TERMINATOR GENISYS T-800
Picture source: http://scifidaily.ru/2015/10/04/32145-prodolzheniya-terminatora-5-genezis-zamorozheny/
The Terminator movie
depicted the takeover of humans by machines that gained self-awareness. The
latter developed human characteristics and began to fight humanity for control
over the Earth. If such were the reality on Earth today, what is the likelihood
of us being able to co-exist together with safety?
Based on McCarthy’s (1998) Definition:
The science and engineering of
making intelligent machines that are able to display intelligent behaviour that
charaterises themselves and allows them to perceive their surrounding
environment. This is sometimes seen as a form of sentience and sapience
(Bostrom, 2011).
How Safe Can AI Be?
From the definition alone, it can
be concluded that the probability of a smart machine having been built with
good intentions is very likely. It is likely that the purpose of AI was
initially to make like more efficient for the everyday user of AI, which in
time was the aim of such a technology.
However, some risks of AI have been
proposed, causing there to be a controversy surrounding the further development
of AI. This has caused many researchers and well-known philosophers to be
afraid of AI.
Three Perspectives:
Hawking (2014), in an interview
with the BBC, claimed that AI could be the determinant factor to the end of the
human race. He went on to convey the threat of AI as something that would
surpass the slow pace of biological evolution of humans, thus taking over from
the humans as the dominant race running the planet. Hence, “humans, who are
limited by slow biological evolution, couldn't compete, and would be
superseded" in relation to AI (Hawking, 2014), such that AI will possibly
destroy humanity within a century (Cellan-Jones, 2014).
Musk (2014), in support of
Hawking's perspective that AI is dangerous if it is not kept under tight
controls, offered a sum of £10million
to projects designed to aid the control of the technology. Musk further added
that the rise of AI would be equivalent to "summoning the demon" and
turning humans into pets of a super-intelligent race of computers that we
helped to create. He also predicted that AI would be more threatening than
Nukes.
Kurzweil (2015), however, suggested
an evolution from the bio-AI perspective, that “hybrid humans” will be born
from integrated technology for space colonisation, giving birth to a race of
humans who possess AI technology, and perhaps a complete AI revolution i.e. AI
takes over from humans.
But, Is It Worth All The Worry?
Both Hawking (2014) and Musk (2014)
have good points about the risk of AI taking over as the "superior
race" from humans, especially as they can improve their intelligence at a
much faster rate than humans can. The technology they can build as technology
can potentially far supersede anything we know
and understand today. For example, not only can AI outsmart the financial
market, and probably guess lottery numbers before they come out by figuring out
mathematical and statistical possibilities, but they could also develop weapons
of mass destruction that we humans may fail to understand.
However, from the prospect of
Bio-AI development, the view that AI will take over from the “hybrid humans”
that aim to replace us (Kurzweil, 2015) may not be so plausible because
technically humans already exist with AI, from those who have bionic arms to
clockwork hearts for medical reasons. Therefore, this argument may be refuted.
Extraordinary Inventions Carry Extraordinary Risks:
One cannot deny the fact that AI is
an extraordinary piece of technology, with "intelligence", suggesting
a personified "smartness" to the technology which allows it to
"think" based on a fixed set of imperatives. In the short-term, AI is
dependent on its controllers, but in the long-term, it can become a fully
self-sufficient, self-maintaining machine that does not rely on human
coordination to function, hence the reason it is deemed to be
"intelligent" as such (Yudkowsky et al., 2010).
However, There Are Benefits To This Kind of Technology:
A new version of Cortana is
currently being developed as a super-intelligent virtual assistant. The
intention is to make people's lives more efficient and not to waste so much
time searching for the information one may need at any given time.
RoboSimian is also known as the
"mechanical monkey" due to its specific purpose and design, which was
intended to venture out into dangerous zones in disaster areas to aid search
and rescue teams in order for humans to stay out of danger themselves.
Medical robots are being made to
assist in surgery in operating theatres to help surgeons to conduct surgery in
a more precise manner, making surgery more accurate and efficient.
Andrew Ng's Perspective on Intelligence vs Sentience
Ng (2014), a chief scientist for the well-known Chinese e-commerce site, Baidu, stated that the difference between intelligence and sentience was very significant. He argued the case for an increase in intelligence not necessarily leading to an assumed increase in sentience.
What's the Logical Approach, You May Ask?
It can be said that the logical
approach to this argument revolves around looking into what we are made of. Our
intelligence, often associated with our brains, is sourced by a neuron system
within us that has evolved in our DNAs over millions, or even billions of
years. The brain and its intelligence, in that sense, is essentially the most
complicated and sophisticated piece of technology that exists at current. So
far, no other race nor intelligence has surpassed it, especially not with the
current stage of technology that this world has available to us.
Recently, a comparison was made
between the current largest artificial neural networks and the human brain,
where the conclusion was that the latter was hundreds of times smaller than the
former (Ng, 2015). In this sense, one could suggest that machines are only
really good at taking in and processing huge amounts of information in a short
space of time, however, machines could never have the same level of
consciousness and independent thought as humans have, limiting aspects like
creativity and innovation, which would essentially be key factors in their
survival and evolution if they were planning to take over.
So, Where Does That Leave Us? Are We Safe?
Therefore,
the safety of AI is limited by the tasks they can do. In one sense, they may be
able to build weapons and try to control human behaviour, but at the same time,
humans may be able to think of a way around their "systems" and a
figure a way out, which will stop the take over from succeeding. Thus, there is
still hope, and who is to say that AI won't be "friendly"? (Yudkowsky
et al., 2010). This is an idea that will
be explored in another article.
No comments:
Post a Comment