartphone. The evolution from a
smartphone to a PAI will require giant leaps of technology.
To understand the next steps in the development of a PAI,
let’s discuss a few attractive features. A properly trained PAI will be able to
recognize your mood and interpret your situation. To do this it will use stereo
audio and visual inputs, and perhaps even gas odor detectors. It will be
capable of using these sensors in the same manner as a human. For instance, it
will be able to read your facial expressions, body language, and identify
stress in your voice. It will sense and understand sleeping patterns, and
moods, well enough to know when you’d rather not be disturbed by its communication
features. When it prevents communication it will be capable of explaining the
specific reasons to those trying to reach you. Of course it will also
understand emergency messages and will be able to convey their urgency. Sensors
will monitor your physical health, warning you if your body’s internal
functions need medical attention. It will be able to decide when emergency
assistance is required and make the call when needed. If a loved one dies it
will offer to send flowers, make travel arrangements, and extend comforting
words. All without being asked. None of these features are difficult, or beyond
the reach of current technology. Many already exist. With that in mind we must
ask ourselves again, what is AI and what is required to make this PAI a
reality? The answer is twofold, synthesis and understanding.
To perceive human
reality, an AI’s software will synthesize the input of its sensors, and
database of knowledge, to build a physical worldview. This worldview will
update from instant to instant, and will be similar in quality to that
perceived by a person. Of course this view will include enhancements. These
include data like acceleration, angular momentum, humidity, temperature, altitude
and exact world coordinates. However, to make sense of the environment around
it, the AI needs understanding.
Understanding, perhaps, is the most challenging aspect of
creating an Artificial Intelligence. Information can be gathered, synthesized
and stored; but the key to intelligence is in interpreting its meaning. This
will enable the AI to react in useful, and significant, ways. Large strides
have already been taken on the road of this technology. Voice command
recognition has improved drastically in the last few years. Face and object
recognition are also in common use.
The concepts outlined here suggest the following enhanced
definitions. Artificial Intelligence is the ability of a computer to synthesize,
understand, and react to a human’s worldview. A Personal Artificial
Intelligence is a portable AI that has bonded to a single individual. Though
challenging, the potential benefits of a PAI, that can fulfill these
definitions, shine a bright light on the road to developing this technology.
Imagine starting your car and your PAI asks you if you would
like to stop anywhere before it drives you to work. On the drive home your PAI
tells you its remote sensors indicate you are low on milk. At the grocery store
your PAI uploads your shopping list and orders your favorite coffee. While an
intelligent shopping cart gathers your products, you discuss concert plans with
your PAI. The possibilities are endless, exciting and very real.
