Just because you made a
machine “smart” doesn’t mean you made it intelligent.
Smart homes, IoT, AI, ubicomp, robotics, ambient media, pervasive
computing – all are words that in one form or another describe our attempt to
embed digital intelligence into the otherwise dumb, lifeless devices around us.
However, as we engineers so often do, many of us have been all-consumed by the
pursuit of a single focus at the expense of other equally, if not more
important factors. The fields of IoT and AI have been racing ahead trying to
get some basic functionality in place, yet in doing so, we’ve created another
problem. We’ve been building greater and greater smarts and functionality, but
in the process, created smart devices that are a bunch of assholes.
What we neglect in the pursuit of intelligence is that there is more
than one type. Our current efforts have understandably chased after the more
obvious, functional definition of intelligence, however, this narrow
perspective on the word has created a bunch of basic gadgetry that no one wants
to interact with. We’re actively building the electromechanical equivalent of
genius savants – all IQ with no EQ.
And in fact, even that statement may be a massive overgeneralization.
Human intelligence is potentially comprised of countless aspects. While there
are many ways we can deconstruct these intelligences, all of which are
incorrect or limiting in one way or another, for the sake of exploration in the
digital sphere, let’s pick one and play.
Howard Gardener’s 1983 book, Frames
of Mind: The Theory of Multiple Intelligences, is one of the more popular
perspectives on the mind which views intelligence as a set of differentiated
modalities, rather than as a single, general trait or ability. Though often
criticized for blurring the line between innate smarts and developed ability,
Gardener’s theory breaks down intelligence into 9 distinct, though somewhat
interrelated types: logical-mathematical, linguistic, bodily, musical, naturalist,
intra-personal, spatial, and existential.
If we hold these 9 as an expanded definition of the word intelligence,
we can compare the technology of today against this expanded standard to see
how smart our devices really are.
Naturalistic
Our Natural Intelligence defines our ability to recognize and interact
with the natural world around us - plants, animals, rocks, clouds, etc. Though
not added until nearly a decade after publishing, Gardener described this
intelligence as core to our evolutionary success as hunters, gatherers, and
farmers.
Developmentally, this intelligence should be marked by little more than
high-resolution vision and olfactory systems coupled with an extremely in-depth
natural database to analyze against. In theory, this seems like a less
important ability for machines to have, however, if we are to rely on them as
tools for our need, you’ll likely want your smartphone not to accidentally
serve you poison-ivy tea. Moreover, even a robot will likely suffer from
mistaking a jaguar for a house cat, so there could be some important survival
aspects here.
Current
Artificial Level: City-Slicker
Leading Example:
Bosch Bonirob
Implication:
Device may be destroyed by nature or angry user.
Bodily-Kinesthetic
Our body-smarts describe our ability to
leverage our body and manipulate objects effectively around us. This
intelligence relies greatly on a good sense of timing, perception, and control,
enabling the perfection of fine or gross motor skills.
Plainly put, machines typically suck at this
one. While a device can be programmed to perfectly optimize a single motion,
such as in manufacturing environments, the moment any variability is
introduced, all hell breaks loose. However, body intelligence is further
complicated by the fact that devices are often build modular and adaptable; the
equivalent of being caught in a permanent state of robotic puberty, where audio
drivers drop a few octaves and new wires start growing where there weren’t any
before.
Current
Artificial Level: Drunken Toddler
Leading Example:
Boston Dynamics Atlas
Implication:
One wrong step and a half-million dollar piece of equipment becomes an
impressive paperweight.
Spatial
Spatial intelligence is a bit of a tricky one. At a glance, it is
described as little more than the ability to think in three dimensions,
however, the implications of this get a bit more interesting. Because of the
human capacity to speculate, spatial intelligence effectively ties into our
imagination and creativity, allowing us to perceive hypothetical versions of
the world around us.
Can a computer think in three dimensions? Of course – it can likely
think in four or five. However, very few machines have taken the next leap into
projecting new thoughts and hypotheses into this three-dimensional world,
meaning that as it stands, our best machines are little more than environment
rendering software.
Current Artificial Level: Following the Lego Instructions
Leading
Example: Autodesk Inventor
Implication: Cold-war era, brutalist design and architecture.
Logical-mathematical
This intelligence represents our ability to
calculate, quantify, consider propositions, and carry out logical operations.
Beyond simply the manipulation of numbers, this intelligence has a heavy
implication on pattern recognition and the ability to develop effective
strategies.
This is the place where computers are
currently kicking our asses. While some people may argue that machines still
trail humans in pattern recognition and strategy development, anyone who has
suffered the embarrassment of playing an RTS on “Insane” mode knows this to be
a fallacy.
Current Artificial Level: Rain Man
Leading Example:
Google AlphaGo
Implication: Younger generations not bothering to learn math and an
impending enslavement of the human race.
Musical
Our musical intelligence – obviously –
pertains to aspects such as pitch, rhythm, and timbre. However, beyond the
simple identification of music, this creative intelligence allows us the
ability to recognize, reflect upon, and even create our own musical patterns.
While the audio processing half of this
equation seems to be little to no problem at all, once again, the greater
creative aspect is where things become a challenge. That said, by analyzing
historic data points of music, we have been able to create quite sophisticated
music generation software. However, to date the music created by these programs
falls into one of two categories: blatant rip off or painfully unlistenable.
Current Artificial Level: Starving Singer/Songwriter
Leading
Example: Melomix109
Implication: We will continue to listen to top 40 songs produced by
human pop-machines.
Linguistic
Our word smarts represent our ability to
understand and express through words and language. However, even beyond this,
Linguistic intelligence helps us to see broader context and apply
meta-linguistic skills to understand the deeper meaning of language and
communication; reading between the lines.
In theory, the artificial version if this
intelligence is simply an effective natural language processing unit. The
problem is that no one has built a good one yet, since human communication is
fraught with subtle nuance, sarcasm, context-sensitivity, and lies. Basic
linguistic intelligence is simply understanding the words – a task nearly
cracked by the world of computational linguistics – but understanding the words
behind the words is a feat that requires an understanding of who those words
are coming from, which leads us to our next point…
Current Artificial Level: Super-Intelligent Parrot
Leading
Example: Bruce Wilcox’s “Rose” Chatbot
Implication: Automated customer support lines will cause you to throw
your phone across the room for at least another decade.
Interpersonal
People smarts define our capacity to
understand and interact with others without making things terribly awkward.
This is the empathy category hinges off the ability for someone to interpret
all types of verbal and nonverbal communication and be sensitive to the
individuals around them.
We’re now getting into the territory where
our technology simply sucks. Our current ability for technology to pick up on
human context and emotion is only made look less pathetic by the ability for
that technology to react to it. We’ve devoted entire fields – UX, HCI, and
others – to attempting to improve this form of artificial intelligence, but to
date, no one has really made a device that at some point doesn’t deserve to
become airborne.
Current Artificial Level: Autistic Teenager
Leading
Example: RoboThespian
Implication: No robot girlfriends just yet.
Intrapersonal
Intrapersonal intelligence describes our
ability to look within and deconstruct ourselves. It is crucial to be able to
identify and understand our own emotions, as well as to put these within a
broader context of what it means to our lives and our futures.
This intelligence is practically non-existent
within the technology world, however, not for lack of success, but lack of even
trying. To date, few people have seen the value in a device that can be
introspective, however, in coming years, this type of intelligence may be the
key to unlocking broader interpersonal and existential smarts. How can we
expect to understand others and the world around us, unless we can first
understand ourselves?
Current Artificial Level: Dog Fighting its Reflection
Leading Example:
Nico Robot
Implication: The emergence of artificial psychiatrists for robots that
live in denial and refuse to acknowledge their own emotions.
Existential
Existential intelligence is likely the
highest form and deals with broader questions of the limits of life and our
existence. It highlights the part of our brains that allows us to think beyond
the literal and contemplate broader concepts like the origins of life,
spirituality, and how they get the caramilk into the chocolate.
At the top of the hill sits the most complex
and elusive form of intelligence that, to our knowledge, few other species let
alone technologies, have ever cracked. While I can’t be certain that pigs don’t
think about where bacon truly comes from, I do know that it will be a long time
before robots start developing their own religion. That said, the interesting
thought exercise emerges once an AI has met and potentially surpassed this
ninth intelligence and begins to explore the realms beyond… but I digress.
Current Artificial Level: Non-Existent
Leading
Example: NA
Implication: DO robots dream of electric sheep? IS there a robo-god?